Cenozoic Mammals of Land and Sea: Tributes to the Career of Clayton E. Ray Robert J. Emry EDITOR Smithsonian Institution Press Washington, D.C. 2002 A B S T R A C T Emry, Robert J., editor. Cenozoic Mammals of Land and Sea: Tributes to the Career of Clay­ ton E. Ray. Smithsonian Contributions to Paleobiology, 93, 372 pages, 197 figures, 11 plates, 46 tables, 2002.—This is a volume of collected papers published to honor the career of Clayton E. Ray, now Curator Emeritus in the Department of Paleobiology, National Museum of Natural History, Smithsonian Institution, and Curator of Late Cenozoic Mammals and of Fossil Marine Mammals in the same department for more than 30 years before his retirement in 1994. The volume includes a preface, a biography and bibliography of Clayton E. Ray, and 19 papers devoted principally to Pleistocene mammals and to fossil marine mammals. Gary Morgan describes late Pleistocene mammalian faunas from several sites in southernmost Florida and discusses the Neotropical influence in Florida's Pleistocene faunas. Richard H. Tedford describes the basicranium of the Pleistocene giant wombat Phascolonus gigas Owen and dis­ cusses its significance in marsupial phylogenetic reconstruction. Gerardo De Iuliis and A. Gor­ don Edmund describe Vassallia maxima Castellanos, the only pre-Pleistocene pampathere known in which a skull and mandible are associated with osteoderms; the range of osteoderm variation in one associated individual allows them to synonymize other taxa that had been based on osteoderm differences. Paul W. Parmalee and Russell Wm. Graham report additional records of the giant beaver, Castoroides, from the mid-South. Frederick Grady, Joaquin Arroyo-Cabrales, and E. Ray Garton report the northernmost known occurrence of vampire bats in the Pleistocene of eastern North America. H. Gregory McDonald reports the second known occurrence of the badger Taxidea taxus in the Pleistocene of Kentucky and discusses the paleoecological implications of the occurrence. Jerry N. McDonald and George E. Lam- mers describe Bison antiquus from Ontario and discuss the evolution of bison in the Holocene of North America. Daryl P. Domning presents a new analysis and interpretation of the terres­ trial posture in desmostylians. Thomas A. Demere and Annalisa Berta describe new material and present a phylogenetic analysis of the Miocene pinniped Desmatophoca oregonensis from Oregon. Irina A. Koretsky and Dan Grigorescu describe and evaluate the systematic position of the fossil monk seal Pontophoca sarmatica from the Miocene of eastern Europe. Irina A. Koretsky and Peter Holec describe a new, primitive, phocid pinniped from the early middle Miocene of Slovakia and discuss its bearing on the phylogeny and classification of pinnipeds. Irina A. Koretsky and Albert E. Sanders report remains of the oldest known phocid pinniped from the late Oligocene of South Carolina. R. Ewan Fordyce describes and discusses a bizarre archaic Oligocene dolphin from the eastern North Pacific, on which he bases a new species, genus, and subfamily. Christian de Muizon, Daryl P. Domning, and Darlene R. Ketten describe and discuss the paleobiology and behavior of an unusual walrus-convergent delphinoid ceta­ cean from the early Pliocene of Peru. Susan D. Dawson and Michael D. Gottfried report paleo- pathologic conditions in a Miocene odontocete cetacean. Albert E. Sanders and Lawrence G. Barnes contribute two papers, both describing and analyzing new, primitive, cetotheriid mys- ticete cetaceans from the late Oligocene of South Carolina. James W. Westgate and Frank C. Whitmore, Jr., describe a new species of bowhead whale from the Pliocene Yorktown Forma­ tion in Virginia. James G. Mead and Rosemary G. Dagit present an account of the search for the 1880s manuscript of J.A. Allen's unpublished monograph on the mammalian orders Cete and Sirenia; the manuscript was not found but the 12 plates that were prepared for it are pub­ lished herein. OFFICIAL PUBLICATION DATE is handstamped in a limited number of initial copies and is recorded in the Institution's annual report, Annals of the Smithsonian Institution. SERIES COVER DESIGN: The trilobite Phacops rana Green. Library of Congress Cataloging-in-Publication Data Cenozoic mammals of land and sea : tributes to the career of Clayton E. Ray/ Robert J. Emry p. cm.—(Smithsonian contributions to paleobiology ; no. 93) 1. Mammals, Fossil. 2. Paleontology—Cenozoic. I. Emry, Robert J. II. Series. QE881 .C46 2002 569-dc2I 2002030521 © The paper used in this publication meets the minimum requirements of the American National Standard for Permanence of Paper for Printed Library Materials Z39.48 1984 Contents Page Preface i Biography and Bibliography of Clayton Edward Ray Ralph E. Eshelman, Robert J. Emry, Daryl P. Domning, and David J. Bohaska 1 Late Rancholabrean Mammals from Southernmost Florida, and the Neotropical In­ fluence in Florida Pleistocene Faunas Gary S. Morgan 15 The Basicranium of the Giant Wombat Phascolonus gigas Owen (Vombatidae: Mar- supialia) and Its Significance in Phylogeny RichardH. Tedford 39 Vassallia maxima Castellanos, 1946 (Mammalia: Xenarthra: Pampatheriidae), from Puerta del Corral Quemado (Late Miocene to Early Pliocene), Catamarca Province, Argentina Gerardo De Iuliis and A. Gordon Edmund 49 Additional Records of the Giant Beaver, Castoroides, from the Mid-South: Alabama, Tennessee, and South Carolina Paul W. Parmalee and Russell Wm. Graham 65 The Northernmost Occurrence of the Pleistocene Vampire Bat Desmodus stocki Jones (Chiroptera: Phyllostomatidae: Desmodontinae) in Eastern North America Frederick Grady, Joaquin Arroyo-Cabrales, and E. Ray Garton 73 Second Record of the Badger Taxidea taxus (Schreber) from the Pleistocene of Ken­ tucky and Its Paleoecological Implications H. Gregory McDonald 77 Bison antiquus from Kenora, Ontario, and Notes on the Evolution of North Ameri­ can Holocene Bison Jerry N. McDonald and George E. hammers 83 The Terrestrial Posture of Desmostylians Daryl P. Domning 99 The Miocene Pinniped Desmatophoca oregonensis Condon, 1906 (Mammalia: Car- nivora), from the Astoria Formation, Oregon Thomas A. Demere and Annalisa Berta 113 The Fossil Monk Seal Pontophoca sarmatica (Alekseev) (Mammalia: Phocidae: Monachinae) from the Miocene of Eastern Europe Irina A. Koretsky and Dan Grigorescu 149 A Primitive Seal (Mammalia: Phocidae) from the Early Middle Miocene of Central Paratethys Irina A. Koretsky and Peter Holec 163 Paleontology of the Late Oligocene Ashley and Chandler Bridge Formations of South Carolina, 1: Paleogene Pinniped Remains; The Oldest Known Seal (Carnivora: Phocidae) Irina A. Koretsky and Albert E. Sanders 179 Simocetus rayi (Odontoceti: Simocetidae, New Family): A Bizarre New Archaic Oli­ gocene Dolphin from the Eastern North Pacific R. Ewan Fordyce 185 Odobenocetops peruvianus, the Walrus-Convergent Delphinoid (Mammalia: Ceta­ cea) from the Early Pliocene of Peru Christian de Muizon, Daryl P. Domning, and Darlene R. Ketten 223 IV SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY Paleopathology in a Miocene Kentriodontid Dolphin (Cetacea: Odontoceti) Susan D. Dawson and Michael D. Gottfried 263 Paleontology of the Late Oligocene Ashley and Chandler Bridge Formations of South Carolina, 2: Micromysticetus rothauseni, a Primitive Cetotheriid Mys- ticete (Mammalia: Cetacea) Albert E. Sanders and Lawrence G. Barnes 271 Balaena ricei, a New Species of Bowhead Whale from the Yorktown Formation (Pliocene) of Hampton, Virginia James W. Westgate and Frank C. Whitmore, Jr. 295 Paleontology of the Late Oligocene Ashley and Chandler Bridge Formations of South Carolina, 3: Eomysticetidae, a New Family of Primitive Mysticetes (Mammalia: Cetacea) Albert E. Sanders and Lawrence G. Barnes 313 Plates for Joel Asaph Allen's Unpublished Monograph on the Mammalian Orders Cete and Sirenia and a Record of the Search for the Manuscript James G Mead and Rosemary G Dagit 357 Preface When Clayton Ray retired from the National Museum of Natural History at the end of 1994, those of us who had been closely associated with him for many years, who had sat down to lunch with him several times each week, who had talked with him almost daily, were left with the feeling that we should do something to recognize his career and his con­ tributions to his profession. Clayton, typically, had discouraged any kind of social event, such as a department party or luncheon, to celebrate the occasion of his retirement. So, mindful of what Clayton's advice would have been had we asked for it, we opted to do something useful instead: we invited some of his colleagues, mainly those who had worked closely with him on certain projects, who had been his coauthors, or who had studied groups of special interest to him, to contribute papers to be assembled as a festschrift. The process of completing this volume has gone on longer than intended (al­ though not so much longer than expected), but this result surely makes the wait worth­ while. Although the "Contents" might suggest that a rather disparate range of topics is rep­ resented, in fact the contributions are all representative of Clayton's diverse interests during his career—late Cenozoic mammals, Caribbean/Neotropical mammalian biogeog­ raphy, marine mammals. There is hardly a paper in this volume to which Clayton hasn't contributed in some way, in some cases quite directly, and many of these papers would simply not have been possi­ ble without some kind of aiding and abetting by him. When we began to assemble this vol­ ume, some of the papers now included herein were in fact being prepared as individual Contributions to Paleobiology, with Clayton's encouragement, and in at least one in­ stance, with his coauthorship. The authors of these contributions were eager to have them included in this tribute, even though some of the papers would undoubtedly have been published sooner as individual contributions. As editor, I take this opportunity to express my gratitude to all of the contributors for their patience in seeing this through the long pro­ cess from conception to publication. I also want especially to thank David J. Bohaska, Irina A. Koretsky, and Robert W. Purdy, without whose ideas, encouragement, and assis­ tance that long process might have been immeasurably longer. Irina A. Koretsky and Mary A. Parrish both stepped into the breach to help with illustration revisions during the final stage of production. Early in the following biography, I mention that Clayton was bom a Hoosier, which I am confident that readers will interpret in its usual meaning: the nickname for a native or resident of Indiana. The term sometimes also connotes a hayseed, a bumpkin, an unskilled rustic. Clayton's well-known sense of humor sometimes includes self-deprecation, so whereas he might not reject this meaning outright, it is hardly appropriate for a naturalized Virginian gentleman farmer. But there is a particular meaning of the term that applies to Clayton. The several office dictionaries that I consulted gave no etymology for hoosier, but the Webster s Third New International Dictionary, Unabridged does have this proba­ ble etymology: "alter, of E dial, hoozer anything large of its kind." So the word does fit. Although Clayton is not large in physical stature, he must have carried some weight; the breadth and depth of the imprints he has left through the course of his professional career are unmistakable evidence of "a large one of his kind," not the Hoosier by biogeographical chance, but the self-made Hoosier. Robert J. Emry >r f Clayton E. Ray, 1979. Biography and Bibliography of Clayton Edward Ray Ralph E. Eshelman, Robert J. Emry, Daryl P. Domning, and David J. Bohaska With this volume we pay tribute to our mentor and colleague, Clayton E. Ray. Scientist, historian, ardent supporter of ama­ teur and professional paleontologists alike, Clayton is most es­ pecially a sound advisor and a valued friend. One of the things we miss most as a result of his retirement is the benefit of his full-time good counsel: we counted on his instinct for the clear­ headed, no-nonsense approach to problems, minor or major, and we relied upon him as a source for thoughtful, good advice, always strengthened with encouragement. In all of his personal interactions, Clayton is invariably perceptive, stimulating, helpful, patient, and kind. All of these qualities, combined with his quick wit and keen sense of humor, have enriched our long associations with him. We all hope that some of his attributes have been communicable—that something might have rubbed off on all of us to make us more like him. When we began to organize this tribute, our intention was to publish it as a surprise to Clayton, but the process has now taken so long that the bigger surprise will be if Clayton does not find out what is afoot. One disadvantage of intending this as a surprise is that it inhibited our grilling him for biographical information, and thus limited our biography essentially to what is known or could be researched by all of us—the professional, public aspect. Consequently, this cannot be expected to be an in-depth biography, but a rather superficial review of the gen­ eral trends in Clayton's professional career. Clayton's wife, Donna, supplied us with a few of the basic facts of his pre-university days. Clayton was born a Hoosier, in Henry County, Indiana, to Lloyd and Ruth Ray, on 6 February 1933 and grew up in central Indiana. Clayton remembers Ralph E. Eshelman and Darryl P. Domning, Research Associates, De­ partment of Paleobiology, National Museum of Natural History, Smithsonian Institution, Washington, D.C. 20560-0121; Robert J. Emry and David J. Bohaska, Department of Paleobiology, National Museum of Natural History, Smithsonian Insttitution, Washington, D.C. 20560-0121. fondly the rural life he experienced around his boyhood home in Indiana, especially the summers spent on his grandparents' farm. He attended high school in Indianapolis, ranked third in his class at graduation, and was awarded a full scholarship to Harvard College. He married his high-school sweetheart, Donna, and together they raised four daughters. Clayton enrolled at Harvard College, Cambridge, Massachu­ setts, in 1951 and received his A.B. degree in geology in 1955. He was elected to Phi Beta Cappa. His advisor as an undergrad­ uate was Alfred S. Romer, one of the most prominent and highly respected vertebrate paleontologists of his time. Toward the end of his senior year, as Clayton met with Romer to dis­ cuss graduate school possibilities, Romer revealed to him, in confidence, that Brian Patterson would be coming to Harvard and that Clayton might want to discuss graduate school possi­ bilities with Patterson when he came to visit. Clayton believes that he was the first "civilian" (as he puts it) to know that Brian Patterson would be moving from Chicago to Harvard. Clayton followed Romer's advice, met with Patterson, and decided to remain at Harvard for advanced degrees with Patterson as his advisor. Clayton does stress, however, that, in the real, practical sense, it was Professor Ernest Williams who served as his grad­ uate advisor and had the greatest impact on his career. It was Williams, for example, who steered Clayton's interests into Quaternary mammals of the Neotropics, and the Caribbean re­ gion in particular. Clayton was awarded his A.M. degree in 1958 and his Ph.D. in 1962 from Harvard, both in geology. In September 1959 Clayton moved to the University of Flor­ ida, Gainesville, where he accepted the positions of Interim As­ sistant Professor in the Department of Biology and Interim As­ sistant Curator at the Florida State Museum (now the Florida Museum of Natural History). Upon receiving his doctorate from Harvard, he was promoted to Assistant Professor and As­ sistant Curator, positions he held until he moved to the Smith­ sonian Institution in 1963. SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY Among his duties in Florida was, of course, fielding ques­ tions about fossils from visitors. One day he was manning the museum's front office in the old Seagle Building on Gaines­ ville's main street when a high school student from Biloxi, Mississippi, named Daryl Domning walked in, his long-suffer­ ing parents in tow, seeking publications and tips on where to collect fossils in the area. Clayton patiently gave them direc­ tions to Paynes Prairie, an area on the edge of town where spoil piles from interstate highway construction offered bones of Pleistocene critters to the casual collector; afterward, he helped them identify their finds. This brief encounter was followed by countless more in subsequent years, as Clayton came to play a major role in fostering Domning's career as a vertebrate pale­ ontologist. The two have since coauthored nearly a dozen pub­ lications on marine mammals. Early in the 1960s, the National Museum of Natural History, Smithsonian Institution, began planning a major new exhibition hall devoted to Pleistocene mammals. When the museum de­ cided in 1963 to hire a new curator to be in charge of developing these new exhibits, Clayton was the choice. Clayton's letter ac­ cepting the new position was typical of him in its thoughtfulness and clarity; he outlined prior obligations that he had to meet and asked to make his move to Washington, D.C, between school semesters to lessen the disruption to his children's education. On 18 December 1963, Clayton became Associate Curator of Later Cenozoic Mammals at the museum, where he remained until his retirement more than 30 years later. It is probably not generally known that when research scien­ tists (i.e., curators) are hired at the National Museum, they do not immediately become federal civil service employees, but must first complete a conditional, term appointment of at least one year. If all is well at the end of that term, the appointment is then converted to Career Civil Service status. C L . (Lew) Gazin, senior vertebrate paleontologist at the Smithsonian at the time Clayton arrived, wrote the following evaluation of Clayton's first year: I have found him to be a highly intelligent and industrious worker. During the rather short time since receiving his doctorate at Harvard he has had published or has completed for publication a surprising number of well-written and care­ fully thought out papers. Much of the time of his first year here had been de­ voted, as anticipated, to planning with the exhibits staff renovation of Hall 6 for a new series of exhibits which are to depict the Pleistocene epoch in vertebrate paleontology. Nevertheless, this has not prevented him from carrying on field work in Mexico during the early part of the summer and in initiating at other times several short trips of profit to the museum. Moreover, in addition to keeping abreast of a rather considerable amount of examination and report work he has managed to squeeze in an appreciable amount of research and give a few lectures. Dr. Ray is a personable young man who gets along well with his colleagues and makes a good impression on outsiders. It gives me pleasure to report that we are entirely satisfied with his work. So began Clayton's career of more than three decades as a Smithsonian Curator. Clayton officially retired at the end of September 1994 and is now a Curator Emeritus in the Depart­ ment of Paleobiology. Fieldwork and Research Clayton's earliest publications were on modern mammals, including remains of mammals recovered from archeological sites, on Pleistocene mammals, and even on an osteological pe­ culiarity of the extant gopher tortoise (see bibliography, follow­ ing). In 1960 he cofounded, with two archeologists at the Uni­ versity of Florida, the first formally constituted museum pro­ gram in archaeozoology. Elizabeth Wing, an early associate in the program, has gone on to become a leading figure in the field, and similar formal programs in archaeozoology are now common in other museums and universities. Clayton's interests in archaeozoology have continued throughout his career, even as his research evolved to focus more on Quaternary mammals and eventually on fossil marine mammals. The latter, however, had been prominent among his interests almost from the start. When Remington Kellogg, the preeminent curator of fossil ma­ rine mammals at the Smithsonian, passed away, Clayton's at­ tention shifted progressively to marine mammals, and by the late 1970s this assumed priority over his work on Quaternary paleontology. The collections and research data on Quaternary mammals that Clayton had brought together were largely be­ queathed to others. For example, bats were turned over to Gary Morgan, ungulates to Jerry McDonald, Caribbean oryzomyine rodents to Mike Carleton, and faunal studies to Ralph Eshel­ man. In concentrating on marine mammal fossils at the Smith­ sonian, Clayton continued a tradition that goes back almost as far as the Institution itself. In fact, the first vertebrate fossil cat­ aloged in the Smithsonian's collections was a marine mammal bone, and work on fossil marine mammals has been conducted by scientists such as F.A. Lucas, Frederick True, Gerrit S. Miller, Jr., Remington Kellogg, and Frank C. Whitmore, Jr. Be­ cause of this long-standing emphasis, the fossil marine mam­ mal collection at the Smithsonian is the largest in the United States and probably in the world; certainly it is one of the two most important fossil marine mammal collections in the world. Clayton's contributions have added substantially to this already strong collection; most important, he was the agent directly re­ sponsible for the Smithsonian's acquiring the Douglas Emlong collection of Tertiary marine mammals from the West Coast, and for nearly three decades he oversaw the collecting of the huge volume of material from the late Tertiary Lee Creek Mine in North Carolina. In his own research, Clayton has taken a lively interest in nearly every group of aquatic mammals, but pinnipeds most of all; among these, thanks largely to the gar­ gantuan Lee Creek project, he has focused mainly on the pho­ cid seals and on walruses (the subject of his first marine mam­ mal paper, way back in 1960). In the 1970s, Clayton and Charles Handley, Curator of Mam­ mals in the Division of Mammalogy of the Department of Ver­ tebrate Zoology, attempted to form a marine mammal study center for both fossil and modern marine mammals at the Smithsonian. An important objective was to bring the separate collections together both administratively and, especially, physically. As anyone who has studied whale skeletons knows, NUMBER 93 FIGURE 1.—Clayton E. Ray, ready for the long drive to Adelaide, after breaking camp at the end of field work, Lake Callabonna, South Australia, 19 September 1970. (Photograph by R.J. Emry.) the specimens are immovable in any practical sense, and com­ parative studies could be greatly facilitated by having all mate­ rial, fossil and modern, stored together. Space was temporarily obtained at a surplused U.S. Navy building (a World War II tor­ pedo factory) in Alexandria, Virginia. The modern whale col­ lections were moved there from the museum, but this space was lost before the fossil collections could be moved, and no other adequate space could be found. These efforts did result, however, in a unified library, the Remington Kellogg Library of Marine Mammalogy. A new curatorial position for modern whales was established, and a monthly seminar on marine mammals was begun in 1979. Now, in 2001, the goal of bring­ ing the modern and fossil skeletal materials together has finally been partially realized; the fossil baleen whale material has fol­ lowed the modern material to the Smithsonian's Garber Facil­ ity, a new storage facility in Silver Hill, Maryland. Clayton's early fieldwork included numerous sites in Flor­ ida, which he worked from 1956 through 1964. In 1958 Clay­ ton conducted fieldwork in the Dominican Republic and Puerto Rico, and in 1963 he went back to the Dominican Republic and also to the Lesser Antilles. In 1964 Clayton conducted fieldwork at Ladds Quarry, Geor­ gia, collecting an extensive fauna from a paleoecologically crit­ ical area that had not previously been studied; this collection represents one of the larger Pleistocene faunas of North Amer­ ica, with the bulk of the material now housed in the Smithso­ nian. In 1964 and 1966 Clayton participated in a multidisciplinary study, sponsored by Harvard University and the National Sci­ ence Foundation, at Valsequillo, Puebla, Mexico, where late Pleistocene deposits contained associated early humans and fossil vertebrates. Much of the material collected was retained by the Mexican government, but some of the fossil vertebrates were accessioned into the collections of the Smithsonian. From 1964 through 1967, Clayton conducted fieldwork in the late Pleistocene deposits at Saltville, Virginia, supported by a grant from the National Science Foundation. Specimens re­ sulting from this work, and associated specimens donated by Virginia Polytechnic Institute, now reside at the Smithsonian. Subsequently, Research Associate Jerry McDonald continued fieldwork at the site for a number of years. In 1966 Clayton participated in fieldwork at Gate City, Vir­ ginia, where parts of a skeleton of Megalonyx were discovered in a cave. The specimen is now part of the Smithsonian collec­ tions. In 1967 Clayton oversaw fieldwork in borrow pits around the Norfolk, Virginia, area; this work produced about 200 late Pleistocene vertebrate specimens, which, for speci­ mens from the Atlantic Coastal Plain, are unusual in having good stratigraphic control. In 1968 Clayton worked in Sardinia, Sicily, Malta, and Ma­ jorca. One objective of this work was to acquire material for the new Pleistocene exhibits then being developed at the Smithsonian, but this work also produced microfaunal material for future research purposes. In 1970 Clayton, accompanied by Chief Preparator Franklin Pierce, joined Dick Tedford and Bob Emry from the American Museum of Natural History for fieldwork in the dry northern interior of South Australia (Figure 1). A primary objective of that expedition was also to acquire material for the Pleistocene exhibits. Ten weeks of collecting in Pleistocene deposits ex­ posed in the dry, deflating bed of Lake Callabonna produced numerous skeletons of extinct Australian megafauna, including SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY giant marsupials and giant birds. Among the specimens person­ ally collected by Clayton were two skeletons of the giant wombat Phascolonus; these specimens are the material basis of Dick Tedford's paper in the present volume. A cast of one of the skeletons is now exhibited at the American Museum of Natural History. Since 1970, most of Clayton's fieldwork has been in connec­ tion with his long-term project at the Lee Creek Phosphate Mine in eastern North Carolina. This project has resulted in the majority of vertebrate fossil specimens accessioned into the Smithsonian's collections during this time. Clayton worked with and advised many amateur collectors who helped gather specimens, and he encouraged them to donate important mate­ rial to the Smithsonian. He has also worked with nearly 50 au­ thors to study and publish the findings. Thus far, three volumes have been published, covering the geology of the site, the plant and invertebrate fossils, and all non-mammalian vertebrates. Clayton is presently close to completing the editing of the fourth and final volume (this one covering mammals, marine and terrestrial) resulting from this research. The Lee Creek Phosphate Mine will certainly be the most comprehensively studied fossil site on the Atlantic Coastal Plain, and, when completed, the four volumes will record what is very likely the taxonomically most diverse fossil assemblage known from a single site. In addition to his own work on Smithsonian collections, Clayton has actively promoted work on the collections by oth­ ers. Much of this was accomplished by his taking full advan­ tage of Smithsonian in-house funding sources, such as the Short-Term Visitor Awards Program, Pre- and Postdoctoral Fellowships, the Remington Kellogg Fund, and the Walcott Fund, as well as outside sources (e.g., the Marine Mammal Commission), to enable non-Smithsonian researchers to study and publish on the collections and thereby add to the knowl­ edge about them. Clayton was advisor to several Smithsonian Predoctoral and Postdoctoral Fellows whose research included using the Smith­ sonian collections. Among them are David Gillette, who re­ viewed the glyptodonts of North America; Russell Graham, who worked up the Valsequillo fauna of Mexico that Clayton had collected; Ewan Fordyce, from New Zealand, who studied primitive cetaceans in the Emlong collection; Kishor Kumar, from Pakistan, who also researched primitive cetaceans; Al Myrick, who studied Miocene porpoises; Jerry McDonald, then of Radford University, who worked on fossil ovibovines (musk oxen); and Irina Koretsky, a specialist on fossil phocid pinni­ peds, who had emigrated to the United States from Kiev, Ukraine. Under Clayton's sponsorship, many others received short- term support for a variety of projects: Joaquin Arroyo-Ca- brales, Mexico, worked on vampire bats; Oscar Carranza, Mex­ ico, Mexican faunas; Pamela Rasmussen and Jonathan Becker, fossil birds from Lee Creek Mine; Castor Cartelle, Brazil, megatheres; Laszlo Kordos, Budapest, Miocene land and ma­ rine faunas; Guram Mchedlidze, Tbilisi, mid-Tertiary Cetacea; Irina Dubrovo, Moscow, fossil cetaceans; Christian de Muizon, Lima (now Paris), Pliocene pinnipeds; Oldrich Fejfar, Prague, Miocene rodents; Mario Cozzuol, Buenos Aires, fossil Ceta­ cea; Annalisa Berta, Tom Demere, and Sharon Messenger, all marine mammal workers from San Diego. Clayton was an Adjunct Professor in the Department of Geological Sciences, Virginia Polytechnic Institute, in 1969. He also served on the doctoral committee of Gregory Mc­ Donald, University of Toronto, whose dissertation dealt with edentates. Clayton has been the long-term sponsor of several Research Associates who have conducted collection-based research at the museum, have added materials to the collections, and have contributed substantially to curation of certain collections: Ta- seer Hussain, Neogene mammals; Annalisa Berta, fossil pinni­ peds; Daryl Domning, fossil Sirenia and Desmostylia; Ralph Eshelman, Neogene faunas; Jerry McDonald, Pleistocene un­ gulates; Charles Repenning, fossil pinnipeds and microtines; and Frank Whitmore, Jr., fossil whales. This collaborative work has resulted in important studies of such topics as faunas and stratigraphy of Paratethys, emphasiz­ ing fossil pinnipeds; archaic whales from New Zealand, Ant­ arctica, and the west and southeast coasts of the United States; the most diverse nonmarine Miocene mammal fauna east of the Mississippi River outside of Florida; and a bizarre cetacean with convergent walrus-like morphology from the Pliocene of Peru. Clayton's recommendations made at the 1967 Darwin Conference led to the Quaternary cave studies conducted by Dave Steadman on the Galapagos Islands. Since 1976, Clayton has been an author of nine publications of the Smithsonian Contributions to Paleobiology (nos. 28, 40, 51, 52, 53, 59, 61, 66, and 90). Number 59 (1986, coauthored with Daryl Domning and Malcolm McKenna), "Two New Oli­ gocene Desmostylians and a Discussion of Tethytherian Sys­ tematics," reported a link among major groups of mammals, with important information bearing on relationships at the su- praordinal level. The paper has since been widely cited by other authors discussing the higher-level classification of mammals. Among Clayton's other more important contributions are his papers on the oldest known pinniped skeleton, which has stim­ ulated and contributed to ongoing intensive debate on the defi­ nition, origin (or origins?), and relationships within and outside the Pinnipedia: is the Pinnipedia a natural monophyletic group or the result of convergence of two groups of carnivores with separate origins? In addition to his scientific publications, Clayton has written book reviews and popular articles. Examples include reviews of the Sierra Club Handbook on Marine Mammals and Sut- cliff's On the Track of Ice Age Mammals. Clayton served as a consultant on the Smithsonian-American Heritage book The Evidence of Evolution, and the Macmillan Company book NUMBER 93 Hunting for Fossils. Popular articles include "The time has come the walrus said..." in Virginia Explorer, "Outerbank Ovi- bovine Adds Link to Musk Ox Research," and "Catawba's wandering walrus: Case of peripatetic pinniped solved," both in Whalebones, Newsletter of the North Carolina State Mu­ seum of Natural History. Clayton has been, of course, a frequent reviewer of proposals and manuscripts for the National Science Foundation, National Geographic Society, Sigma Xi, American Philosophical Soci­ ety, Journal of Paleontology, Journal of Vertebrate Paleontol­ ogy, and other similar journals and grant-awarding agencies. He has a well-deserved reputation as a thorough reviewer who not only finds mistakes but also offers solutions. He never met a split infinitive that he could not repair, and it is almost a sure bet that he will notice some in this volume. Service to His Profession During his freshman year at Harvard College, Clayton re­ ceived a $50 grant from a fund that had been established in the seventeenth century to enable students to return to their homes to be with their families during the Christmas holidays. To this event Clayton attributes the beginning of his continuing inter­ est in establishing and supporting endowment funds. He was instrumental in establishing the Cooper Fund and the Kellogg Fund at the Smithsonian, and the Society of Vertebrate Paleon­ tology Endowment Fund. The Society of Vertebrate Paleontology Endowment Fund did not yet exist when Clayton became Chair of the Develop­ ment Committee in 1986. Clayton had been concerned that the society might not be able to continue its long support of the Bibliography of Fossil Vertebrates, an important reference se­ ries within the field, but one that was becoming increasingly expensive to produce. Clayton urged the society to establish an endowment fund that would ensure the continuation of the bib­ liography, and this led to his being invited to chair an endow­ ment committee. His committee organized a fund-raising cam­ paign with a goal of $500,000 within five years. That goal was surpassed within three years; by 1992 the balance was over $750,000 and now is more than twice that amount. At the 1989 annual meeting, the Society of Vertebrate Paleontology pre­ sented Clayton with a special citation and plaque for his service on the committee. Clayton's response to the membership is so typical of him that it is worth quoting here for the insight it gives into his character: On a personal note, as this is my last communication to you from the Develop­ ment Committee, 1 was surprised at the business meeting and very pleased to receive a plaque from the Society in recognition of my service on the commit­ tee. Rather I owe thanks to the Society for enabling me to pursue on its behalf a cause that I believe to be profoundly important to its future, for making that pursuit successful, and for affording me the opportunity to work closely with the best of the best in VP. Most of all I am proud of my good judgment as to the character of our members. Clayton was also the leader in persuading the Society of Ver­ tebrate Paleontology to assume responsibility for publishing the Journal of Vertebrate Paleontology. His effort began with an open letter to the society, which resulted in the formation of the Journal of Vertebrate Paleontology Committee. He was asked to chair this committee, and he did so from 1982 to 1985. Work toward this end involved distribution of a mail ballot to the membership for approval to take on such an effort, neces­ sary revision of the society's constitution, and negotiations with the University of Oklahoma and with Jiri Zidek, the jour­ nal's first editor, who had founded the journal with support from the University of Oklahoma. The Journal of Vertebrate Paleontology has continued to grow as a professional, highly respected publication and has prompted renewed efforts to in­ crease the society's endowment in order to ensure timely publi­ cation of accepted manuscripts. From October 1975 through December 1977, Clayton served as Chairman of the Committee of Scientific Advisors on Ma­ rine Mammals for the Marine Mammal Commission. John Twiss, Executive Director of the Commission, wrote, in a letter to Bob Emry, 18 August 1999, that Clayton "was an able chair­ man who had considerable influence upon the Committee's ac­ tivities. Particularly noteworthy were his efforts to focus atten­ tion on the plight of the West Indian manatee (Trichechus manatus). In no small measure as a result of his work, the Commission put into place aggressive research and manage­ ment programs which succeeded in bringing greatly increased Federal and state resources to bear on this species' protection and conservation." In the 1960s, Clayton served on the American Geological In­ stitute-Geological Society of America Geoscience Information Committee. Clayton served on the Advisory Panel on the Fu­ ture of Mammalogy at the Museum of Comparative Zoology, Harvard University, in 1985; he was a member of the Search Committee for Curator of Vertebrate Paleontology, American Museum of Natural History, New York, in 1986. Clayton was the keynote speaker at the first annual meeting of the Marine Mammal Society and authored a paper that was pub­ lished in the first issue of its journal, Marine Mammal Science. Many of Clayton's research papers exhibit a strong historical component, and he has a keen interest in the history of our pro­ fession and the lives and careers of its early practitioners, those giants whose shoulders we all stand on. Among the accom­ plishments that Clayton is most proud of is his proposal to name the auditorium at the National Museum of Natural His­ tory after Spencer Fullerton Baird, followed by a successful campaign to persuade the museum's administration to do so. Clayton was banquet speaker at the 1989 Florida Paleontologi­ cal Society meeting on the occasion of the centennial of the first publication on Florida vertebrate paleontology by Joseph Leidy. He was also the invited speaker at the 1991 ceremony dedicating the Wagner Free Institute of Science, Philadelphia, as a National Historic Landmark on the centennial of Joseph Leidy's death. SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY FIGURE 2 (left).—Clayton E. Ray reviewing the modeled jaws of Carcharodon megalodon containing actual fossil teeth from Lee Creek Mine, prior to installation in the paleontology exhibits at the National Museum of Natural History, 1985. From left: Rob­ ert J. Emry, Curator, Department of Paleobiology; Thomas R. Harney, Public Information Officer, Office of the Director; Clay­ ton E. Ray, Curator, Department of Paleobiology; an unidenti­ fied person obscured by Clayton Ray; Victor G. Springer, Cura­ tor, Department of Vertebrate Zoology; Ian G. Maclntyre, Curator and then Chairman, Department of Paleobiology; Walter E. Hock, Sr., Modelmaker, Smithsonian Exhibits Central; Robert W. Purdy, Museum Specialist, Department of Paleobiology. All NMNH staff except Hock. (Photograph by Chip Clark, NMNH.) Professional Affiliations Clayton has been a Life Member of the American Society of Mammalogists since 1956. He is an Honorary Member of the Marine Mammal Society; a member of the Biological Society of Washington; Florida Paleontological Society; Georgia Acad­ emy of Science; Gruppo Grotte Nuorese, Associazione Speleo- logica; Paleontological Society of Washington; William Pen- gelly Cave Studies Association; and United States Association for Quaternary Research. He is the only Honorary Charter Member of the North Carolina Fossil Club. He is also a mem­ ber of the Society of Vertebrate Paleontology, where he served as Chair of the Journal of Vertebrate Paleontology Committee, 1982-1985, Chair of the Development Committee, 1986-1989, and Vice President and President-Elect, 1990-1992. Exhibits Early in the planning phase of the Pleistocene Hall at the Na­ tional Museum, Clayton suggested a multidisciplinary ap­ proach of using "living fossils" and introducing the relation­ ship of early humans with the flora and fauna. During development of the Pleistocene Hall from 1964 through 1971, Clayton advised on the preparation of all skeletal mounts and murals, and he also arranged for acquisition of specimens needed for the exhibit by exchange, gift, or field collection. For example, the Alaskan material (consisting of specimens from the Fairbanks area frozen gravels, including a composite skele­ ton of a woolly mammoth and a Bison mummy) and the Ari­ zona glyptodont specimens, reputed to be the best examples from North America, were acquired by exchange from the American Museum of Natural History. Of the Australian mate­ rial collected in 1970 by a joint Smithsonian Institution-Amer­ ican Museum of Natural History team, only a skull of the giant marsupial Diprotodon was eventually used. Clayton completed the gift of a composite skeleton of the rare goat-antelope, Myo­ tragus balearicus, from the Deya Archeological Museum in Majorca. His fieldwork in the Mediterranean produced the ma­ terial needed to make composite skeletons of the dwarf giant deer from Crete, Myotragus from Majorca, Prolagus from Sar­ dinia, and giant dormouse from Sicily; these materials remain in the museum's collections but were not assembled for the ex­ hibit. Clayton also coordinated the collection of dwarf hippo, lemur, tortoise, and elephant bird remains from the Pleistocene of Madagascar. NUMBER 93 Clayton proposed the "Jaws" exhibit on fossil sharks and was the primary advisor on its development (Figure 2). The original set of fossil teeth that were mounted in the recon­ structed jaws of Carcharodon megalodon were acquired through Clayton's Lee Creek Mine project, mainly through the generosity of amateur collector Peter Harmatuk. The exhibit was completed in October 1985, and the poster that accompa­ nied the exhibit, featuring members of the exhibit team framed by the reconstructed jaws, quickly became a collectors' item. Clayton also assisted in the acquisition of new fossils for the "Life in the Ancient Seas" exhibit, which opened in May 1990, and he supervised the mounting or remodeling of several ma­ rine mammal skeletons, plus creation of the mural, for this ex­ hibit. Clayton worked actively with the natural history museum community outside the Smithsonian. For example, he testified before the Virginia legislature in favor of the state establishing a museum of natural history, served first as an advisor in 1988, and was then appointed by the Governor of Virginia as a trustee of the Virginia Museum of Natural History in 1989. Among his other services, he helped write the collections policy manual for that museum. Because Thomas Jefferson was an early prac­ titioner of our profession of vertebrate paleontology, and one of Clayton's heroes, Clayton was especially proud to be awarded the Jefferson Medal in 1993 by Governor Wilder and the Vir­ ginia state legislature for his outstanding contributions to sci­ ence, and especially for his work with the Virginia Museum of Natural History. Clayton is an Advisory Member of the Saltville Foundation and its Museum of the Middle Appalachians. He has served as an advisor to the Calvert Marine Museum in Solomons, Mary­ land, and its Calvert Cliffs fossil exhibit; to the Cincinnati Mu­ seum of Natural History on its Pleistocene exhibit; and to the Aurora Fossil Museum in Aurora, North Carolina. Library Anyone who has known Clayton for any time at all will know that he is an avid bibliophile and a strong advocate of building and maintaining good research libraries. He has served on a succession of Smithsonian Institution library com­ mittees through the years, and he served as the Department of Paleobiology Representative on the Library Liaison Committee before his retirement. Clayton was instrumental in arranging for the donation of Remington Kellogg's personal marine mammal library to the Smithsonian and establishing it as part of the Institution's library system, set up in a separate room near the fossil marine mammal collections. Clayton almost sin­ gle-handedly has seen to it that the Kellogg Library has been properly cared for, that new literature and unpublished works are continually added, and that an active reprint exchange pro­ gram is maintained. Clayton also facilitated the donation of the C L . Gazin and R.H. Reinhart personal libraries to the verte­ brate paleontology library holdings of the Smithsonian. He has made numerous donations to the Smithsonian Institution Ar­ chives and has arranged for all Society of Vertebrate Paleontol­ ogy documents to be deposited in the Smithsonian Institution Archives as well. Clayton's bibliophilia bore early fruit in one of his first pub­ lications, a bibliography of the fossil vertebrates of Florida (1957). He later remarked to Daryl Domning that having once produced such a work doomed one forever to be sought out as an advisor on other bibliographic projects. This destiny, how­ ever, he willingly embraced; nor did his warning dissuade Domning from using Clayton's Florida work as a model for his own first publication (a bibliography on fossil vertebrates of Louisiana and Mississippi, 1969). The Florida bibliography- also served as an early inspiration for Domning's much later (1996) and much larger bibliography of the Sirenia and Des­ mostylia—another major (and Smithsonian-published) project to which Clayton for years lent his enthusiastic encouragement and tangible support. Extracurricular Pursuits Clayton in his extracurricular pursuits, and especially in his retirement, practices sustainable agriculture, animal husbandry, and silviculture on his farm near Fredericksburg, Virginia (Fig­ ure 3). He regards this as a natural extension of his career in natural history. Some might call him a purist in these pursuits, as he maintains "a holistic commitment to an environmental lifestyle." He eschews air conditioning, uses no motorized farm equipment, and practices organic farming methods. In line with this commitment, he collects and restores horse-drawn vehicles such as wagons, buggies, and bobsleds, as well as agricultural implements. To "power" this equipment he keeps and breeds Suffolk draft horses, maintaining at least one working team. He and Donna make frequent trips, mainly to southeastern Penn­ sylvania, to attend farm auctions and shows. He usually comes home with what he considers a prize. He loaned one of his farm wagons for a temporary Hidatsa Indian exhibit and helped the Smithsonian's Folklife Festival with information on an oat- threshing demonstration. One of Clayton's major projects since he retired from the museum has been barn razing and barn raising. In 1995, a large (35 by 80 feet) dairy barn, built early in the last century, was about to be bulldozed to make way for construction of a shop­ ping mall. Clayton was given the barn in exchange for remov­ ing it from the property. With the help of his extended family, Clayton dismantled the barn piece by piece and has since reas­ sembled it on his own property. Clayton still spends occasional days (typically one day each week) at the museum, continuing his research and finishing long-term research projects. The lunchtime gathering on those days is sometimes highlighted by new photographs, perhaps of his proud team of Suffolks pulling some new implement, or maybe of a pony recently acquired for the grandchildren, or, re­ cently, baby pictures—a new foal born to one of his Suffolk SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY FIGURE 3.—Clayton E. Ray and grandson Danny afield in 1991, Stafford County, Virginia. mares. We've learned that we cannot expect to see Clayton in the museum if there is new snow in Stafford County, because that far south in Virginia, any opportunity to hitch up the bob­ sled is too rare to be missed. Finally, we take this opportunity to publish, this time com­ plete and unexpurgated, a letter as written by Clayton to the ed­ itor of Smithsonian magazine (1995). February first had begun in the dark as a humdrum day like many others with that somnambulant 50-mile drive up 1-95. Then right there in the middle of M Street, N.W., opposite my wife's office, suddenly materialized a beautiful tar- covered ball peen hammer with the handle only slightly split. I made it mine as her cautions about traffic echoed behind me. Then, in her office I find the new Smithsonian with John Neary's article about junk—this was a day to remem­ ber! I always skim through Smithsonian eagerly to keep track of what you are doing (I am a curator in NMNH), to find your mistakes, and to sniff at the superficial potboilers written by hacks just for the money. Suddenly I realized why I have been so critical of you all these years and where you have gone wrong—so lit­ tle of your stuff is written by real experts who know and love their subject. Neary's article is a sparkling exception. I assure you that my assessment has nothing to do with why I gave up jogging to and from my office back when we lived in South Arlington. The daily pick­ ings in fender washers, bolts, and screw drivers alone were too heavy and their gleaning too time consuming, to say nothing of the occasional chain or tool box that required phoning my wife from the Jefferson Memorial to come and haul me home with what I had dragged from the 14th Street bridge. Nor is my judg­ ment weighted by my assemblage of blacksmith's vises (8, 9, 10, but who's counting); the next one auctioned at a bargain bid will certainly have some beautilitarian nuance to complement my holdings. Congratulations on finally publishing something authoritative and of inherent interest to every reader. As you might imagine, and to Clayton's chagrin, his letter was heavily edited and shortened when published in the maga- Bibliography of Clayton E. Ray 1957 Pollock, H.E.D., and CE. Ray. Notes on Vertebrate Animal Remains from Mayapan. Carnegie Institution of Washing­ ton, Department of Archaeology, Current Reports, 41: 633-656. Ray, CE. A List, Bibliography, and Index of the Fossil Verte­ brates of Florida. Florida Geological Survey, Special Pub­ lication, 3: 175 pages. Ray, CE. Palaeontology. Vertebrate. In Walter Yust, editor, Britannica Book of the Year, 1957, pages 595-596. Chi­ cago, London: Encyclopedia Britannica. Ray, CE. Pre-Columbian Horses from Yucatan. Journal of Mammalogy, 38(2):278. 1958 Ray, CE. Fusion of Cervical Vertebrae in the Erethizontidae and Dinomyidae. Breviora, 97:1-13, figures 1, 2. Ray, CE. Additions to the Pleistocene Mammalian Fauna NUMBER 93 from Melbourne, Florida. Bulletin of the Museum of Com­ parative Zoology, 119(7):419-450, figures 1-5. 1959 Ray, CE. A Sesamoid Bone in the Jaw Musculature of Go- pherus polyphemus (Reptilia: Testudininae). Anatomischer Anzeiger, 107(l/5):85-91, figures 1-4. 1960 Ray, CE. Trichecodon huxleyi (Mammalia: Odobenidae) in the Pleistocene of Southeastern United States. Bulletin of the Museum of Comparative Zoology, 122(3): 127-142, text figure 1, plates 1, 2. Ray, CE. The Manatee in the Lesser Antilles. Journal of Mammalogy, 41(3):412—413. 1961 Ray, CE. The Monk Seal in Florida. Journal of Mammalogy, 42(1):113. Waters, J.H., and CE. Ray. Former Range of the Sea Mink. Journal of Mammalogy, 42(3):380-383, figure 1. 1962 Ray, CE. New Finds of Old Bones. Florida Alumnus, Uni­ versity of Florida, 14(4):20-22, 3 figures. 1963 Ray, C.E., S.J. Olsen, and H.J. Gut. Three Mammals New to the Pleistocene Fauna of Florida, and a Reconsideration of Five Earlier Records. Journal of Mammalogy, 44(3):373- 395, text figures 1-7, plates l^t. 1964 Gut, H.J., and CE. Ray. The Pleistocene Vertebrate Fauna of Reddick, Florida. Quarterly Journal of the Florida Acad­ emy of Sciences, 26(4):315—328. Hooijer, DA., and CE. Ray. A Metapodial of Acratocnus (Edentata: Megalonychidae) from a Cave in Hispaniola. Proceedings of the Biological Society of Washington, 11: 253-258, figure 1. Ray, CE. A New Capromyid Rodent from the Quaternary of Hispaniola. Breviora, 203:1^1, figure 1. Ray, CE. A Small Assemblage of Vertebrate Fossils from Spring Bay, Barbados. Journal of the Barbados Museum and Historical Society, 31 (1): 11 -22, figures 1 - 3 . Ray, CE. Tapirus copei in the Pleistocene of Florida. Quar­ terly Journal of the Florida Academy of Sciences, 27(1): 59-66, plate 1. Ray, CE. The Jaguarundi in the Quaternary of Florida. Jour­ nal of Mammalogy, 45(2):330-332, figure 1. Ray, CE. The Taxonomic Status of Heptaxodon and Dental Ontogeny in Elasmodontomys and Amblyrhiza (Rodentia: Caviomorpha). Bulletin of the Museum of Comparative Zoology, 131(5): 107-127, figures 1,2. 1965 Hibbard, C.W., CE. Ray, D.E. Savage, D.W. Taylor, and J.E. Guilday. Quaternary Mammals of North America. In H.E. Wright, Jr., and D.G. Frey, editors, The Quaternary of the United States, pages 509-525. Princeton: Princeton University Press. Ray, CE. A Glyptodont from South Carolina. Charleston Museum Leaflet, 27:1-12, plates \-4. Ray, CE. A New Chipmunk, Tamias aristus, from the Pleis­ tocene of Georgia. Journal of Paleontology, 39(5): 1016-1022, text figure 1. Ray, CE. The Relationships of Quemisia gravis (Rodentia: ?Heptaxodontidae). Smithsonian Miscellaneous Collec­ tions, 149(3): 1-12, text figures 1, 2, plate 1. Ray, CE. Variation in the Number of Marginal Tooth Posi­ tions in Three Species of Iguanid Lizards. Breviora, 236:1-15, figures 1-5. Ray, C.E., and L. Lipps. An Assemblage of Pleistocene Ver­ tebrates and Mollusks from Bartow County, Georgia. [Abstract.] Bulletin of the Georgia Academy of Science, 23(2):67. 1966 Ray, CE. The Identity of Bison appalachicolus. Notulae Naturae of the Academy of Natural Sciences of Philadel­ phia, 384:1-7, figures 1, 2. Ray, CE. The Status of Bootherium brazosis. Pearce-Sell- ards Series, 5:1-7, figures 1, 2. 1967 Lipps, L., and CE. Ray. The Pleistocene Fossiliferous Deposit at Ladds, Bartow County, Georgia. Bulletin of the Georgia Academy of Science, 25(3): 113-119, figures 1, 2. Ray, CE. Pleistocene Mammals from Ladds, Bartow County, Georgia. Bulletin of the Georgia Academy of Science, 25(3): 120-150, figures 1-5. Ray, C.E., B.N. Cooper, and W.S. Benninghoff. Fossil Mam­ mals and Pollen in a Late Pleistocene Deposit at Saltville, Virginia. Journal of Paleontology, 41(3):608-622, text fig­ ures 1-4, plates 65, 66. 1968 Ray, C.E., and L. Lipps. Additional Notes on the Pleistocene Mammals from Ladds, Georgia. Bulletin of the Georgia Academy of Science, 26(2):63. Ray, C.E., A. Wetmore, D.H. Dunkle, and P. Drez. Fossil Vertebrates from the Marine Pleistocene of Southeastern Virginia. Smithsonian Miscellaneous Collections, 153(3): 1-25, text figures 1, 2, plates 1, 2. Ray, C.E., D. Wills, and J.C. Palmquist. Fossil Musk Oxen of 10 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY Illinois. Transactions of the Illinois Academy of Science, 61(3):282-292, figures 1-5. Wing, E.S., CA. Hoffman, Jr., and CE. Ray. Vertebrate Remains from Indian Sites on Antigua, West Indies. Car­ ibbean Journal of Science, 8(3-4): 123-139, figures 1^1. 1970 Ray, C.E., CS . Denny, and M. Rubin. A Peccary, Platygonus compressus LeConte, from Drift of Wisconsinan Age in Northern Pennsylvania. American Journal of Science, 268(l):78-94, 1 text figure, 2 plates. Ray, C.E., and L. Lipps. Southerly Distribution of Porcupine in Eastern United States during Late Quaternary Time. [Abstract.] Bulletin of the Georgia Academy of Science, 28(2):24. 1971 Ray, CE. Polar Bear and Mammoth on the Pribilof Islands. Arctic, 24(1):9-18, text figure 1. 1973 Ray, CE. Review of The Distributional History of the Biota of the Southern Appalachians, Perry C Holt, editor. Sys­ tematic Zoology, 22(2): 199-201. Ray, C.E., and F.C Whitmore, Jr. Paleontology. In T. Simkin, W.G. Reeder, and C MacFarland, editors, Galapagos Sci­ ence, 1972 Status and Needs, pages 67-68. Report of Gal­ apagos Science Conference, October 6-8, 1972, Wash­ ington, D. C. [sponsored by the National Science Founda­ tion, Smithsonian Institution, and the University of Wis­ consin]. Washington, D.C: Smithsonian Institution. 1975 Ray, CE. The Relationships of Hemicaulodon effodiens Cope 1869 (Mammalia: Odobenidae). Proceedings of the Biological Society of Washington, 88(26):281-304, plates 1-6. Ray, CE. The Geography of Phocid Evolution. [Abstract.] American Zoologist, 15(3):812. Ray, CE. Field Conference in Marine Mammal Beds of Coastal Oregon, August 18-20, 1975. [Unpublished field guide], 33 pages, 2 tables, 7 maps. 1976 Ray, CE. Phoca wymani and Other Tertiary Seals (Mamma­ lia: Phocidae) Described from the Eastern Seaboard of North America. Smithsonian Contributions to Paleobiol­ ogy, 28: 36 pages, 3 text figures, 11 plates. Ray, CE. [Anonymous, with assistance and recommenda­ tions of others.] Marine Mammal Names. 8 pages. Wash­ ington, D.C: Marine Mammal Commission. [A list of the marine mammals of the world, scientific and common names, published by the commission and distributed worldwide, to serve as a standard nomenclature.] Ray, CE. Procrastination Week [letter to the editor]. Wash­ ington Post, March 17, page A18. Ray, CE. The Gray Seal in U.S. Waters. 11 pages. [Unpub­ lished "white paper" for Marine Mammal Commission.] Repenning, C.A., and CE. Ray. The Origin of the Hawaiian Monk Seal. Proceedings of the Biological Society of Wash­ ington, 89(58):667-688, plates 1-3. 1977 Ray, CE. The Geography of Phocid Evolution. Systematic Zoology, 25(4):391-406, figures 1-6. Ray, CE. Fossil Marine Mammals of Oregon. Systematic Zoology, 25(4):420-436, figures 1, 2. Ray, CE. Seals and Walruses of Florida. Plaster Jacket, 27: 1-13, figures 1-3, cover illustration. 1979 Ray, CE. Chelonia couperi Harlan 1842, a Supposed Turtle Based on the Clavicle of a Megathere (Mammalia: Eden­ tata). Notulae Naturae of the Academy of Natural Sciences of Philadelphia, 455: 16 pages, 4 figures. Ray, C.E., and D.E. Wilson. Evidence for Macrotus californi- cus from Terlingua, Texas. Occasional Papers, The Museum, Texas Tech University, 57:1-10, 2 figures. Repenning, C.A., CE. Ray, and D. Grigorescu. Pinniped Biogeography. In J. Gray and A.J. Boucot, editors, Histor­ ical Biogeography, Plate Tectonics, and the Changing Environment: Proceedings of the 37th Annual Biology Colloquium and Selected Papers, pages 357-369. Corval- lis: Oregon State University Press. 1980 Morgan, G.S., CE. Ray, and O. Arredondo. A Giant Extinct Insectivore from Cuba (Mammalia: Insectivora: Soleno- dontidae). Proceedings of the Biological Society of Wash­ ington, 93(3):597-608, 1 figure. Ray, CE. Douglas Ralph Emlong, 1942-1980. [Obituary.] Society of Vertebrate Paleontology News Bulletin, 120:45— 46, 1 photograph. 1981 Gillette, D.D., and CE. Ray. Glyptodonts of North America. Smithsonian Contributions to Paleobiology, 40: 255 pages, 97 figures. Ray, CE. Catawba's Wandering Walrus: Case of Peripatetic Pinniped Solved. Whalebones [quarterly newsletter of the North Carolina State Museum of Natural History], 13(Fall):3, 1 figure [unnumbered pages]. Ray, CE. Marine Mammal Information [guest editorial], December 1981, pages 1-3. Ray, C.E., E. Anderson, and S.D. Webb. The Blancan Carni- NUMBER 93 11 vore Trigonictis (Mammalia: Mustelidae) in the Eastern United States. Brimleyana, 5:1-36, 7 figures. Ray, C.E., and J.K. Ling. A Well Documented Early Record of the Australian Sea Lion. Archives of Natural History, 10(1):155-171, 7 figures. Ray, C.E., and A.E. Spiess. The Bearded Seal, Erignathus barbatus, in the Pleistocene of Maine. Journal of Mam­ malogy, 62(2):423-427, 2 figures. 1982 Domning, D.R, G.S. Morgan, and CE. Ray. North American Eocene Sea Cows (Mammalia: Sirenia). Smithsonian Con­ tributions to Paleobiology, 52: 69 pages, 34 figures. Ray, CE. The Journal of Vertebrate Paleontology and the SVP. Society of Vertebrate Paleontology News Bulletin, 125:5-7. Ray, CE. Review of "Handbook of Marine Mammals, Vol­ ume 1: The Walrus, Sea Lions, Fur Seals, and Sea Otter, and Volume 2: Seals," S.H. Ridgway and R.J. Harrison, editors. Journal of Mammalogy, 63(4):721-724. Ray, C.E., F. Reiner, D.E. Sergeant, and CN. Quesada. Notes on Past and Present Distribution of the Bearded Seal, Erig­ nathus barbatus, around the North Atlantic Ocean. Memo- rias do Museu do Mar, Serie Zoologica (Cascais, Portu­ gal), 2(23): 1-32, 10 figures. Steadman, D.W., and CE. Ray. The Relationships of Megao- ryzomys curioi, an Extinct Cricetine Rodent (Muroidea: Muridae) from the Galapagos Islands, Ecuador. Smithso­ nian Contributions to Paleobiology, 51: 23 pages, 11 fig­ ures. 1983 Barnes, L.G., D.P. Domning, and CE. Ray. Status of Studies on Fossil Marine Mammals. [Abstract.] Fifth Biennial Conference on the Biology of Marine Mammals, Novem­ ber 27-December I, 1983, Boston, Massachusetts, page 1. Ray, CE. Journal of Vertebrate Paleontology Committee Report. Society of Vertebrate Paleontology News Bulletin, 127:8-9, 17-19. Ray, CE. Outerbank Ovibovine Adds Link to Musk Ox Research. Whalebones [quarterly newsletter of the North Carolina State Museum of Natural History], 15(Spring):3, 1 figure [unnumbered pages]. Ray, CE. [Anonymous.] President Robert L. Carroll's Report on Amendments and the JVP. Society of Vertebrate Pale­ ontology News Bulletin, 128:1-2 [also pages 2-9, amend­ ments to SVP constitution also written by CE. Ray]. Ray, CE. Hooded Seal, Cystophora cristata: Supposed Fos­ sil Records in North America. Journal of Mammalogy, 64(3):509-512, 4 figures. Ray, C.E., editor. Geology and Paleontology of the Lee Creek Mine, North Carolina, I. Smithsonian Contributions to Paleobiology, 53: 529 pages, frontispiece, 95 figures, 101 plates. Ray, CE. Prologue. In CE. Ray, editor, Geology and Paleon­ tology of the Lee Creek Mine, North Carolina, I. Smithso­ nian Contributions to Paleobiology, 53:1-14. 1984 Ray, CE. Fossil Seals from the Calvert Formation of the Pamunkey River, Hanover County, Virginia. In L.W. Ward and Kathleen Krafft, editors, Stratigraphy and Paleontol­ ogy of the Outcropping Tertiary Beds in the Pamunkey River Region, Central Virginia Coastal Plain: Guidebook for Atlantic Coastal Plain Geological Association 1984 Field Trips, pages 232-235. Atlantic Coastal Plain Geo­ logical Association. Ray, C.E., and A.E. Sanders. Pleistocene Tapirs in the East­ ern United States. In H.H. Genoways and M.R. Dawson, editors, Contributions in Quaternary Vertebrate Paleontol­ ogy: A Volume in Memorial to John H. Guilday. Special Publication, Carnegie Museum of Natural History, 8:283- 315. 1985 Barnes, L.G., D.P. Domning, and CE. Ray. Status of Studies on Fossil Marine Mammals. Marine Mammal Science, 1(1): 15-53, 9 figures. 1986 Cifelli, R.L., and CE. Ray. Review of Vertebrate Zoogeogra­ phy and Evolution in Australasia, M. Archer and G. Clay­ ton, editors. Journal of'Vertebrate Paleontology, 6(2):203- 205. Domning, D.P, and CE. Ray. The Earliest Sirenian (Mam­ malia: Dugongidae) from the Eastern Pacific Ocean. Marine Mammal Science, 2(4):263-276, figures 1-6. Domning, D.R, CE. Ray, and M.C. McKenna. Two New Oligocene Desmostylians and a Discussion of Tethythe­ rian Systematics. Smithsonian Contributions to Paleobiol­ ogy, 59: iv + 56 pages, 23 figures. Ray, CE. [Anonymous.] Development Committee Message. News Bulletin of the Society of Vertebrate Paleontology, 138:4 [also annual auction announcement, page 57]. Ray, CE. [Anonymous.] Development Committee Report. Society of Vertebrate Paleontology News Bulletin, 137:5-6 [also raffle announcement, page 49]. Ray, CE. [Letter to the Editor.] The Four Star SIWC [Smith­ sonian Institution Women's Council] Newsletter, 7(1 ):8. Ray, C.E., and D.P. Domning. Manatees and Genocide. Marine Mammal Science, 2(1):77—78. [Letter to the editor regarding "Conservation and Protection of Marine Mam­ mals...."] 1987 Germon, R.N., Lauck W. Ward, and CE. Ray. Ecphora: 12 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY Important Fossil from the Miocene Strata on the Chesa­ peake Bay. Maryland Naturalist, 31(l):25-33, figures 1- 11. McDonald, J.N., and CE. Ray. Bootherium bombifrons, the Autochthonous Low-Horned Musk Ox of Pleistocene North America. [Abstract.] Second International Muskox Symposium, October 1-4, 1987, University of Saskat­ chewan, page 22. Ray, CE., editor. Geology and Paleontology of the Lee Creek Mine, North Carolina, II. Smithsonian Contributions to Paleobiology, 61: 283 pages, 49 figures, 80 plates. Ray, CE. Foreword. In CE. Ray, editor, Geology and Pale­ ontology of the Lee Creek Mine, North Carolina, II. Smith­ sonian Contributions to Paleobiology, 61:1-8. Ray, CE. Development Committee Message. Society of Ver­ tebrate Paleontology News Bulletin, 139:24-25 [also SVP Archives, page 63]. Ray, CE. Development Committee Activities. Society of Ver­ tebrate Paleontology News Bulletin, 140:2. 1988 McDonald, J.N., and CE. Ray. Zoogeography of the Musk Ox Genus Praeovibos Staudinger, 1908. [Abstract.] Jour­ nal of Vertebrate Paleontology, supplement, 8(3):21A- 22A. Morgan, G.S., O.J. Linares, and CE. Ray. New Species of Fossil Vampire Bats (Mammalia: Chiroptera: Desmodon- tidae) from Florida and Venezuela. Proceedings of the Bio­ logical Society of Washington, 101(4):912-928, 8 figures. Ray, CE. Development Committee Report and Second Development Workshop. Society of Vertebrate Paleontol­ ogy News Bulletin, 142:3-7. Ray, CE. [Anonymous.] Development Committee Note. Soci­ ety of Vertebrate Paleontology News Bulletin, 143:1-2. Ray, CE. Review of "A Short History of Vertebrate Palaeon­ tology," by Eric Buffetaut. Earth Sciences History, 7(2): 162. Ray, C.E., O.J. Linares, and G.S. Morgan. Paleontology. In A.M. Greenhall and U. Schmidt, editors, Natural History of Vampire Bats, pages 19-30 [Chapter 3]. Boca Raton, Florida: CRC Press. 1989 Berta, A., CE. Ray, and A. Wyss. Skeleton of the Oldest Known Pinniped, Enaliarctos mealsi. Science, 244(4900): 60-62, figures 1-3. McDonald, J.N., L.E. Freeman, and CE. Ray. The Osteology of Bootherium bombifrons: A Comparative Osteological Atlas. [Abstract.] Journal of Vertebrate Paleontology, sup­ plement, 9(3):32A. McDonald, J.N., and CE. Ray. Bootherium bombifrons, the Autochthonous Low-Horned Muskox of Pleistocene North America. [Abstract.] Canadian Journal of Zoology, 67(5): A64. McDonald, J.N., and CE. Ray. The Autochthonous North American Musk Oxen Bootherium, Symbos, and Gidleya (Mammalia: Artiodactyla: Bovidae). Smithsonian Contri­ butions to Paleobiology, 66: 77 pages, 64 figures. Ray, CE. Development Committee Report. Society of Verte­ brate Paleontology News Bulletin, 145:9. 1990 Berta, A., and CE. Ray. Skeletal Morphology and Locomo­ tor Capabilities of the Archaic Pinniped Enaliarctos mealsi. Journal of Vertebrate Paleontology, 10(2): 141- 157, 7 figures, cover illustration. Buffrenil, V. de, A. de Ricqles, CE. Ray, and D.P. Domning. Bone Histology of the Ribs of the Archaeocetes (Mamma­ lia: Cetacea). Journal of Vertebrate Paleontology, 10(4): 455-466, 4 figures. McDonald, H.G., and CE. Ray. The Extinct Sloth, Megalo­ nyx (Mammalia: Xenarthra), from the United States Mid- Atlantic Continental Shelf. Proceedings of the Biological Society of Washington, 103(1): 1-5, 2 figures. 1991 Boms, H.W., Jr., and CE. Ray. Historic Bovid Teeth from Gardiner, Maine. Current Research in the Pleistocene, 8: 85-86. Domning, D.P., CE. Ray, and M.C. McKenna. A New Speci­ men of Behemotops proteus (Mammalia: Desmostylia) from the Oligocene of Washington. [Abstract.] Journal of Vertebrate Paleontology, supplement, 11(3):26A. McDonald, J.N., CE. Ray, and CR. Harington. Taxonomy and Zoogeography of the Musk Ox Genus Praeovibos Staudinger, 1908. In J.R. Purdue, W.E. Klippel, and B.W. Styles, editors, Beamers, Bobwhites, and Blue Points: Tributes to the Career of Paul W. Parmalee. Scientific Papers, Illinois State Museum, 23:285-314. [Also pub­ lished as The University of Tennessee, Department of Anthropology, Report of Investigations, 52.] Tedford, R.H., L.G. Barnes, and CE. Ray. Earliest Miocene Littoral Arctoid, Kolponomos. [Abstract.] Journal of Ver­ tebrate Paleontology, supplement, 11(3):57A. Ubelaker, D.H., H.E. Berryman, T.R Sutton, and CE. Ray. Dif­ ferentiation of Hydrocephalic Calf and Human Calvariae. Journal of Forensic Sciences, 36(3):801—812, 9 figures. 1992 Inuzuka, N., D.P. Domning, CE. Ray, and Y. Hasegawa. Summary of Taxa and Distribution of Desmostylia. Abstracts, 29th International Geological Congress, Kyoto, Japan, 24 August-3 September 1992, 2:349. Koretsky, LA., and CE. Ray. Fossil True Seals (Family Pho­ cidae) from the Lee Creek Mine of Southeastern USA. [Abstract.] Journal of Vertebrate Paleontology, supple­ ment, 12(3):37A-38A. Miyazaki, S., H. Horikawa, N. Kohno, K. Hirota, M. Kimura, NUMBER 93 13 Y. Hasegawa, Y. Tomida, L.G. Barnes, and CE. Ray. Summary of the Fossil Record of Pinnipeds of Japan, and Comparisons with Those from the Eastern North Pacific. Abstracts, 29th International Geological Congress, Kyoto, Japan, 24 August-3 September 1992, 2:352. Muizon, C. de, and C E . Ray. A Walrus-Like Cetacean from the Early Pliocene of Peru. [Abstract.] Journal of Verte­ brate Paleontology, supplement, 12(3):45A. Ray, CE. Foreword. In R.R. Reeves, B.S. Stewart, and S. Leatherwood, The Sierra Club Handbook of Seals and Sirenians, pages xi-xii. San Francisco: Sierra Club Books. Ray, CE. "The time has come the walrus said...." 77ie- Vir­ ginia Explorer, 8(1):2—7. 1993 McDonald, J.N., and CE. Ray. Records of Musk Oxen from the Atlantic Coastal Plain of North America. The Mosa- saur, 5:1-18, figures 1-13. Ray, CE. [Letter to Editor.] Fossil Collecting and Govern­ ment Regulation. Science, 259(5095):581. 1994 Inuzuka, N., D.P. Domning, and CE. Ray. Summary of Taxa and Morphological Adaptations of the Desmostylia. The Island Arc, 3(4):522-537. Koretsky, LA., and CE. Ray. Cryptophoca, New Genus for Phoca maeotica (Mammalia: Pinnipedia: Phocinae), from Upper Miocene Deposits in the Northern Black Sea Region. Proceedings of the Biological Society of Washing­ ton, 107(l):17-26, figures 1^1. Miyazaki, S., H. Horikawa, N. Kohno, K. Hirota, M. Kimura, Y. Hasegawa, Y. Tomida, L.G. Barnes, and CE. Ray. Summary of the Fossil Record of Pinnipeds of Japan, and Comparisons with That from the Eastern North Pacific. The Island Arc, 3(4):361-372. Ray, CE. [Letter to the Editor]. The Junkoholic. Smithso­ nian, 25(1): 12. Ray, CE. Presentation of the Harrell L. Strimple Award of The Paleontological Society to Peter J. Harmatuk. Journal of Paleontology, 68(4):920-923, 1 figure [with statements by F.C. Whitmore, Jr., and F.M. Huber; response by Peter J. Harmatuk]. Ray, C.E., D.P. Domning, and M.C. McKenna. A New Speci­ men of Behemotops proteus (Order Desmostylia) from the Marine Oligocene of Washington. Proceedings of the San Diego Society of Natural History, 29:205-222, figures 1-15. Tedford, R.H., L.G. Barnes, and CE. Ray. The Early Mio­ cene Littoral Ursoid Carnivoran Kolponomos: Systematics and Mode of Life. Proceedings of the San Diego Society of Natural History, 29:11-32, figures 1-15. 1996 McDonald, J.N., CE. Ray, and F. Grady. Pleistocene Caribou (Rangifer tarandus) in the Eastern United States: New Records and Range Extensions. In K. Stewart and K. Sey­ mour, editors, Paleoecology and Paleoenvironments of Late Cenozoic Mammals; Tributes to the Career of C. S. (Rufus) Churcher, pages 406-430. Toronto: University of Toronto Press. 1997 Arroyo-Cabrales, J., and CE. Ray. Revision de los Vampiros Fosiles (Chiroptera: Phyllostomidae, Desmodontinae) de Mexico. In Joaquin Arroyo-Cabrales and Oscar J. Polaco, editors, Homenaje al Profesor Ticul Alvares, pages 69-86. Mexico, D.F. [Mexico City]: Coleccion Cientifica, Insti- tuto Nacional de Antropologia e Historia. 2001 Ray, CE., editor. Geology and Paleontology of the Lee Creek Mine, North Carolina, III. Smithsonian Contributions to Paleobiology, 90: 365 pages, 127 figures, 45 plates, 32 tables. Ray, CE. Prodromus. In CE. Ray, editor, Geology and Pale­ ontology of the Lee Creek Mine, North Carolina, III. Smithsonian Contributions to Paleobiology, 90:1-20. Late Rancholabrean Mammals from Southernmost Florida, and the Neotropical Influence in Florida Pleistocene Faunas Gary S. Morgan A B S T R A C T The Cutler Hammock and Monkey Jungle Hammock sites, both located near the Atlantic coast in Dade County at the southern tip of the Florida peninsula, have yielded the southernmost late Pleis­ tocene (Rancholabrean) vertebrate faunas known from the conti­ nental United States. These two fossil assemblages were deposited in small sinkholes or sinkhole/cave systems that developed in the marine Pleistocene Miami Limestone when sea level was consid­ erably lower than present, probably during the latter part of the Wisconsinan glacial interval between 20 and 11 Ka. The mammalian fauna from Cutler Hammock consists of 47 species, including 16 members of the extinct Pleistocene mega­ fauna. The most common larger mammals from Cutler Hammock are Mylohyus nasutus, Equus sp., Bison antiquus, and Canis dirus. Small vertebrates, including toads, snakes, birds, bats, rodents, and rabbits, are abundant in this site. The presence of a diverse fauna of large carnivores, particularly the dire wolf, Canis dirus, and the fragmentary condition of most ungulate long bones, some of which show clear evidence of gnawing, suggest that Cutler Ham­ mock functioned, in part, as a carnivore den. Monkey Jungle Ham­ mock has 41 species of mammals, nine of which are of extinct megafauna. Small mammals, including 12 species of rodents and eight species of bats, dominate the Monkey Jungle fauna. Mammals from Cutler and/or Monkey Jungle that are unknown in Florida before the late Rancholabrean include Panthera atrox, Puma concolor, Ursus americanus, Sciurus niger, Pitymys pine- torum, Sigmodon hispidus, and Bison antiquus. The West Palm Beach site, located about 100 km north of Cutler and Monkey Jun­ gle in Palm Beach County, has 11 species of extinct megafauna among its 17 species of mammals. The most common mammals in this fauna are the proboscideans Mammut americanum and Mam­ muthus columbi. Mammals of South American origin, including species of ground sloths, armadillos, glyptodonts, vampire bats, and capyba- ras, are present in many Florida Pleistocene faunas. These South American species occur in Florida faunas along with species of North American origin that either are restricted to the Neotropics Gary S. Morgan, New Mexico Museum of Natural History, 1801 Mountain RoadNW, Albuquerque, New Mexico 87104. at the present time (e.g., jaguar, Panthera onca) or are closely related to living Neotropical species (e.g., extinct Florida cave bear, Tremarctos floridanus, and extinct tapir, Tapirus veroensis). The evolutionary history and biogeography of Florida's Pleis­ tocene mammals with Neotropical affinities are thoroughly ana­ lyzed. Introduction Florida is well known for its rich Rancholabrean vertebrate faunas, most of which are concentrated in the northern half of the peninsula (Webb, 1974a; Kurten and Anderson, 1980). The karst topography of northern Florida has strongly affected the distribution of Rancholabrean sites and has greatly increased their number, particularly faunas derived from sinkholes, fis­ sures, or caves (e.g., Arredondo sites, Devil's Den, Haile sites, Reddick, and Sabertooth Cave) and faunas found in spring-fed rivers and streams (e.g., Aucilla River, Ichetucknee River, Santa Fe River, and Waccasassa River). The only diverse late Pleistocene sites previously reported from the southern half of the Florida peninsula (e.g., Melbourne, Seminole Field, and Vero) were deposited in marshes, ponds, or rivers along the At­ lantic or Gulf coasts (Webb, 1974a; Kurten and Anderson, 1980). Two deep, water-filled sinkholes located in southwest­ ern Florida not far inland from the Gulf coast, Warm Mineral Springs (McDonald, 1990) and Little Salt Spring (Clausen et al., 1979; Holman and Clausen, 1984), have yielded limited late Rancholabrean faunas, including the ground sloth Megalo­ nyx jeffersonii, the sabertooth cat Smilodonfatalis, and the ex­ tinct giant land tortoise Hesperotestudo (=Geochelone) cras- siscutata. This study reviews three late Pleistocene (late Ranchola­ brean) mammalian faunas from southernmost peninsular Flor­ ida: Cutler Hammock, Monkey Jungle Hammock, and West Palm Beach (Figure 1). Southernmost peninsular Florida is herein defined as the southern one-third of the peninsula from Lake Okeechobee southward (south of latitude 27°N). The first 15 16 83 ( 82' SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY 8 1 ° 8 0 ° __27' 28' __26' __25' 24' FIGURE 1.—Map of southernmost peninsular Florida showing location of Rancholabrean and Holocene fossil sites discussed in text: 1 =West Palm Beach, late Rancholabrean; 2=Miami, late Rancholabrean; 3=Cutler Ham­ mock, late Rancholabrean; 4=Monkey Jungle Hammock, late Rancholabrean, and Monkey Jungle 2, Holocene; 5=Nichol's Hammock, Holocene; 6=Florida City, late Rancholabrean. fossil vertebrate fauna reported from this region was from Nichol's Hammock, a Holocene sinkhole site located south of Miami in Dade County (Hirschfeld, 1968). This site provides an interesting view of the vertebrate fauna of southern Florida prior to the extensive human impact of the last 500 years, but it completely lacks extinct species. Although located somewhat north of the study area covered in this paper, the Vero site, lo­ cated not far inland from the Atlantic coast in Indian River County, is probably the best known Rancholabrean vertebrate fauna from the southern half of the Florida peninsula (Weigel, 1962; Webb and Wilkins, 1984; Morgan, 1985). Fossils were found near West Palm Beach in Palm Beach County in 1969 during the excavation of a drainage ditch for a commercial development. Converse (1973) presented a prelim­ inary faunal list and discussion of the West Palm Beach site. This site is similar in faunal composition and depositional set­ ting to several other Florida coastal Rancholabrean sites, in particular Seminole Field, Melbourne, and Vero. Bones and teeth of extinct Pleistocene megafauna were recovered in 1969 from a small sinkhole at the Monkey Jungle tourist attraction about 25 km southwest of Miami. This discovery was summa­ rized in a popular article (Ober, 1978), but until now only a few studies on selected vertebrate taxa have mentioned the Monkey Jungle Hammock site. Cutler Hammock is by far the largest of the three south Florida Rancholabrean sites, having produced literally thousands of vertebrate fossils. This site was discov- NUMBER 93 17 ered in 1985 by archaeologists conducting survey work near Perrine south of Miami. Like Monkey Jungle, the Cutler Ham­ mock site was found in a small sinkhole on the Atlantic Coastal Ridge in an area now covered by a tropical hardwood ham­ mock. Small terrestrial vertebrates, including toads, snakes, birds, and rodents, are abundant at Cutler Hammock. Sixteen species of Pleistocene megafauna occur in this fauna. Carr (1987) presented a popular account of the Cutler Hammock site, Morgan (1991) reviewed the bat fauna, Morgan and Sey­ mour (1997) described the sample of Florida panthers (Puma concolor), and Emslie (1998) analyzed the avifauna. Emslie and Morgan (1995) summarized the taphonomy of this site, concluding that many of the large mammal remains were con­ centrated there as a result of the activities of large carnivores, particularly the dire wolf. The Cutler Hammock and Monkey Jungle Hammock sites are the two southernmost Rancholabrean vertebrate faunas in the continental United States. This fact is integrally related to the second important aspect of these two faunas stressed in this study—the number of Florida Pleistocene vertebrate taxa that have affinities with the Neotropics. Morgan (1985, 1991) noted that many of the bat species from these two sites, and from Vero, were derived from either the West Indies or Middle America. Becker (1985), Emslie and Morgan (1995), and Emslie (1998) mentioned the Neotropical affinities of several species of birds from West Palm Beach and Cutler Hammock. The strong Neotropical component in the Cutler Hammock and Monkey Jungle Hammock vertebrate faunas has prompted an overview of all Florida Rancholabrean mammals that have af­ finities with the Neotropical Region. All fossils discussed in this paper are housed in the verte­ brate paleontology collections of the Florida Museum of Natu­ ral History (formerly the Florida State Museum), University of Florida, Gainesville, Florida. More detailed locality and map data, field notes, and additional information on Cutler Ham­ mock, Monkey Jungle Hammock, West Palm Beach, and other sites described herein are available in the vertebrate paleontol­ ogy locality files of the Florida Museum of Natural History. Names and distributions for Pleistocene mammals follow Kurten and Anderson (1980) and Hulbert (2001); those for liv­ ing mammals follow Wilson and Reeder (1992). ACKNOWLEDGMENTS.—The Cutler Hammock site was exca­ vated by archaeologists from the Metro-Dade Archaeology Di­ vision and the Dade County Archaeological Conservancy un­ der the direction of Robert Carr. Their careful excavation techniques, screening of all sediments removed from the de­ posit, and detailed field notes greatly augmented the amount and quality of scientific data available from this important site. Access to the Cutler Hammock site was granted by the Deering family, who generously donated the entire Cutler collection of vertebrate fossils to the Florida Museum of Natural History. Lewis D. Ober, William G. Weaver, and several of their stu­ dents from Miami-Dade Community College excavated the Monkey Jungle Hammock site and deposited the fossils in the Florida Museum of Natural History. They screened much of the sediment from Monkey Jungle, which led to the recovery of important samples of bats and rodents. Frank Dumond, general manager of Monkey Jungle, kindly permitted access to the property. The West Palm Beach site was excavated by the late Howard H. Converse, Jr., and a large volunteer crew from the West Palm Beach area. The paleontological community owes a debt of gratitude to Howard for preserving the West Palm Beach fossils and for his many other important contributions to Florida paleontology. His untimely death saddened all of us who knew him. Russell W. Graham and S. David Webb pro­ vided helpful comments on the manuscript. Last, but certainly not least, I would like to thank my friend and colleague, Clay­ ton E. Ray. Almost 30 years ago Clayton took time out from his always-too-busy schedule to give me a guided tour of the Smithsonian's vertebrate paleontology collection and prepara­ tion lab. At the time I was an impressionable undergraduate student considering graduate study in invertebrate paleontol­ ogy. I was so captivated by vertebrate fossils that I went on to complete an undergraduate thesis on the Desmostylia, with much help and encouragement from Clayton, and then pursued graduate work and eventually a career in vertebrate paleontol­ ogy. Clayton, many thanks for all your help through the years and congratulations on a long and productive career. ABBREVIATIONS: BP before present LF local fauna NALMA North American land mammal age Ka thousands of years before present Ma millions of years before present UF Florida Museum of Natural History, University of Florida Description of Fossil Sites CUTLER HAMMOCK The Cutler Hammock site is located 13 km southwest of Mi­ ami and 4 km east of Perrine in Dade County (25°37'N, 80°19'W), near the southern tip of the Florida peninsula. Fos­ sils were discovered at Cutler Hammock in 1985 and were ex­ cavated during 1985 and 1986 by the Historic Preservation Di­ vision of Dade County under the direction of Robert Carr and by personnel from the Florida Museum of Natural History. The fossils occur in a sinkhole in a tropical hardwood hammock on the Atlantic Coastal Ridge less than 5 m above sea level and only about 0.3 km inland from the Atlantic Ocean. This sink­ hole formed in the oolitic facies of the marine late Pleistocene Miami Limestone. Since its formation in the late Pleistocene, the Cutler Hammock sinkhole has filled with sediments com­ posed of sand, clay, and pieces of limestone eroded from the walls of the sinkhole, as well as with abundant vertebrate fos­ sils. The Cutler Hammock site was carefully excavated using a square-meter grid system over the entire surface of the sink­ hole, which is approximately 6 m in diameter (Carr, 1987). 18 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY Most of the 1 x 1 m squares were excavated in arbitrary 10 cm levels to a depth of at least 2 m. The excavation levels did not follow natural stratigraphic units. The majority of the fossils were derived from the lowermost stratigraphic unit, a layer composed of brownish silt and sand, bones, and limestone frag­ ments. In places, this layer was semi-indurated by a calcareous cement. Cores indicated as much as 3 m of additional unexca- vated sediment below the local water table. All sediments re­ moved from Cutler Hammock, amounting to more than 5 met­ ric tons, were washed through 1.5 mm mesh window screen. The screening resulted in the recovery of thousands of bones and teeth of small vertebrates, which provided an important ad­ dition to the abundant samples of large vertebrates removed us­ ing standard excavation techniques. Emslie and Morgan (1995) provided a vertebrate faunal list for Cutler Hammock and discussed the taphonomy and paleo­ ecology of the site, and Morgan (1991) reviewed the chi- ropteran fauna, Morgan and Seymour (1997) described and il­ lustrated the Puma concolor fossils, and Emslie (1998) reviewed the avian fauna. The Cutler Hammock vertebrate fauna currently totals 119 species, including five fish, seven amphibians, nine reptiles, 51 birds, and 47 mammals. The fau­ nal list almost certainly will increase as more work is con­ ducted on the extensive microvertebrate sample, only a small fraction of which has been sorted and identified. In particular, additional species of toads, frogs, lizards, snakes, passerine birds, and bats almost surely will be added to the faunal list. Terrestrial vertebrates dominate the Cutler Hammock LF, both in number of species and individuals; however, the faunal list does include a few freshwater fish (gar and bowfin), sala­ manders (siren and amphiuma), aquatic turtles (mud turtle and pond turtle), water birds (loon, grebe, anhinga, and several ducks), and an otter. The most abundant species in the Cutler Hammock LF are the spadefoot toad (Scaphiopus holbrooki) and colubrid snakes, in particular the indigo snake (Drymar- chon corais). Land tortoises are common, including the extant gopher tortoise (Gopherus polyphemus) and the extinct dwarf tortoise (Hesperotestudo incisa). The absence of the extinct gi­ ant land tortoise Hesperotestudo crassiscutata is notable, as it is one of the most abundant large vertebrates encountered in Pleistocene sinkhole, fissure, and cave sites in northern Florida. The large avifauna, which includes five extinct species, reflects a diversity of habitats (Emslie and Morgan, 1995; Emslie, 1998). Two large scavenging birds found at Cutler Hammock no longer occur in Florida; the California condor (Gymnogyps californianus) is found only in the western United States and the teratorn (Teratornis merriami) is extinct. The Cutler Hammock mammalian fauna is composed of 47 species, including 16 members of the extinct Pleistocene mega­ fauna (Table 1). The most abundant ungulates in the fauna are all extinct species, including the long-nosed peccary (Mylohyus nasutus), a small species of Equus, and the extinct bison Bison antiquus. Large carnivores are relatively common in the Cutler Hammock LF, in particular, the dire wolf (Canis dirus) and the Florida cave bear (Tremarctos floridanus), both of which are extinct, and the extant jaguar (Panthera onca). The abundance of dire wolf remains and the fragmentary condition of most un­ gulate limb bones led Emslie and Morgan (1995) to hypothe­ size that the Cutler Hammock site was a carnivore den. Small mammals are common, in particular the eastern cottontail rab­ bit (Sylvilagus floridanus), the cotton rat, (Sigmodon hispidus), and the eastern woodrat (Neotoma floridana). Extant species in the fauna include two Neotropical taxa no longer found in Flor­ ida, the jaguar and the ghost-faced bat (Mormoops megalo- phylla), as well as two species of rodents, Geomys pinetis and Pitymys pinetorum, and two species of bats, Eptesicus fuscus and Myotis austroriparius, that are now restricted to the north­ ern half of the Florida peninsula (Morgan, 1991). MONKEY JUNGLE HAMMOCK The Monkey Jungle Hammock site also is located in Dade County in southernmost peninsular Florida. The site is on the property of the Monkey Jungle tourist attraction, 5 km west of Goulds and 12 km southwest of Cutler Hammock (25°34'N, 80°26'W). The fossils were discovered in a small sinkhole less than 5 m above sea level and approximately 10 km inland from the Atlantic Ocean on the Atlantic Coastal Ridge. Like Cutler Hammock, the fossils at Monkey Jungle were found in a sink­ hole in a tropical hardwood hammock developed in the marine late Pleistocene Miami Limestone. Both the Monkey Jungle and Cutler sites were formed during periods of much lower sea level and correspondingly lowered water tables. Consequently, a very late Pleistocene (late Rancholabrean) age is most likely. Lewis D. Ober, William G. Weaver, and several students from Miami-Dade Community College first found fossils at Monkey Jungle in 1969. Through their efforts the fossiliferous organic sediments were removed from the sinkhole and washed through window screens to recover the rich microvertebrate fauna. No stratigraphic units were recorded. Ober (1978) dis­ cussed the discovery of vertebrate fossils at the Monkey Jungle Hammock site and provided a preliminary faunal list. Webb (1974b) mentioned isolated teeth of the stout-legged llama (Palaeolama mirifica) from Monkey Jungle, Martin (1977) and Morgan (1985) reported the large molossid bat Eumops glauci- nus, Morgan (1991) reviewed the entire chiropteran fauna, and Morgan and Seymour (1997) described and illustrated a Puma concolor P4. The presence of a diverse fauna of large mammalian carni­ vores and the fragmentary condition of most of the larger bones suggests that, like Cutler Hammock, the Monkey Jungle Ham­ mock sinkhole may have been a carnivore den (Emslie and Morgan, 1995). The abundance of small vertebrates, including many species of frogs, lizards, snakes, birds, and small mam­ mals, indicates that avian predators, most likely owls, roosted in the sinkhole/cave system and concentrated bones there as well. Three species of cave-dwelling bats, Mormoops megalo- phylla, Pteronotus cf. P. pristinus, and Myotis austroriparius, NUMBER 93 19 TABLE 1.—Mammalian faunas from the late Pleistocene (late Rancholabrean) Cutler Hammock, Monkey Jun­ gle Hammock, and West Palm Beach local faunas, southern peninsular Florida. (x=present; -=not present; t= extinct species; #=extirpated in Florida, but still extant in Neotropical Region; *=no longer found in southern Florida, but still occurring in northern half of state.) Taxa Order MARSUPIALIA Family DlDELPHlDAE Didelphis virginiana Order XENARTHRA Family DASYPODIDAE Dasypus bellus] Family PAMPATHERIIDAE Holmesina septentrional is] Family MYLODONTIDAE Paramylodon harlani] Order INSECTIVORA Family SORICIDAE Cryptotis parva Family TALP1DAE Scalopus aquaticus Order CHIROPTERA Family MORMOOP1DAE Mormoops megalophylla# Pteronotus cf. P. pristinus] Family VESPERTILIONIDAE Eptesicus fuscus * Lasiurus sp. Myotis austroriparius* Nycticeius humeralis Family MOLOSSIDAE Eumops glaucinus floridanus Tadarida brasiliensis Order CARNIVORA Family CANIDAE Canis dirus] Canis la trans* Urocyon cinereoargenteus Family URSIDAE Tremarctos floridanus] Ursus americanus Family PROCYONIDAE Procyon lotor Family MUSTELIDAE Lutra canadensis Mephitis mephitis Spilogale putorius Family FELIDAE Lynx rufus Panthera atrox] Panthera oncaU Puma concolor Sm ilodon fatalist Order RODENTIA Family SCIURIDAE Cutler Hammock X X X X X X X - X - X X - - X X X X X X X X X X X X X X Monkey Jun- West Palm gle Hammock Beach X X X - - X - X X X X X X - X - X X - X X - X X X X - X - X X - - X X X - x - X - Taxa Glaucomys volans Sciurus carolinensis Sciurus niger Family CASTORIDAE Castor canadensis* Family GEOMYIDAE Geomys pinelis * Family HYDROCHAER1DAE Neochoerus pinckneyi] Family MURIDAE Subfamily ARVICOLINAE Neofiber alleni Pitymys pinetorum * Synaptomys australis] Subfamily SlGMODONTINAE Neotoma floridana Oryzomys palustris Peromyscus gossypinus Peromyscus polionotus Podomys floridanus Sigmodon hispidus Order LAGOMORPHA Family LEPORIDAE Sylvilagus floridanus Order PERISSODACTYLA Family TAPIRIDAE Tapirus veroensis] Family EQUIDAE Equus small sp.f Equus large sp.f Order ARTIODACTYLA Family TAYASSUIDAE Mylohyus nasutus'] Platygonus compressus] Family CAMELIDAE Hemiauchenia macrocephala] Palaeolama mirifica] Family CERVIDAE Odocoileus virginianus Family BOVIDAE Bison antiquus] Order PROBOSCIDEA Family MAMMUTIDAE Mammut americanum] Family ELEPHANTIDAE Mammuthus columbi] Total number of species Cutler Hammock X X X - X - X X - X X X X X X X - X X X X X X X X X X 47 Monkey Jun­ gle Hammock X X X X - - X X - X X X X X X X - X - X X X X X - - - 41 West Palm Beach - - - - - X - - X - - - - - X X X X - X - - X X X X X 17 have been identified from Monkey Jungle, suggesting that this fossil site was a dry cave or sinkhole/cave system in the late Pleistocene. These three species no longer occur in southern Florida, presumably because there are currently no extensive dry cave systems in this region that would provide suitable roosting sites for cavernicolous bats. At present there are no caves in Florida south of Orange and Hernando Counties (about 28°30'N) that are inhabited by bats (Rice, 1957; Jen­ nings, 1958; Morgan, 1985, 1991). During the Wisconsinan glaciation, when both sea level and local water tables were much lower than present, southernmost peninsular Florida and the Florida Keys probably consisted of a low-relief carbonate platform as much as 100 m above sea level. It is likely that an extensive karst terrain developed in southern Florida at that 20 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY time, as the Miami Limestone and Key Largo Limestone would have been highly susceptible to the formation of caves, sink­ holes, and other karst features. The Monkey Jungle Hammock and Cutler Hammock sinkholes or sinkhole/cave systems al­ most certainly formed during the Wisconsinan and subse­ quently became filled with sediments near the end of this inter­ val between 20 and 11 Ka. The mammalian fauna from the Monkey Jungle Hammock site consists of 41 species, including nine extinct members of the Pleistocene megafauna (Table 1). Like Cutler Hammock, large carnivores are a common element of the fauna, particu­ larly Canis dirus, Panthera onca, Puma concolor, and Trem­ arctos floridanus. Large ungulates are uncommon, and Bison antiquus and proboscideans are absent. Another similarity be­ tween the Monkey Jungle and Cutler faunas is the presence of two extant species of Neotropical mammals no longer found in Florida, Panthera onca and Mormoops megalophylla. Monkey Jungle records the southernmost Florida occurrence of four species of mammals that previously were unknown in the mod­ ern and late Pleistocene fauna south of Lake Okeechobee, the bats Myotis austroriparius and Eptesicus fuscus and the rodents Castor canadensis and Pitymys pinetorum. Monkey Jungle Hammock (Morgan, 1991) and Clark's Cave, Virginia (Guilday et al., 1977), each contain eight species of bats; thus they are the most diverse Pleistocene chiropteran faunas in North America (Kurten and Anderson, 1980). WEST PALM BEACH The West Palm Beach site was located in a shell rock pit near the city of West Palm Beach in Palm Beach County (26°42'N, 80°10'W), approximately 100 km north of the Cutler Ham­ mock and Monkey Jungle Hammock sites (Figure 1). Like so much of southern Florida, this area was subsequently devel­ oped for residential housing, and the fossil site no longer exists. The site was discovered in 1969 when a dragline operator dig­ ging a drainage canal unearthed the partial skeleton of a mam­ moth. The late Howard H. Converse, Jr., and a large volunteer crew spent the following two months excavating the large sam­ ple of vertebrates that now constitute the West Palm Beach LF. All of the fossils were excavated from a small area of the pit that measured 27 meters by 16 meters. Converse (1973) dis­ cussed the discovery and excavation of the West Palm Beach locality and presented a preliminary vertebrate faunal list. The vertebrate fauna from West Palm Beach differs from those of Cutler Hammock and Monkey Jungle Hammock in the dominance of aquatic vertebrates, the abundance of large ungu­ lates, and the rarity of small mammals and carnivores. The aquatic component of the West Palm Beach fauna consists of rich samples of freshwater fish, including bowfin (Amia calva), gar (Lepisosteus sp.), and catfish (Ameiurus sp.), pond turtles (Trachemys scripta and Pseudemys nelsoni), and alligators (Al­ ligator mississippiensis). Two large mammals from West Palm Beach generally are associated with freshwater aquatic sites in Florida as well (Webb, 1974a), the giant capybara (Neochoerus pinckneyi) and the Vero tapir (Tapirus veroensis). The closest living relatives of these extinct species are presently restricted to the Neotropics and prefer aquatic habitats. Most of these aquatic vertebrates are either absent or rare at the Cutler and Monkey Jungle sites. The abundance of freshwater taxa, ab­ sence of estuarine and nearshore marine species, and presence of numerous well-preserved fossils of large terrestrial verte­ brates suggest that the West Palm Beach site was deposited in a freshwater depositional environment, probably in a pond, marsh, or slow-moving stream located in a grassland savanna habitat. The 17 species of mammals identified from the West Palm Beach LF are listed in Table 1. Large mammals dominate the fauna, in particular, American mastodont (Mammut ameri­ canum), Columbian mammoth (Mammuthus columbi), and Bi­ son antiquus. Unlike Cutler and Monkey Jungle, there are nu­ merous complete mandibles, limb bones, and a few skulls of large mammals from West Palm Beach. In addition to the par­ tial mammoth skeleton mentioned above, an associated skele­ ton of a juvenile mastodont and a nearly complete skull and several mandibles of adult mastodonts also were collected. Two xenarthrans occur at West Palm Beach; the giant armadillo (Holmesina septentrionalis) is represented by a single osteo­ derm, whereas Harlan's ground sloth (Paramylodon harlani) is known from a complete mandible, humerus, and several other postcranial elements. Carnivores are rare, being represented by one identifiable element each of Canis dirus and raccoon (Pro- cyon lotor). Neochoerus pinckneyi and Tapirus veroensis are the only large mammals from West Palm Beach that are absent at Cutler and Monkey Jungle (Table 1). A skull of N. pinckneyi from West Palm Beach is one of the most nearly complete known specimens of this species (Ahearn, 1981). This site was not extensively sampled for microvertebrates, but preliminary washing suggested that small mammals were uncommon. Di- delphis virginiana, Sylvilagus floridanus, Synaptomys austra­ lis, and Sigmodon hispidus are the only small mammals known in the fauna, and all of these species are represented by limited samples. The West Palm Beach LF is the southernmost record for the extinct bog lemming (Synaptomys australis), which also is the only small mammal from this locality that is absent from both the Cutler and Monkey Jungle sites. Despite the importance of the West Palm Beach LF, it has re­ ceived scant mention in the scientific literature. Besides Con­ verse's (1973) semipopular account of the site, Robertson (1974) described a skull and two horn cores of Bison antiquus and Becker (1985) reported the avifauna. The fossil birds from West Palm Beach consist of eight species, two of which are ex­ tinct (a stork, Ciconia maltha, and a hawk eagle, Spizaetus cf. S. grinnelli), and two species that now have very restricted ranges in western North America (the California condor, Gymnogyps californianus, and the whooping crane, Grus americanus). NUMBER 93 21 OTHER SOUTH FLORIDA RANCHOLABREAN AND HOLOCENE SITES Although Cutler Hammock, Monkey Jungle Hammock, and West Palm Beach are the largest Rancholabrean faunas known from southern peninsular Florida, a few fossils are known from several other localities in this region. A few isolated vertebrate fossils of late Pleistocene age from Dade County are worthy of note (Figure 1). Two teeth of the extinct tapir, Tapirus veroen­ sis (UF 1539), found in 1918 near Florida City in Dade County about 15 km southwest of Monkey Jungle, actually are the southernmost well-documented vertebrate fossils known from the Florida peninsula. A set of mandibles with both lower m3s of Mammuthus columbi (UF 12562) was collected in 1923 dur­ ing the dredging of a ship basin within the city limits of Miami in northern Dade County. Although no associated fauna was re­ covered with the tapir or mammoth specimens, a late Rancho­ labrean age is most probable. Persistent rumors exist of mam­ moth teeth, presumably Mammuthus columbi, collected in shallow waters off the Florida Keys, but no specimens have yet been positively identified. There is every reason to believe that Rancholabrean vertebrate fossils eventually will be discovered in the Florida Keys, as these islands were directly connected to the Florida peninsula during the late Pleistocene low sea-level stand. The Nichol's Hammock LF was collected 1 km northeast of Princeton, less than 5 km southeast of Monkey Jungle in south­ ern Dade County (25°32'N, 80°25'W). Hirschfeld (1968) re­ viewed the Nichol's Hammock vertebrate fauna. Like the Mon­ key Jungle and Cutler sites, the fossils from Nichol's Ham­ mock were deposited in a small sinkhole in a tropical hard­ wood hammock. Unlike those two Rancholabrean sites, how­ ever, the Nichol's Hammock fauna consists exclusively of ex­ tant species and thus is almost certainly Holocene in age. The only mammals from Nichol's Hammock that do not currently occur in southeastern peninsular Florida are the eastern wood- rat (Neotoma floridana) and the red wolf (Canis rufus). An en­ demic subspecies of N. floridana now occurs on Key Largo, but woodrats are otherwise absent south of Lake Okeechobee. The record of the red wolf is more problematic. The Nichol's Hammock record, a mandible with a nearly complete dentition, represents the only fossil occurrence in Florida of C. rufus, a species that inhabited the state until the early twentieth century. The complete absence of C. rufus from Rancholabrean fossil sites and Holocene archaeological sites in Florida suggests that the red wolf appeared in the state fairly recently. Judging from the occurrence of C. rufus, the Nichol's Hammock LF may be only a few hundred years old (Hirschfeld, 1968), although no radiocarbon dates from the site are currently available to con­ firm or refute this hypothesis. The Monkey Jungle 2 LF and the Monkey Jungle Hammock LF are both located in the Monkey Jungle tourist attraction near Goulds in Dade County. Although both sites were depos­ ited in small sinkholes, they are quite different in age and fau­ nal composition. The Monkey Jungle 2 LF is similar to Nichol's Hammock in its lack of extinct species. A late Ho­ locene age is supported by the presence of the black rat (Rattus rattus), an Old World species introduced into Florida within the last 500 years. Morgan (1991) reported the large molossid bat Eumops glaucinus from Monkey Jungle 2, a species with Neo­ tropical affinities that still occurs in the Miami area. Age of Rancholabrean Faunas from Southernmost Florida The vertebrate fossils from both the Cutler Hammock and Monkey Jungle Hammock sites accumulated in sinkholes in the marine Pleistocene Miami Limestone. Oolites from the Mi­ ami Limestone in nearby Coral Gables, Florida, have been dated at 140-110 Ka (Osmond et al., 1965). These and similar ages for the contemporaneous Key Largo Limestone of the Florida Keys indicate that the Miami Limestone was deposited during the last (Sangamonian) interglacial, when sea level was approximately 6 m higher than present (Bloom, 1983), com­ pletely inundating the Florida Keys and much of southern pen­ insular Florida. The last interglacial high sea-level stand, which corresponds to isotope substage 5e, has been dated at many sites worldwide to between 130 and 120 Ka (Harmon et al., 1983). The Cutler Hammock and Monkey Jungle Hammock sites accumulated during the subsequent Wisconsinan glacial interval between 120 and 10 Ka, a period during which sea lev­ els were persistently lower than present. The late Rancholabrean comprises the time interval from the beginning of the last interglacial (130 Ka) until the end of the Pleistocene (10 Ka). The Cutler Hammock and Monkey Jungle Hammock local faunas are clearly late Rancholabrean by this definition and are most likely very late Pleistocene in age (20-11 Ka), corresponding to the 100 m drop in sea level dur­ ing the late Wisconsinan glacial maximum. Further refinement of the ages of these two sites may not be possible because of the lack of directly associated radiocarbon dates. Tests per­ formed by the University of Arizona radiocarbon laboratory confirmed that bones from both Cutler and Monkey Jungle are too highly leached (nitrogen content below 0.1%) to yield ac­ curate radiocarbon dates. The subtropical climate of Florida, which is characterized by high temperature, rainfall, and hu­ midity, coupled with acidic soils, accounts for the nearly ubiq­ uitous diagenetic alteration, particularly leaching, observed in bones from most Florida late Rancholabrean fossil sites. Many of these same processes, along with the intense activity of soil invertebrates, also have eliminated wood and other datable or­ ganic materials from Florida sites. The few late Rancholabrean sites from the state that have yielded accurate radiocarbon dates apparently have been continuously submerged since the end of the Pleistocene, including the Aucilla River sites in the eastern panhandle (Dunbar et al., 1989) and Little Salt Spring 22 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY (Clausen et al., 1979) and Warm Mineral Springs (McDonald, 1990), both of which are deep, water-filled sinkholes located in southwestern Florida. These underwater sites often have excel­ lent preservation of wood that has been dated to provide com­ parative radiocarbon ages for dates obtained from bone col­ lagen. Although no radiocarbon dates from Cutler Hammock were obtained from layers containing remains of extinct mammals, there is one reliable date from an overlying layer. A charcoal sample from a layer 1.25 m below the surface containing a hearth and human remains was radiocarbon dated at 9670± 120 years before present (BP) (Beta-16750), indicating a very early Holocene age (Carr, 1987). No extinct species of mammals were found in association with this radiocarbon date; however, many extinct megafaunal taxa were recovered from layers im­ mediately below the hearth. This early Holocene date provides a minimum age for the Cutler Hammock LF. A piece of mastodont rib from West Palm Beach was radio­ carbon dated at 21,150±400 years BP (Buckley and Willis, 1972; Converse, 1973), indicating a late Rancholabrean (Wis­ consinan) age. This is a whole-bone date based upon the com­ bined inorganic and organic carbon fractions of the bone. The more rigorous radiocarbon-dating techniques now applied to bones date only the bone collagen (i.e., the organic carbon frac­ tion), which constitutes about 20% of the carbon in bones. The inorganic component of bone carbon is highly subject to diage­ netic alteration and routinely gives erroneous radiocarbon dates. Although this radiocarbon date is suspect, it probably does provide a minimum date for the site. Mammalian biochronology provides the most useful infor­ mation on the age of the Cutler Hammock, Monkey Jungle Hammock, and West Palm Beach local faunas. The first ap­ pearance of Bison in North America at about 300 Ka is gener­ ally used to define the base of the Rancholabrean NALMA (Savage, 1951; Lundelius et al., 1987). Although the presence of Bison is convenient for defining the Rancholabrean, this ge­ nus is often absent in Florida Rancholabrean faunas; therefore, other faunal criteria have been established for recognizing Ran­ cholabrean faunas in Florida (Morgan and Hulbert, 1995). Mammals from the Cutler Hammock and/or Monkey Jungle Hammock local faunas (see faunal lists in Table 1) that are re­ stricted to Rancholabrean faunas in Florida are Canis dirus, Tremarctos floridanus, Ursus americanus, Smllodon fatalis, Panthera atrox (=P. leo atrox of Kurten and Anderson, 1980), Puma concolor, Sciurus niger, Pitymys pinetorum, Oryzomys palustris, Sigmodon hispidus, and Bison antiquus. All of these species are known from Cutler Hammock, and all but three, Ursus americanus, Smilodon fatalis, and Bison antiquus, occur at Monkey Jungle Hammock. Three of these species, Canis di­ rus, Sigmodon hispidus, and Bison antiquus, have been identi­ fied from West Palm Beach. Furthermore, more than one-half of these species occur only in late Rancholabrean faunas in Florida (Morgan and Hulbert, 1995): U. americanus, P. atrox, P. concolor, S. niger, P. pinetorum, S. hispidus, and B. antiquus. Rancholabrean Mammals from Southernmost Florida MARSUPIALIA.—The opossum (Didelphis virginiana) is present in the West Palm Beach, Cutler Hammock, and Mon­ key Jungle Hammock local faunas. This species is common in many Florida Rancholabrean faunas and is currently wide­ spread throughout the state. Didelphis virginiana is the only member of the Neotropical family Didelphidae found in Ran­ cholabrean faunas north of Mexico. Marsupials went extinct in North America in the middle Miocene and did not reappear un­ til Didelphis arrived from South America in the middle Pleis­ tocene (late Irvingtonian). The earliest North American records of D. virginiana are from two Florida late Irvingtonian faunas, Coleman 2A in Sumter County (Martin, 1974) and Sebastian Canal in Brevard County (Morgan and Portell, 1996). XENARTHRA.—The beautiful armadillo (Dasypus bellus) is the most common xenarthran in the two Dade County fossil sites, and it is the only member of this group present at Monkey Jungle. The giant armadillo or pampathere (Holmesina septen- trionalis) and Harlan's ground sloth (Paramylodon harlani) both occur in the Cutler Hammock and West Palm Beach fau­ nas, although neither species is common. The presence of D. bellus, H. septentrionalis, and P. harlani in southernmost Flor­ ida is not unexpected, as these are the three most common and widespread of the six species of xenarthrans known from Flor­ ida Rancholabrean faunas. The giant ground sloth Eremothe- rium laurillardi (see Cartelle and De Iuliis, 1995, for use of this species name rather than E. mirabile or E. rusconii) and the glyptodont Glyptotherium floridanum are found primarily in Florida coastal faunas and are rare in Rancholabrean sinkhole/ cave sites. Jefferson's ground sloth (Megalonyx jeffersonii) has a rather spotty distribution in Florida but occurs most often in sinkhole/cave and river sites in the northern part of the state. Five of these six genera (the cingulates Dasypus, Holmesina, and Glyptotherium and the ground sloths Eremotherium and Paramylodon=Glossotherium, in part; see McDonald, 1995) first entered Florida in the late Pliocene (late Blancan) as early participants in the Great American Faunal Interchange. The en­ tire assemblage of xenarthrans found in Florida Rancholabrean faunas went extinct at the end of the Pleistocene. The nine- banded armadillo (Dasypus novemcinctus) is now common throughout the state but is unknown from Florida Ranchola­ brean faunas. INSECTIVORA.—Two species of insectivores have been iden­ tified from Cutler Hammock and Monkey Jungle Hammock, the least shrew (Cryptotis parva) and the eastern mole (Scalo- pus aquaticus). Both species still inhabit southern peninsular Florida. The third insectivore currently found in southern Flor­ ida, the short-tailed shrew (Blarina carolinensis) is curiously absent from these two faunas. The only other insectivore known from Florida, the southeastern shrew (Sorex longiros- tris), is unknown from the southern two-thirds of the peninsula. Insectivores, in particular soricids, are more speciose in tem­ perate and boreal regions, so their lack of diversity in the late Pleistocene and modern faunas of Florida is not unexpected. NUMBER 93 23 CHIROPTERA.—Eight species of bats representing three fam­ ilies are known from Monkey Jungle Hammock (Morgan, 1991): the mormoopids Mormoops megalophylla and Pterono- tus cf. P. pristinus; the vespertilionids Myotis austroriparius, Eptesicus fuscus, Lasiurus sp., and Nycticeius humeralis; and the molossids Tadarida brasiliensis and Eumops glaucinus floridanus. Four of these species, M. megalophylla, M. austror­ iparius, E. fuscus, and N. humeralis, have been identified from Cutler Hammock. No bats occur in the West Palm Beach LF. The two mormoopids are no longer found in Florida; P. pristi­ nus is extinct and M. megalophylla is now restricted to the southwestern United States and the Neotropics. Myotis austror­ iparius and E. fuscus are presently unknown in the southern one-third of the Florida peninsula. Mormoops megalophylla, P. pristinus, and M. austroriparius almost certainly inhabited caves in southern Florida. Eptesicus fuscus commonly roosts in caves as well, especially in the West Indies, although this spe­ cies is currently rare in Florida caves. The bat faunas from both Monkey Jungle and Cutler indicate that extensive cave systems existed in southernmost Florida during the late Pleistocene, de­ spite the absence of dry caves in this region at the present time. The disappearance of these four species of bats from southern Florida probably resulted from the rise in sea level and water tables since the end of the Pleistocene, which caused the flood­ ing of formerly dry cave systems and the subsequent loss of suitable roosting sites (Morgan, 1991). Late Pleistocene and Holocene extinctions of cave-dwelling bats occurred through­ out the West Indies as well, particularly on small, low lime­ stone islands such as the Bahamas and Cayman Islands (Mor­ gan and Woods, 1986; Morgan, 1989a, 1991, 1994a, 2001). These localized extinctions of bats in the Antilles also have been attributed to the post-glacial flooding of caves that were dry during the late Pleistocene low sea-level stand. Fossils of Mormoops megalophylla from Cutler Hammock and Monkey Jungle Hammock constitute the second and third records of this species from the Pleistocene of Florida (Mor­ gan, 1991). Ray et al. (1963) and Wilkins (1983) reported M. megalophylla from the Rancholabrean Rock Springs LF in Or­ ange County about 350 km north of the Dade County fossil sites. Mormoops megalophylla is currently found in the United States only in southern Texas and Arizona. It is primarily a Neotropical species occurring from Mexico south throughout Central America to northern South America and Trinidad (Smith, 1972). Other late Pleistocene fossils of M. megalo­ phylla from outside its modern range are known from Cuba, Hispaniola, Jamaica, Abaco, Andros, and Tobago (Silva Ta- boada, 1974; Eshelman and Morgan, 1985; Morgan and Woods, 1986; Morgan, 1989a, 2001). Two mandibles tenta­ tively referred to Pteronotus pristinus, an extinct species de­ scribed from late Quaternary cave deposits in Cuba (Silva Ta- boada, 1974), were reported from Monkey Jungle by Morgan (1991). No living or fossil species of Pteronotus had been re­ corded previously from the United States, although several spe­ cies of Pteronotus are found in northern Mexico and Cuba (Smith, 1972). Three Cuban species of Pteronotus were identi­ fied from late Quaternary cave deposits on Abaco, Andros, and New Providence, although all three are now extinct in the Ba­ hamas (Morgan, 1989a, 2001). Considering the diversity of mormoopids in Cuba and Bahamas during the late Pleistocene (Silva Taboada, 1974; Morgan, 1989a, 2001), and the close proximity of these islands to Florida, it is not entirely unex­ pected to find an Antillean species of Pteronotus in a late Pleis­ tocene fossil site in southern Florida. Myotis austroriparius does not presently inhabit the southern half of the Florida peninsula. This is almost certainly related to the absence of dry caves in southern Florida, as this species usually roosts in caves for at least part of the year (Rice, 1957). Myotis austroriparius is the most abundant fossil bat in north­ ern Florida, where it has been identified from more than 10 Rancholabrean cave, sinkhole, and fissure deposits (Martin, 1972). Eptesicus fuscus is common in both the Monkey Jungle Hammock and Cutler Hammock faunas. The big brown bat is one of the most widespread bats in North America, but it is cur­ rently unknown from the southern one-third of the Florida pen­ insula and is uncommon throughout the remainder of the state. Fossils of E. fuscus also are rare in Florida, having been re­ ported from only three other Rancholabrean sites, Vero, Arredondo 2A, and Reddick IA (Morgan, 1985). A small spe­ cies of Lasiurus, either L. borealis or L. seminolus, occurs at Monkey Jungle. The fragmentary condition of the fossils does not permit a more precise identification. Lasiurus seminolus is the only small Lasiurus now found in southern Florida; how­ ever, the presence of E. fuscus and M. austroriparius at Mon­ key Jungle, both far south of their modern ranges, suggests that L. borealis also could have extended its range southward. Fos­ sils of small Lasiurus are known from two other Florida Ran­ cholabrean sites, Reddick IA and Vero (Morgan, 1985). The evening bat (Nycticeius humeralis) now occurs throughout the Florida peninsula (Layne, 1974). In addition to the Monkey Jungle record, the only other Pleistocene fossils of N. humera­ lis are from Vero (Morgan, 1985) and Baker's Bluff Cave, Ten­ nessee (Guilday et al., 1978). The Mexican free-tailed bat (Tadarida brasiliensis) occurs in the Holocene Nichols Hammock LF (Morgan, 1991), as well as Monkey Jungle Hammock. Tadarida brasiliensis is now the most abundant bat in southern Florida (Layne, 1974). This spe­ cies is not known to roost in caves in Florida (Jennings, 1958; Morgan, 1991), in marked contrast to the southwestern United States, where T. brasiliensis inhabits caves in huge numbers. Tadarida brasiliensis now roosts almost exclusively in human- made structures in Florida, although before the twentieth cen­ tury they probably lived in trees. The only other Florida Ran­ cholabrean records of T. brasiliensis are from Reddick 1A and Vero (Morgan, 1985, 1991). A distal humerus of Tadarida from the late Blancan Macasphalt Shell Pit LF in Sarasota County in southwestern Florida is not identifiable to the species level (Morgan and Ridgway, 1987; Morgan, 1991). Wagner's mastiff bat (Eumops glaucinus) occurs in both the Monkey Jungle 24 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY Hammock and Holocene Monkey Jungle 2 faunas (Martin, 1977; Morgan, 1991). Other Florida fossil records of this spe­ cies are from the Holocene stratum at Vero (Morgan, 1985) and the late Rancholabrean Melbourne LF (Allen, 1932; Ray et al., 1963; Koopman, 1971). Eumops glaucinus currently inhabits southernmost peninsular Florida, Cuba, Jamaica, and the main­ land Neotropics from Mexico to tropical South America. All Neotropical representatives of E. glaucinus have been referred to the nominal subspecies. Only the Florida population is suffi­ ciently distinct to be regarded as a different subspecies, E. glaucinus floridanus, which was first described as a fossil from Melbourne (Allen, 1932; Koopman, 1971; Eger, 1977). CARNIVORA.—Diverse carnivore assemblages characterize Cutler Hammock and Monkey Jungle Hammock, whereas only two carnivores, Canis dirus and Procyon lotor, are known from West Palm Beach (Table 1). Fourteen species of mammalian carnivores have been identified from Cutler Hammock, 10 of which also occur at Monkey Jungle. Emslie and Morgan (1995) summarized the rich carnivore fauna from Cutler Hammock and concluded that the site represented a dire wolf den, at least in part. They identified 42 individuals of Canis dirus from Cut­ ler Hammock, many of which were juveniles. Furthermore, the remains of the most abundant ungulates in the site, particularly small peccaries, consist almost entirely of isolated teeth and ends of limb bones, many of which appear to have been gnawed. Other common large carnivores at Cutler Hammock (in decreasing order of abundance) are Tremarctos floridanus, Canis latrans, Panthera onca, and Lynx rufus. Four other large carnivores are represented by one individual each: Ursus amer­ icanus, Panthera atrox, Puma concolor, and Smilodon fatalis. Cutler Hammock is one of the few Rancholabrean sites in North America that records the association of four large cats, Panthera atrox, Panthera onca, Puma concolor, and Smilodon fatalis. Among Florida's large Rancholabrean felids, only Ho- motherium serum is absent from Cutler Hammock. Homothe- rium serum has been reported from Reddick (Waldrop, 1974) but is unknown from the southern half of the state. Ursus amer­ icanus, Lutra canadensis, Mephitis mephitis, and Smilodon fa­ talis are the only carnivores in the Cutler fauna that are absent from Monkey Jungle. There is some question about the identification of Canis la- trans from Cutler and Monkey Jungle. The specimens tenta­ tively identified as C. latrans from Cutler Hammock, including maxillae, mandibles, and isolated teeth, are considerably smaller than typical western coyotes. The presence of human remains in association with extinct Pleistocene megafauna in several stratigraphic levels at the Cutler Hammock site (Carr, 1987; Emslie and Morgan, 1995) suggests the intriguing possi­ bility that the small Cutler Canis may actually represent an early domestic dog, C.familiaris. Only further careful study of the well-preserved fossils of small Canis from Cutler Ham­ mock, as well as of the less complete fossils from Monkey Jun­ gle and several other Florida Rancholabrean sites, will clarify their taxonomic status. RODENTIA AND LAGOMORPHA.—Thirteen species of rodents have been identified from Cutler Hammock or Monkey Jungle Hammock (Table 1). Eleven of these species are shared be­ tween the two faunas, and each site has one rodent not found in the other. These latter two species are the beaver (Castor ca­ nadensis) from Monkey Jungle and the pocket gopher (Geomys pinetis) from Cutler Hammock. Both of these rodents are now restricted to the northern half of the Florida peninsula. Geomys pinetis has specialized habitat requirements, specifically dry, loose sandy soils for burrowing, that appear to be lacking in southern peninsular Florida (Wilkins, 1984). The pine vole (Pitymys pinetorum) occurs at both Cutler and Monkey Jungle, but like the beaver and pocket gopher it is now absent in south­ ern Florida. Two other rodents identified from these two sites, the Florida mouse (Podomys floridanus) and the beach mouse (Peromyscus polionotus) are now absent or are very rare in southernmost peninsular Florida. The most abundant rodent in the Cutler and Monkey Jungle faunas is the cotton rat (Sigmo­ don hispidus), which also is one of the most common living ro­ dents in southern Florida. Sigmodon hispidus also occurs in the West Palm Beach LF, as do two other rodents, Neochoerus pinckneyi and Synaptomys australis, neither of which is known from Cutler or Monkey Jungle. West Palm Beach is the south­ ernmost Florida record for both of these extinct species. The presence in the Cutler and Monkey Jungle sites of all three species of squirrels that currently inhabit the Florida pen­ insula is notable. Sciurids are generally rare or absent in most Florida Rancholabrean faunas. Sciurus carolinensis and S. ni­ ger are not particularly common in the two Dade County sites, but flying squirrels (Glaucomys volans) are second only to Sig­ modon hispidus in abundance among rodents at Monkey Jungle and also are fairly common at Cutler. Flying squirrels are known from several other Florida late Pliocene and Pleistocene sinkhole or cave sites. A possible reason for the abundance of flying squirrels in sinkhole/cave sites is predation by owls, par­ ticularly the barn owl (Tyto alba). The eastern woodrat (Neotoma floridana) is one of the most common rodents at both Cutler and Monkey Jungle. This spe­ cies no longer occurs in the southern one-third of the Florida peninsula; however, an isolated endemic subspecies, N. flori­ dana smalli, inhabits Key Largo, the northernmost of the Flor­ ida Keys, located about 30 km south of the Dade County sites. The Cutler and Monkey Jungle records of N. floridana, along with a Holocene record from Nichol's Hammock (Hirschfeld, 1968), indicate that the woodrat's range extended to the south­ ern tip of the Florida peninsula until recently. The cottontail rabbit (Sylvilagus floridanus) occurs in the Cutler Hammock, Monkey Jungle Hammock, and West Palm Beach faunas and is especially common in the Cutler site. The marsh rabbit (S. palustris) has not been identified from any of these sites, even though it is fairly common in southern penin­ sular Florida and is one of the few indigenous mammals known from the Florida Keys (Lazell, 1989). The absence of marsh rabbits at Cutler and Monkey Jungle is probably related to this NUMBER 93 25 species' preference for wet—marshy or swampy—habitats. Most of the vertebrates from these two sites are more indicative of drier forested or savanna habitats. PERISSODACTYLA.—Two species of horses of the genus Eq­ uus are known from Cutler Hammock, one of which also oc­ curs in the Monkey Jungle Hammock and West Palm Beach faunas. Because of the taxonomic confusion that plagues the North American members of this genus, the Equus fossils from these three sites are not identified to species. The smaller horse from the Cutler site belongs to the E. alaskae group of Equus species, whereas the larger species belongs to the E. laurentius group, according to Winans's (1989) study of North American fossil Equus. The smaller Equus is the second most abundant ungulate at Cutler Hammock. The most interesting aspect of the Cutler horse fossils is that virtually the entire sample con­ sists of isolated deciduous premolars and unerupted or newly erupted premolars and molars. Considering the abundance of Canis dirus in the Cutler fauna, Emslie and Morgan (1995) hy­ pothesized that juvenile horses probably constituted a signifi­ cant portion of the dire wolves' diet, exceeded in numbers only by peccaries. There is a small sample of Tapirus veroensis from West Palm Beach. Tapirs are absent from Cutler and Monkey Jungle, probably because conditions at these sites were more favorable to terrestrial taxa. ARTIODACTYLA.—Six species of artiodactyls have been identified from the Cutler Hammock LF. Five of these species occur at Monkey Jungle and four are known from West Palm Beach. The most abundant large mammal from Cutler Ham­ mock is the long-nosed peccary (Mylohyus nasutus). Emslie and Morgan (1995) counted 99 individuals of M. nasutus from Cutler Hammock; however, that number will increase substan­ tially when the fauna is completely curated. The Cutler Mylo­ hyus sample consists of hundreds of isolated adult and juvenile teeth, a small sample of mandibular and maxillary fragments, and numerous ends of long bones, carpals, tarsals, and phalan­ ges. The fragmentary condition of most of the M. nasutus fos­ sils from the Cutler site, as well as the large number of gnawed limb ends of this species, led Emslie and Morgan (1995) to hy­ pothesize that peccaries composed a large portion of the diet of the carnivores that inhabited the cave, particularly Canis dirus. A few fossils of M. nasutus are known from Monkey Jungle Hammock and West Palm Beach. Florida's other Ranchola­ brean tayassuid, the flat-headed peccary, Platygonus compres­ sus, is uncommon at Cutler and Monkey Jungle and is absent from West Palm Beach. Camels are not particularly common in any of the three fau­ nas under discussion. Both the large-headed llama (Hemi- auchenia macrocephala) and the stout-legged llama (Palae- olama mirifica) are represented by small samples of isolated teeth from Cutler and Monkey Jungle. Webb (1974b) listed P. mirifica from Monkey Jungle, and Converse (1973) mentioned this species from West Palm Beach, the only species of camel from the latter fauna. White-tailed deer (Odocoileus virgin- ianus) are fairly common in all three faunas. The extinct bison Bison antiquus is among the most abundant large mammals in both the Cutler Hammock and West Palm Beach faunas. No horn core is known from the Cutler site, where the bison sam­ ple consists almost exclusively of isolated teeth, and thus the identification of B. antiquus is based upon the late Ranchola­ brean age of the fauna. A partial bison skull with both horn cores from West Palm Beach is definitely identifiable as B. an­ tiquus (Robertson, 1974). PROBOSCIDEA.—Both species of proboscideans found in Florida late Rancholabrean sites, the American mastodont (Mammut americanum) and the Columbian mammoth (Mam­ muthus columbi), occur at Cutler Hammock and West Palm Beach. No proboscideans are known from Monkey Jungle. The samples of these two species from Cutler Hammock are very sparse, consisting of isolated enamel plates and one deciduous premolar of M. columbi and a single enamel fragment of M. americanum. Both mastodonts and mammoths are common at West Palm Beach. The M. americanum sample includes the complete skeleton of a juvenile mounted at the West Palm Beach Science Center, as well as a complete skull and several mandibles in the UF collection. A partial M. columbi skeleton also was discovered at this site. The rarity of mastodonts and mammoths at Cutler and their absence at Monkey Jungle is not unexpected, as these two species are generally uncommon in Florida Rancholabrean sinkhole and cave sites. Florida Neogene and Recent Vertebrates with Neotropical Affinities Florida has long been known for its rich Pleistocene fauna of Neotropical mammals, including ground sloths, glyptodonts, giant armadillos, and capybaras, among others. All of these groups of mammals evolved in South America. Most of them first reached North America in the late Pliocene and early Pleistocene (late Blancan and early Irvingtonian) as partici­ pants in the Great American Faunal Interchange, which oc­ curred during the million-year period following the final con­ nection of these two continents at about 2.5 Ma. Late Blancan and early Irvingtonian sites from Florida document some of the earliest arrivals in North America of South American mammals that dispersed northward across the Panamanian isthmus fol­ lowing the onset of the Great American Faunal Interchange (Webb, 1976, 1985; Webb and Wilkins, 1984; Morgan and Hulbert, 1995). South American immigrants recorded from Florida Blancan and Irvingtonian sites include the armadillos Dasypus, Holmesina, and P achy ar mother ium; the glyptodont Glyptotherium; the ground sloths Eremotherium, Paramylodon (^Glossotherium, in part), and Nothrotheriops; the porcupine Erethizon; the capybaras Hydrochaeris and Neochoerus; and the giant flightless bird Titanis. After the northward flow of South American immigrants essentially ceased in the early Irv­ ingtonian (after about 1.5 Ma), most of these genera became in­ tegral members of the temperate North American fauna and un­ derwent evolutionary and biogeographic changes unrelated to 26 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY their Neotropical origin. Some of these genera quickly spread throughout North America (e.g., Paramylodon and Erethizon), whereas others were restricted to more subtropical climates in the southeastern United States, particularly Florida (e.g., Holm­ esina, Pachyarmatherium, Eremotherium, Hydrochaeris, and Titanis). Among the 11 genera listed above, which constitute most of the South American contribution of large vertebrates to the Great American Interchange, all but two, Pachyarmathe­ rium and Titanis, survived in Florida until the major megafau­ nal extinctions at the end of the Pleistocene. Dasypus and Ere­ thizon are still extant in North America, although the porcupine is no longer found in Florida, and the capybara Hydrochaeris still inhabits parts of tropical Middle America and South Amer­ ica. The bizarre armadillo, Pachyarmatherium leiseyi, a new genus and species recently described from the Leisey Shell Pit and several other Florida sites (Downing and White, 1995), is known only from the late Blancan and early Irvingtonian. Ti­ tanis walleri, the only phorusrhacid bird known outside of South America, is restricted in Florida to three late Pliocene (late Blancan) sites (Brodkorb, 1963; Chandler, 1994). A spec­ imen recently collected from the Gulf Coastal Plain of southern Texas (Baskin, 1995) represents the only Pleistocene record of the genus. To better understand the geographic origin of Florida's mam­ mals with Neotropical affinities, it is helpful to define the boundaries and subdivisions of the Neotropical Region. It should be emphasized that these boundaries are based upon the living fauna and flora, although many vertebrate paleontolo­ gists have used the term "Neotropical" to refer to extinct verte­ brates originally derived from South America. The latter usage may be misleading because several species with South Ameri­ can origins are restricted to the Nearctic Region in the late Pleistocene. Hershkovitz (1958) divided the Neotropical Re­ gion into the Patagonian, Brazilian, and West Indian Subre- gions. The Patagonian Subregion is composed of temperate South America, the West Indian Subregion includes the Anti­ llean islands, and the Brazilian Subregion comprises most of the mainland Tropics between Mexico and Brazil. Hershkovitz further subdivided the Brazilian Subregion into four provinces. The Middle American Province includes the northern boundary of the Neotropical Region in Mexico. The Nearctic Region in­ corporates all of temperate North America, including northern Mexico, the continental United States, Canada, and Alaska. Koopman and Martin (1959) placed the northeastern bound­ ary of the Neotropical mammal fauna between 23°N and 24°N latitude in the southern half of the state of Tamaulipas, Mexico, corresponding to the disappearance of tropical deciduous for­ ests. Neotropical mammals, particularly bats, extend their ranges somewhat farther north along the Pacific coast of west­ ern Mexico, with the northern limit of the Neotropics at about 28°N in southern Sonora and southern Chihuahua. There is no strong faunal or floral evidence to suggest that the northern boundary of the Neotropical Region extended much farther north or east during the late Pleistocene. Another explanation must be proposed for the rather common occurrence of what are now considered Neotropical mammals, such as jaguar, oce­ lot, margay, and several bats, in Rancholabrean faunas in Flor­ ida. Either Neotropical species expanded their ranges north­ ward into temperate North America in the late Pleistocene or they inhabited this region throughout the Pleistocene only to have their ranges retract southward into tropical America over the past 10,000 years. The Pleistocene and modern distribu­ tions of Florida Rancholabrean mammals should elucidate cer­ tain patterns, although many species, particularly bats and other small mammals, have inadequate fossil records to assess the full extent of their Pleistocene distribution accurately. The evolutionary and biogeographic history of the numerous Florida mammals with Neotropical connections reveals some interesting trends. The most obvious species with Neotropical affinities are those of South American origin. These include Didelphis, all members of the Xenarthra, and all North Ameri­ can caviomorph rodents. This group can be further subdivided because South American mammals reached temperate North America during three widely separate time intervals: late Mi­ ocene, late Pliocene/early Pleistocene, and late Pleistocene/Ho­ locene. Two unrelated ground sloths, the mylodont Thinoba- distes and the megalonychid Pliometanastes, appeared in North American faunas at the beginning of the Hemphillian NALMA (late Miocene, about 9 Ma), long before the final connection of the two continents (Hirschfeld and Webb, 1968; Webb, 1989). These early arrivals apparently island-hopped or rafted across the water barrier separating North and South America in the late Miocene (Webb, 1976, 1985). At about this same time, procyonids dispersed from North America to South America (Baskin, 1989a). Earlier in the Miocene, megalonychid sloths from South America also reached the West Indies, where they are known from the early Miocene of Cuba (MacPhee and Itur- ralde-Vinent, 1994) and the late Quaternary of Cuba, Hispani­ ola, and Puerto Rico (Paula Couto, 1967; Morgan and Woods, 1986). The genus Thinobadistes went extinct in North America in the late Hemphillian (Webb, 1989). The mylodonts that reached North America in the late Blancan, the Glossotherium- Paramylodon lineage (see McDonald, 1995, for usage of Paramylodon for most North American mylodonts more com­ monly referred to Glossotherium), are unrelated to the older Thinobadistes. All temperate North American megalonychids, including two species of Pliometanastes and five species of Megalonyx, appear to be derived from the original megalony­ chid that emigrated to North America in the early Hemphillian. Although Thinobadistes, Pliometanastes, and Megalonyx were derived from South America, these three genera are unknown from that continent. Moreover, most species of North Ameri­ can mylodonts and megalonychids are restricted to fossil sites in the Nearctic Region. The only members of these two fami­ lies known from tropical Middle America are two species of megalonychids, Megalonyx obtusidens and Meizonyx salvador- ensis, described from the early Pleistocene of El Salvador NUMBER 93 27 (Webb and Perrigo, 1985), and the mylodont Glossotherium garbanii from central Mexico (Montellano-Ballesteros and Carranza-Castaneda, 1986). Table 2 attempts to demonstrate both the origin and the bio­ geographic distribution of the mammals with Neotropical af­ finities recorded from Florida Rancholabrean sites. This table lists all species of Florida Rancholabrean mammals that have some connection with the Neotropical Region and evaluates their evolutionary origin (i.e., South America or North Amer­ ica) and present or late Pleistocene distribution. Many of these species have been lumped together in discussions of Florida Neotropical mammals, when they actually have quite different evolutionary and/or biogeographic histories. Florida's Neotro­ pical mammals fall into three broad categories: (1) species of South American origin that either reached North America in the late Miocene after crossing the Central American Seaway or entered North America as participants in the Great American Faunal Interchange in the late Pliocene or early Pleistocene; (2) species of North American origin that have a predominantly Neotropical distribution at the present time; and (3) species of North American origin that now occur primarily in the Nearctic Region but do extend their ranges into the northernmost Neo­ tropical Region in Mexico and Central America. MAMMALS OF SOUTH AMERICAN ORIGIN Didelphis virginiana is one of only two non-volant mammals of South American origin that reached temperate North Amer­ ica after the early Irvingtonian. The earliest North American records of D. virginiana are from two Florida late Irvingtonian sites, Coleman 2A (Martin, 1974) and Sebastian Canal (Mor­ gan and Portell, 1996). Virtually all fossil occurrences of the Virginia opossum outside of Florida are from the late Rancho­ labrean (Kurten and Anderson, 1980). Didelphis virginiana currently occurs widely throughout the Nearctic Region and in the Middle American Province of the Neotropical Region as far south as Costa Rica. Most Rancholabrean Xenarthra found in Florida, although of South American origin, occur primarily in fossil sites in tem­ perate North America. Two of the four species of Ranchola­ brean ground sloths, Megalonyx jeffersonii and Paramylodon harlani, are widespread throughout North America, and M. jef­ fersonii is known from sites as far north as central Alaska and the Yukon Territory of Canada (Kurten and Anderson, 1980). A third species, the Shasta ground sloth (Nothrotheriops shasten- sis), is found principally in the arid southwest. The only North American Rancholabrean ground sloth that ranged widely into the Neotropics is the giant megatheriid Eremotherium lauril- lardi. Cartelle and De Iuliis (1995) synonymized all of the large late Pleistocene species of Eremotherium from North America, Central America, and South America, including E. mirabile and E. rusconii, under the oldest available name, E. laurillardi. A single Pan-American species of Eremotherium therefore occurred on the Atlantic and Gulf Coastal Plains in the southeastern United States from South Carolina to Florida to Texas, south through Middle America (Gazin, 1956; Webb and Perrigo, 1984), and throughout much of South America as far south as southern Brazil (Cartelle and De Iuliis, 1995). The three shelled xenarthrans recorded from Florida Rancholabrean faunas, Dasypus bellus, Holmesina septentrionalis, and Glyp­ totherium floridanum, are primarily found in fossil sites on the Atlantic and Gulf Coastal Plains in the southeastern United States. The last xenarthran to reach temperate North America, the living nine-banded armadillo (Dasypus novemcinctus), is unknown in the Nearctic Region before the Holocene. This ar­ madillo dispersed naturally into the Florida panhandle in the 1970s, although it was introduced into the peninsula earlier in the last century (Humphrey, 1974). Eleven species of bats identified from Florida Rancholabrean sites are included in Table 2, as are two living bats unknown as fossils from the state. Morgan (1991) presented a detailed dis­ cussion of late Pliocene, Pleistocene, and extant bats from Flor­ ida with Neotropical affinities, so only a brief synopsis is pre­ sented herein. Bats are the only group of mammals listed in Table 2 for which the origin of some species may be in ques­ tion. This uncertainty results from the poor Tertiary fossil record of bats in North and South America. Although the only pre-Pleistocene fossil record of the Mormoopidae is from the late Oligocene and Miocene of Florida (Morgan, 1989b), Mor­ moops megalophylla and Pteronotus cf. P. pristinus are listed as being of Neotropical origin in Table 2 because all living spe­ cies of the family are restricted to the Neotropics. Molossids are known from the Miocene of Florida (Morgan, 1989b), but the great majority of New World members of this family are Neotropical in distribution. Molossids have been identified from the Oligocene and Miocene of South America (Czap­ lewski, 1997). The three molossids known from Florida Ran­ cholabrean sites, Eumops glaucinus, E. underwoodi, and Ta­ darida brasiliensis, probably were derived from the Neotro­ pics. The two Florida phyllostomids listed in Table 2, the extinct vampire bat (Desmodus stocki) and the living Jamaican fruit-eating bat (Artibeus jamaicensis), are almost certainly South American in origin. The five Florida vespertilionids that occur in the Neotropics, Eptesicus fuscus, Lasiurus borealis, L. intermedius, Nycticeius humeralis, and Pipistrellus subflavus, are clearly of North American origin. The Vespertilionidae is the best-represented bat family in the Tertiary of North Amer­ ica (Morgan, 1989b), whereas vespertilionids are unknown in South America prior to the Pleistocene (Czaplewski, 1997). The late Pleistocene and modern distributions of the 13 spe­ cies of bats in Table 2 are more indicative of the bats' biogeo­ graphic affinities than of their origins. Five of the bats, Mor­ moops megalophylla, Artibeus jamaicensis, Eumops glaucinus, Eumops underwoodi, and Molossus molossus, are living species that are principally restricted to the Neotropical Region. Mor­ moops megalophylla is now found in the United States only in southern Texas and Arizona. The majority of its range is in the Neotropics, from Mexico south throughout Central America to 28 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY TABLE 2.—Florida Rancholabrean mammals with Neotropical affinities. With three exceptions, all species listed in this table have been identified, on the basis of Webb (1974a), Hulbert (2001), and Morgan and Hulbert (1995), from one or more Florida Rancholabrean faunas. For the sake of completeness, this list includes three species of mammals (Dasypus novemcinctus, Artibeus jamaicensis, and Molossus molossus) with Neotropical affinities that occur in the modern fauna of Florida but are unknown as fossils. Each species is evaluated according to its evo­ lutionary origin (South America or North America) and its late Pleistocene or Holocene distribution. For extinct species, the distribution indicated is the late Rancholabrean distribution; for extant species it is the present distri­ bution. Many of the distributional patterns of the individual species are not strictly Neotropical or Nearctic, hence the modifier "primarily." The third category is for species that occur widely in both biogeographic regions. A number of the species have peculiarities concerning their origin, distribution, or taxonomy. These are ex­ plained either in footnotes to this table or in the text. (x=present; -=not present; t=extinct species; #=extirpated in Florida, but still extant in the Neotropical Region.) Taxa Origin Late Pleistocene/Holocene distribution Neotropical (South/ Nearctic (North Middle America) America) Primarily Neotropical Primarily Widely in both Nearctic regions MARSUPIAUA Didelphis virginiana XENARTHRA Dasypus bellus] Dasypus novemcinctus Holmesina septentrionalis] Glyptotherium floridanum] Eremotherium laurillardi]' Megalonyx jeffersonii] Paramylodon harlani] INSECTIVORA Cryptotis parva CHIROPTERA Mormoops megalophylla^ Pteronotus pristinus] Artibeus jamaicensis Desmodus stocki] Eptesicus fuscus Lasiurus borealis Lasiurus intermedius Nycticeius humeralis Pipistrellus subflavus Eumops glaucinus Eumops underwoodi^ Molossus molossus Tadarida brasiliensis CARNIVORA Canis dirus] Canis latrans2 Urocyon cinereoargenteus Tremarctos floridanus]^ Monachus tropicalis] Procyon lotor Conepatus leuconotus# Conepatus robustus] Mus tela frenata Spilogale putorius Leopardus pardalis§ Leopardus wiedii/amnicolatt]4 Lynx rufus Panthera atrox] Panthera oncaU Puma concolor Smilodon fatalis] RODENTIA Erethizon dorsatum5 Hydrochaeris holmesi] Neochoerus pinckneyi] Glaucomys volans x? X x x X NUMBER 93 29 TABLE 2.—Continued. Taxa Oryzomys palustris6 Podomys floridanus Sigmodon hispidus LAGOMORPHA Sylvilagus floridanus PERISSODACTYLA Tapirus veroensis]-* ARTIODACTYLA Mylohyus nasutus]* Platygonus compressus]* Hemiauchenia macrocephala]* Palaeolama mirifica]* Odocoileus virgin ianus Bison antiquus] Bison latifrons] PROBOSCIDEA Mammut americanum] Cuvieronius tropicus] Mammuthus columbi] SIRENIA Trichechus manatus Neotropica Orig (South/ Middle America) X - - - - - - - - - - - X n Nearctic (North America) X X X X X X X X X X X X X X X - Late Pleistocene/Holocen Primarily Primarily Neotropical - - - - - - - - - - - - - X Nearctic X X - - X X X X X - X X X - X - e distribution Widely in both regions - - X X - - - X - - - X - - 'Cartelle and De Iuliis (1995) recently synonymized all North, Central, and South American late Pleistocene Er­ emotherium under the single Pan-American species, E. laurillardi. 2May be referable to Canis familiaris instead of C. latrans; see discussion in text. 3These extinct species are known primarily from Rancholabrean sites in North America north of Mexico and are of Nearctic origin; they are included in this list because their closest living relatives are restricted to the Neotro­ pical Region. 4There is some disagreement as to whether this small cat represents an extinct species, Leopardus amnicola. known only from the late Rancholabrean of Florida, or is an extinct subspecies of the Neotropical margay, L. wiedii. 5The porcupine Erethizon dorsatum is a caviomorph rodent of South American origin, but it is now restricted to the Nearctic Region and is unknown from the Holocene fauna of Florida. 6The genus Oryzomys is ultimately of North American origin, but the living rice rat, O. palustris. apparently evolved in Middle or South America and invaded the southeastern United States in the Rancholabrean. northern South America and Trinidad (Smith, 1972). Lazell and Koopman (1985), Lazell (1989), and Frank (1997a) reported an isolated population of the Neotropical fruit bat Artibeus jamai­ censis in the Florida Keys, substantiating a nineteenth-century record of this species from Key West. Artibeus jamaicensis is one of the most widespread of all Neotropical phyllostomids, ranging from the northern limits of the Neotropical Region in Mexico and the West Indies south to northern South America. Artibeus jamaicensis is unknown as a fossil in Florida, suggest­ ing that this species dispersed to the Florida keys from Cuba sometime during the Holocene. With the exception of the records from southernmost peninsular Florida (Koopman, 1971; Eger, 1977; Belwood, 1981), Eumops glaucinus has a strictly Neotropical distribution. It occurs from southern Mexico to tropical South America, as well as in Cuba and Jamaica. Eu­ mops underwoodi is found in the northern part of the Middle American Province from Nicaragua to Mexico and also extends its range northward into southern Arizona (Morgan, 1991). The small molossid Molossus molossus recently was reported from three islands in the Florida Keys, representing the first records of this bat from the United States (Frank, 1997b). Molossus mo­ lossus is a widespread Neotropical bat found from northern Mexico south to southern South America and throughout the West Indies. Frank (1997b) identified the bats from the Florida Keys as the Cuban subspecies M. molossus tropidorhynchus. Molossus molossus has no fossil record in Florida, indicating that, like Artibeus jamaicensis, this species probably reached the Florida Keys by overwater dispersal from Cuba during the Ho­ locene. Each of these five bat species has a marginal range in the southernmost portion of the Nearctic Region, either in extreme southern Florida (A. jamaicensis, E. glaucinus, and M. molos­ sus) or in southern Arizona and/or Texas just north of the Mexi­ can border (M. megalophylla and E. underwoodi). Tadarida brasiliensis also probably belongs with this group of bats. It has a wider distribution in the southern United States than the previ­ ous five bats and also occurs throughout most of the Neotropical Region from Mexico south to Argentina and Chile and on many islands in the West Indies. One of the most interesting mammals of Neotropical origin recorded from Florida Rancholabrean faunas is Desmodus 30 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY stocki, a large extinct species of vampire bat well known from Reddick and several other late Pleistocene faunas in the north­ ern half of the state (Gut, 1959; Olsen, 1960; Morgan et a l , 1988; Ray et al., 1988; Morgan, 1991). Desmodus stocki is considerably larger than the closely related living Neotropical vampire bat, D. rotundus, which is distributed throughout much of the Neotropical Region from its northern limits in Mexico south to Argentina and Chile. Desmodus rotundus does not presently occur in the West Indies, but it is known from late Quaternary deposits on Cuba (Koopman, 1958). Desmodus stocki is known primarily from late Pleistocene sites in the southern United States and northern Mexico (Ray et al., 1988), almost all of which are located in the Nearctic Region. The combination of the larger body size of D. stocki and the milder winters that characterized the southern part of the Nearctic Re­ gion in the late Pleistocene may account for the more temperate distribution of this extinct vampire, compared with the distribu­ tions of the three living species of vampire bats, all of which are confined to the Neotropics (McNab, 1973; Morgan, 1991). A smaller and more primitive species of Desmodus, D. archae- odaptes, was described from one latest Pliocene and two early Pleistocene localities in northern peninsular Florida (Morgan et al., 1988). Desmodus apparently was a participant in the Great American Faunal Interchange, presumably following one of its principal food supplies (large xenarthrans?) northward in the late Pliocene. Before the discovery at Monkey Jungle Hammock of Pter­ onotus pristinus, a bat with indisputable Antillean affinities (Morgan, 1991), a Middle American origin was proposed for most species of Florida's Neotropical bats (Webb and Wilkins, 1984; Morgan, 1985). The occurrence of P. pristinus in the Rancholabrean of Florida suggests the possibility of a West In­ dian origin for several other bats found in the state (Morgan, 1991). Pteronotus pristinus is an extinct species known else­ where only from the late Quaternary of Cuba (Silva Taboada, 1974), the most likely source for West Indian bats that may have dispersed to Florida. Cuba is almost certainly the source for populations of Artibeus jamaicensis and Molossus molossus that are known from the Florida Keys (Lazell and Koopman, 1985; Lazell, 1989; Frank, 1997a, 1997b), particularly consid­ ering how near Cuba is to the Florida Keys and the great dis­ tance separating Florida from the closest mainland populations of these two species. Two other Florida Neotropical bats, Mor­ moops megalophylla and Eumops glaucinus, may have been derived from the West Indies. Mormoops megalophylla is known from late Quaternary cave deposits in the Bahamas, Cuba, Hispaniola, Jamaica, and Tobago but is now extinct in the West Indies (Silva Taboada, 1974; Eshelman and Morgan, 1985; Morgan and Woods, 1986; Morgan, 1989a, 2001). The Florida M. megalophylla cannot be confidently assigned to ei­ ther the extant mainland Neotropical or the extinct West Indian populations of this species, and its origin within the Neotropi­ cal Region is therefore unclear. Likewise, it is not possible to determine whether the endemic Florida subspecies, Eumops glaucinus floridanus, was derived from the mainland Neotro­ pics or from the Greater Antilles, specifically Cuba, because both regions are inhabited by the nominal subspecies, E. g. glaucinus (Eger, 1977). Both Middle American and Antillean origins for E. glaucinus have be"en postulated (Layne, 1974; Baker and Genoways, 1978; Morgan, 1985). Tadarida brasil­ iensis also occurs throughout the Greater Antilles and Baha­ mas, but the various West Indian subspecies are smaller than the Florida race, which is more closely related to T. brasiliensis from the southwestern United States and Mexico. Tadarida is first recorded in Florida from the late Pliocene (late Blancan); thus, it appears likely that T. brasiliensis, or a very closely re­ lated species, reached Florida by way of Middle America as an early participant in the Great American Faunal Interchange (Morgan and Ridgway, 1987). It is surprising that Antillean bats have not been reported from southern Florida more frequently considering the close proximity of Cuba and the Bahamas. The Florida Keys are 150 km north of Cuba, and less than 100 km would have separated Florida from Cuba and the Bahamas during the late Pleistocene low sea-level stand. Layne (1974) offered several suggestions for the rarity of West Indian bats in the modern fauna of south­ ern Florida, including their inability to cross water barriers and the paucity of suitable tropical habitats. Layne's hypothesis may explain why some West Indian bats, particularly frugivo- rous phyllostomids, do not occur in southern Florida. The ab­ sence of dry caves in the southern two-thirds of the peninsula probably accounts for the absence of cave-dwelling bats, such as mormoopids, natalids, and brachyphylline phyllostomids. The three genera of South American rodents known from Florida, Hydrochaeris, Neochoerus, and Erethizon, all first oc­ curred in the state in the late Blancan (Webb, 1974a, 1976; Ahearn and Lance, 1980; Ahearn, 1981; Morgan and White, 1995). All three genera occur sporadically in Florida sites throughout the remainder of the Pliocene and Pleistocene and disappear at the end of the Rancholabrean. Hydrochaeris holm- esi is known from the late Pliocene through the Pleistocene in Florida. Neochoerus dichroplax occurs in several Florida late Blancan faunas (Ahearn and Lance, 1980), whereas N. pinck­ neyi appears to be restricted to the Rancholabrean (Ahearn, 1981; Morgan and White, 1995). The extant genus Hydrocha­ eris and the extinct Neochoerus also were widespread in South America in the Pleistocene, but the species H. holmesi and N. pinckneyi are found only in the southeastern United States. Ere­ thizon is clearly of South American origin, although the genus is unknown from the Neotropical Region. Species of the prehen­ sile-tailed porcupine, Coendou, are the typical Neotropical por­ cupines. The oldest Florida records of porcupines are Erethizon poyeri from the late Blancan Haile 7C LF (Hulbert, 1997) and the small endemic species, E. kleini, from the latest Blancan In­ glis IA LF (Frazier, 1981). The living porcupine (E. dorsatum) was fairly widely distributed in Florida during the Irvingtonian, but there are just three verified Rancholabrean records of this species from the state, Aucilla River, Seminole Field, and Wac- NUMBER 93 31 casassa River (Frazier, 1981; Morgan and White, 1995). Porcu­ pines no longer occur in Florida; the nearest populations in cen­ tral Texas and the Appalachian Mountains of eastern Tennessee. Erethizon dorsatum is now found primarily in temperate conif­ erous forests in the Nearctic Region, although porcupines also occur in the southwestern desert region. MAMMALS OF NORTH AMERICAN ORIGIN All species listed in Table 2 belonging to the Insectivora, Carnivora, Lagomorpha, Perissodactyla, Artiodactyla, and Pro­ boscidea, as well as the sigmodontine rodents and vespertili- onid bats, are of North American origin. The remainder of this discussion will attempt to characterize their evolutionary and biogeographic relationships. A few mammals of North Ameri­ can origin are widely distributed in the Nearctic Region and just barely extend their ranges into the northernmost portion of the Neotropical Region in Mexico. The vespertilionid bats Pip- istrellus subflavus and Nycticeius humeralis fit this distribu­ tional pattern, as do Canis latrans and Lynx rufus. Other Flor­ ida mammals are widely distributed in both the Nearctic and Neotropical Regions. Some of these species range well into Middle America; among these are Lasiurus intermedius, Cryp- totis parva, Glaucomys volans, and Spilogale putorius, as well as the extinct species Mammut americanum, Mammuthus columbi, Bison latifrons, and B. antiquus. Still other extant spe­ cies, including Eptesicus fuscus, Lasiurus borealis, Urocyon ci- nereoargenteus, Procyon lotor, Mustela frenata, Puma con­ color, Sigmodon hispidus, Sylvilagus floridanus, and Odocoileus virginianus, are widely distributed in both North America and South America. Canis dirus, Panthera atrox, and Smilodon fatalis are extinct species that occurred on both conti­ nents. These three large carnivores are known from numerous late Pleistocene faunas in North America and have been identi­ fied from sites west of the Andes as far south as Peru (Kurten and Anderson, 1980; Kurten and Werdelin, 1990). A second, larger species of Smilodon, S. populator, occurs in late Pleis­ tocene faunas east of the Andes (Kurten and Werdelin, 1990) and also is known from older (early to middle Pleistocene) fau­ nas in South America (Berta, 1985). The gomphothere Cuvier- onius was widespread in Middle and South America during the late Pleistocene and thus is generally considered a Neotropical genus, although it also occurred in the southern portion of the Nearctic Region (Kurten and Anderson, 1980). The distribu­ tion of the various species of Cuvieronius is not well under­ stood because the taxonomy of the genus is in need of revision. Cuvieronius tropicus occurs in a limited number of Ranchola­ brean sites in the southeastern United States along the Atlantic and Gulf coasts, and it ranged southward into Middle America. Even though many of the species listed in this paragraph are/ were widely distributed in the Neotropical Region, with a few exceptions (e.g., Cuvieronius) they are rarely included in dis­ cussions of Florida mammals with Neotropical affinities. Pre­ sumably this is a result of their widespread occurrence in the Nearctic Region and their unambiguous North American ori­ gin. Another biogeographic pattern observed in Florida Rancho­ labrean mammals with Neotropical affinities is shown by ex­ tant species that have been extirpated from Florida and are now primarily restricted to the Neotropical Region. The best exam­ ples of this pattern are the jaguar (Panthera onca), ocelot (Leopardus pardalis), margay (Leopardus wiedii), and hog- nosed skunk (Conepatus leuconotus). Conepatus leuconotus (including C. mesomelas) is fairly widespread from the south­ western United States southward into Middle America. It is in­ cluded with this group because species of Conepatus are the typical Neotropical skunks. The jaguar and ocelot do occasion­ ally wander northward into the extreme southern portion of the Nearctic Region, but for the most part these two species and the margay have Neotropical distributions. The occurrence of L. wiedii in Florida requires additional comment. Fells amni- cola, an extinct species of small cat described from the Ran­ cholabrean Aucilla River fauna in northern Florida (Gillette, 1976), was later regarded as an extinct subspecies of Felis (= Leopardus) wiedii by Werdelin (1985). Regardless of which author's interpretation is correct, it seems clear that amnicola is an extinct species or subspecies of small cat that is closely re­ lated to, if not conspecific with, the tropical American margay. Two extinct species of large mammals of North American origin listed in Table 2, Tremarctos floridanus and Tapirus ve­ roensis, belong to genera that are now principally restricted to the Neotropical Region. For this reason, these two species are almost always mentioned in discussions of Florida mammals with Neotropical affinities, despite the fact that they are known from fossil sites only in the Nearctic Region. Tremarctos flori­ danus occurs in many Rancholabrean sites in Florida and in a few other late Pleistocene localities in the southern United States and northern Mexico (Kurten and Anderson, 1980). The only living member of the genus Tremarctos is the spectacled bear (T. ornatus), a species now restricted to the Andes Moun­ tains in South America. Tapirus veroensis also is common in Florida Rancholabrean sites and occurs in scattered localities elsewhere in the southeastern United States (Kurten and Anderson, 1980). The genus Tapirus has a long history in North America extending well back into the Miocene (Hulbert, 1995), but tapirs went extinct in the Nearctic Region at the end of the Pleistocene. Three species of Tapirus, T. bairdii, T.pin- chaque, and T. terrestris, survive today in the Neotropics. The genera Tremarctos and Tapirus were far more widespread in the New World as recently as the late Pleistocene, but they are now restricted to the Neotropical Region because the temperate species in each genus went extinct. Conepatus robustus, a large extinct species of hog-nosed skunk known only from the Ran­ cholabrean Haile 14A LF in northern Florida (Martin, 1978), may represent a third species with this same distributional pat­ tern. As noted above, living members of the genus Conepatus occur primarily in the Neotropics, with only a single species 32 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY making a small incursion into the southernmost portion of the Nearctic Region. Four artiodactyls from the Florida Rancholabrean, the pec­ caries Mylohyus nasutus and Platygonus compressus and the camels Hemiauchenia macrocephala and Palaeolama mirifica, are often included in discussions of Neotropical taxa. These four species are known only from Nearctic fossil sites and be­ long to extinct genera. They are included in Table 2 and this discussion because, with one exception, all living members of the Tayassuidae and the llama tribe (Lamini) of the Camelidae are restricted to the Neotropical Region. Species of Platygonus, Hemiauchenia, and Palaeolama also are known from the Pleis­ tocene of South America (Kurten and Anderson, 1980; Webb, 1974b). Like tapirs, tayassuids and camelids have long evolu­ tionary histories in North America and, except for one species, disappeared from the Nearctic Region in the late Pleistocene. One living species of tayassuid, the collared peccary or javelina (Dicotyles tajacu), barely enters the Nearctic Region in the southwestern United States and northern Mexico. This species is principally Neotropical in distribution, occurring throughout Middle America and much of South America. The absence of D. tajacu from North American late Pleistocene fossil sites (Kurten and Anderson, 1980) suggests that the northward spread of this species occurred during the Holocene, after the extinction of Mylohyus and Platygonus. It had been suggested that Palaeolama mirifica evolved in South America and then reinvaded North America (Webb, 1974b); however, recent dis­ coveries of early Irvingtonian Palaeolama in Florida (Webb and Stehli, 1995) indicate that this genus originated in North America, perhaps in tropical Middle America, and reached South America in the late Pliocene or early Pleistocene. One of the sigmodontine rodents found in Florida Ranchola­ brean faunas, the cotton rat (Sigmodon hispidus), now occurs from the southern United States southward through Middle America to northern South America. The genus Sigmodon orig­ inated in North America during the early Pliocene (early Blan­ can) and then dispersed into South America after the onset of the Great American Interchange (Martin, 1979; Jacobs and Lindsay, 1981, 1984). The two other sigmodontines listed in Table 2, the rice rat (Oryzomys palustris) and the Florida mouse (Podomys florida­ nus), do not fit comfortably into any of the biogeographic pat­ terns so far discussed. Although O. palustris is now found only in the southeastern United States, it is included in Table 2 be­ cause all other members of the genus Oryzomys are Neotropi­ cal in distribution. Neotropical sigmodontine rodents, includ­ ing Oryzomys, had their origin in North America during the late Miocene and early Pliocene (Baskin, 1978, 1986, 1989b; Jacobs and Lindsay, 1981, 1984). These authors have docu­ mented that sigmodontines underwent their initial diversifica­ tion in southern North America prior to the Great American Faunal Interchange. They further postulated that when these ro­ dents first reached South America in the Pliocene they had al­ ready begun the explosive radiation that characterizes the evo­ lutionary history of this group in the Neotropical Region. Oryzomys palustris probably evolved in tropical Middle Amer­ ica and then invaded temperate North America in the Rancho­ labrean. The first appearance of O. palustris in Florida is one of the most reliable biochronological indicators for the beginning of the Rancholabrean NALMA (Morgan and Hulbert, 1995). This species is known from Bradenton, Oldsmar, and several other Florida early Rancholabrean faunas (Morgan, 1991; Mor­ gan and White, 1995; Karrow et al., 1996), as well as many late Rancholabrean sites, but is unknown from Florida Irvingtonian faunas. The Florida mouse, Podomys floridanus, is endemic to the peninsula and eastern panhandle of Florida. This species is in­ cluded in Table 2 because the closest sister taxa of Podomys, the genera Habromys and Neotomodon (see Carleton, 1980), are Neotropical in distribution. The single species of Neotomo­ don and the four species of Habromys are all found at high ele­ vations in Middle America (Hall, 1981). The oldest record of the genus Podomys is from the early Irvingtonian Leisey Shell Pit and Haile 21A local faunas in Florida (Morgan and White, 1995). The living species, P. floridanus, first appeared in the late Irvingtonian Coleman 2A LF (Martin, 1974). MARINE MAMMALS Table 2 includes two marine mammals with Neotropical af­ finities recorded from Florida Rancholabrean faunas, the West Indian monk seal (Monachus tropicalis) and the West Indian manatee (Trichechus manatus). Ray (1958) identified M. tropi­ calis from the late Rancholabrean Melbourne LF, located along Florida's central Atlantic coast. Other fossils tentatively re­ ferred to M. tropicalis are known from the early Irvingtonian Leisey Shell Pit and Rigby Shell Pit local faunas along the Gulf coast in southwestern Florida (Morgan, 1994b; Morgan and Hulbert, 1995). Prior to its extinction within the last 40 years (Kenyon, 1977), the historic range of M. tropicalis included the southernmost portion of the Florida peninsula and the Florida Keys, as well as the Caribbean Sea and the southern Gulf of Mexico (Wing, 1992). Trichechus manatus is known from nu­ merous Rancholabrean river and spring deposits in northern Florida (Domning, 1982). The oldest verified fossil record of T. manatus is from the early Pleistocene Leisey Shell Pit (Mor­ gan, 1994b; Morgan and Hulbert, 1995). The current range of the West Indian manatee includes coastal marine habitats and freshwater rivers throughout the Florida peninsula, as well as the Greater Antilles, the Gulf and Caribbean coasts of Middle America, and northern and eastern South America. NEOTROPICAL BIRDS IN FLORIDA PLIO-PLEISTOCENE AND RECENT FAUNAS A strong Neotropical influence also is apparent in both the Plio-Pleistocene and modern avifaunas of Florida. The most spectacular Florida fossil bird of Neotropical origin is Titanis walleri, the only North American member of the otherwise strictly South American family Phorusrhacidae (Brodkorb, NUMBER 93 33 1963). Titanis first appears in Florida in the late Blancan along with a large suite of South American Interchange mammals (Webb, 1976; Chandler, 1994; Morgan and Hulbert, 1995). The rich Pleistocene record of fossil birds in Florida adds several more species with Neotropical affinities. The jacana (Jacana spinosa), a Neotropical waterbird found from Mexico south throughout much of Middle America, as well as the West In­ dies, is known from two Florida Rancholabrean faunas (Emslie, 1995, 1998). Two species of raptors found in Cutler Hammock and several other Florida Rancholabrean faunas, the extinct species Spizaetus grinnelli and the extant species Mil- vago chimachima, are members of genera that are now re­ stricted to Middle and South America (Emslie, 1995, 1998). Spizaetus grinnelli also occurs in the West Palm Beach LF (Becker, 1985). Three species of jays described from Florida Rancholabrean sites belong to extinct genera that have their closest living relatives in Middle America, according to Brod­ korb (1957) and Holman (1959). Protocitta, including the spe­ cies P. ajax and P. dixi, is closely related to Middle American jays of the genus Calocitta, whereas Henocitta brodkorbi is re­ lated to Middle American jays of the genus Psilorhinos (Brod­ korb, 1957; Holman, 1959). Likewise, two extinct genera of ic- terids known from Rancholabrean faunas in northern Florida, Cremaster and Pandanaris, have their closest living relatives in Middle America. Cremaster tytthus is an arboreal icterid most closely related to the living Neotropical oropendolas and caciques (Brodkorb, 1959). Pandanaris floridana has close af­ finities with Central American blackbirds (Hamon, 1964). At least 12 species of extant land birds found in Florida have affinities with the Neotropical Region (Robertson and Kushlan, 1974). These birds exhibit the same general biogeographic pat­ terns as Florida's Neotropical mammals discussed above, but birds have the added complication of migration. Many land birds that breed in Florida and elsewhere in temperate North America migrate to the Neotropics in the winter months. Only those land birds found in Florida that breed primarily in the Neotropical Region are discussed herein, including both migra­ tory and nonmigratory species. Seven of these species are West Indian in origin, two species are found primarily in Middle America, and three species could have been derived from either the West Indies or Middle America. The seven land birds with Antillean affinities include white-crowned pigeon (Columba leucocephala), mangrove cuckoo (Coccyzus minor), smooth- billed ani (Crotophaga ani), Antillean nighthawk (Cordeiles gundlachii), gray kingbird (Tyrannus dominicensis), black- whiskered vireo (Vireo altiloquus), and yellow warbler (Dend- roica petechia). None of these species occur in Florida Ran­ cholabrean fossil sites, suggesting that they may have dispersed northward from the West Indies fairly recently, perhaps during the Holocene. Two Florida birds probably were derived from Middle America, the short-tailed hawk (Buteo brachyurus) and the black-shouldered kite (Elanus caeruleus), although the lat­ ter species occurs in the southwestern United States as well. Three birds found in Florida, the snail kite (Rostrhamus socia- bilis), caracara (Polyborus plancus), and burrowing owl (Athene cunicularia), axe widespread Neotropical species that occur throughout the West Indies, Central America, and South America. The caracara and burrowing owl also are known from several Florida Rancholabrean faunas. These two species also are found in the arid regions of the western United States but do not occur between peninsular Florida and Texas. Southwestern Vertebrates in Florida Plio-Pleistocene and Recent Faunas Coinciding with the appearance of Neotropical species in the Florida peninsula was an influx of species from the arid south­ west. The dispersal of certain species of western vertebrates into the Florida peninsula probably was facilitated by the drier conditions, accompanied by more widespread grassland and scrub vegetation, that characterized the climate of glacial inter­ vals in Florida (Watts and Hansen, 1988). Texas is the closest state where many of the western species found in Florida Pleis­ tocene faunas still occur. Among mammals, these western forms include jackrabbits (Lepus), smooth-toothed pocket go­ phers (Thomomys), ground squirrels (Spermophilus), pallid bats (Antrozous), and possibly hog-nosed skunks (Conepatus). Western birds found in Florida fossil sites include two species of small owls of the genus Glaucidium (see Emslie, 1998), Cal­ ifornia condor (Gymnogyps californianus), prairie chicken (Tympanuchus cupido), and scrub jay (Aphelocoma coerule- scens). All of these western genera and/or species of mammals and birds are known from Florida Irvingtonian and/or Rancho­ labrean faunas, but only the scrub jay survives there today (Emslie, 1996). Several of these taxa are known only from ear­ lier glacials prior to the latest Pleistocene Wisconsinan glacia­ tion. The best-known pre-Wisconsinan fauna from Florida that clearly represents a glacial interval is the latest Blancan (latest Pliocene) Inglis IA LF, located just inland from the Gulf of Mexico in the north-central portion of the peninsula. Western species identified from Inglis (Meylan, 1982; Morgan, 1991; Emslie, 1996, 1998) include Lepus sp., Antrozous sp., two spe­ cies of Glaucidium owls, scrub jay (Aphelocoma coerule- scens), alligator lizard (Gerrhonotus sp.), and hog-nosed snake (Heterodon nasicus). None of these taxa are known from Flor­ ida Rancholabrean sites. Certain other species of mammals and birds found in Florida Pleistocene faunas occur in both the arid southwest and the Neotropics, including the bats Mormoops megalophylla, Eumops underwoodi, and Tadarida brasiliensis, the burrowing owl (Athene cunicularia), and the caracara (Po­ lyborus plancus). For these species it is difficult to determine the geographic origin of the Florida fossil population. The two birds and T brasiliensis still survive in Florida. The two other bats have been extirpated from the state. The burrowing owl and caracara are clearly relicts in Florida, as no other popula­ tions of these two species exist in eastern North America, whereas T. brasiliensis is more widespread in the southeastern United States. 34 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY Conclusions The most plausible explanation for the diversity of Neotropi­ cal vertebrates in Florida late Cenozoic faunas involves the combination of a subtropical climate and the intermittent pres­ ence of a broad savanna corridor connecting Florida with tropi­ cal Middle America around the northern and western margins of the Gulf of Mexico. This Gulf coastal corridor was a particu­ larly important avenue of dispersal during glacial intervals in the late Pliocene and Pleistocene (Auffenberg and Milstead, 1965; Webb, 1976; Webb and Wilkins, 1984; Morgan, 1985, 1991; Emslie, 1998). The breadth of the Gulf coastal corridor increased during glacials as lower sea levels exposed the shal­ low continental shelf along the western and northern Gulf coast in Florida and westward around the Gulf of Mexico to southern Mexico. The dispersal of Neotropical vertebrates into Florida probably was facilitated by the more equable climate that char­ acterized much of the unglaciated region of North America during glacial intervals, especially during the Wisconsinan gla­ ciation (Graham and Lundelius, 1984). Perhaps the most im­ portant climatic factor influencing the occurrence of tropical vertebrates in Florida is the presence or absence of prolonged freezing temperatures in winter. The region south of the 10°C winter isotherm, which currently passes through the central portion of the Florida peninsula at about the latitude of Tampa and Orlando (between 27°N and 28°N), only rarely experi­ ences freezing temperatures in winter, and the freezes that do occur are of short duration. Prolonged freezes of several days to a week in northern Florida and along the northern Gulf coast may limit the dispersal of mainland Neotropical vertebrates into southern peninsular Florida, where many species of tropi­ cal vertebrates could survive. Introduced tropical vertebrates frequently occur in the wild in southern Florida; these include two species of monkeys, the rhesus macaque (Macaca mulata) and the squirrel monkey (Saimiri sciureus), and the red-bellied squirrel (Sciurus aureogaster), as well as numerous species of tropical birds and reptiles. During glacial intervals, when win­ ters supposedly were somewhat milder, much of the Gulf Coastal Plain was probably frost-free, permitting tropical spe­ cies to reach the Florida peninsula from Middle America. Land tortoises of the genus Geochelone (now placed in the endemic New World genus Hesperotestudo; see Meylan, 1995) have long been used as indicators of Pleistocene climate (e.g., Hibbard, 1960). Living species of giant tortoises are restricted to tropical regions. They are intolerant of freezing temperatures and are too large to burrow and hibernate to escape the cold (Auffenberg and Milstead, 1965). The presence of giant tor­ toises (Hesperotestudo crassiscutata and related species) in North American late Cenozoic sites is therefore generally thought to indicate a frost-free climate (Hibbard, 1960). Two species of Hesperotestudo, H. crassiscutata and the extinct dwarf tortoise, H. incisa, were common throughout Florida in the Rancholabrean. The southernmost occurrence of H. incisa is at Cutler Hammock, whereas the southernmost Ranchola­ brean record of H. crassiscutata in Florida is from the West Palm Beach site. The occurrence of both species of Hesper­ otestudo throughout the northern half of the Florida peninsula during the late Rancholabrean indicates a frost-free climate for this region. The principal dispersal route for Neotropical species that reached Florida was from tropical Middle America northward around the Gulf of Mexico. With several exceptions, most no­ tably the armadillo (Dasypus novemcinctus), very few species of tropical vertebrates have reached Florida using the Gulf coastal corridor since the end of the Pleistocene. A second bio­ geographic pathway involved species that reached southern Florida by overwater dispersal from the West Indies, including several bats and birds discussed above, the green anole (Anolis carolinensis), and numerous plants (Long, 1974). Most Florida mammals with Neotropical affinities disap­ peared at the end of the Pleistocene. Whatever the reason for these extinctions, the rich diversity of Neotropical mammals in Florida Rancholabrean faunas has been reduced to just a few species today. Southernmost peninsular Florida, one of the few subtropical regions in the continental United States, now sup­ ports only five Neotropical mammals, the large molossid bat Eumops glaucinus, the small molossid Molossus molossus, and the Jamaican fruit bat, Artibeus jamaicensis, as well as the more widely distributed Didelphis virginiana and Dasypus novemcinctus. Eumops glaucinus reached the Florida peninsula during the Rancholabrean, and it is known there from three sites, two of which are north of the species' current range (Morgan, 1985, 1991). Artibeus jamaicensis and Molossus mo­ lossus were almost certainly Holocene arrivals. The closest geographic region to Florida that supports a Neotropical fauna is the West Indies, the northernmost extensions of which, Cuba and the Bahamas, are separated from southernmost peninsular Florida by narrow (about 100 km wide), deep-water gaps. Not surprisingly, most Antillean species that apparently reached southern Florida in the Holocene are volant forms, including as many as seven species of birds (Robertson and Kushlan, 1974) and two bats. Most Neotropical vertebrates that arrived in Flor­ ida in the late Pliocene and Pleistocene traveled the much longer mainland route from Middle America, suggesting that distance is less of a barrier to dispersal than are oceanic water gaps. With the exception of the very recent arrival of Dasypus novemcinctus, post-Pleistocene dispersal of Neotropical verte­ brates into Florida from tropical Middle America has almost entirely ceased, probably owing to inhospitable climatic condi­ tions along the Gulf coastal corridor between southern Florida and northern Mexico. NUMBER 93 35 Literature Cited Ahearn, M.E. 1981. A Revision of the North American Hydrochoeridae. 99 pages. Mas­ ter's thesis, University of Florida, Gainesville. Ahearn M.E., and J.F. Lance 1980. A New Species of Neochoerus (Rodentia: Hydrochoeridae) from the Blancan (Late Pliocene) of North America. Proceedings of the Bio­ logical Society of Washington, 93:435-442. Allen, G.M. 1932. A Pleistocene Bat from Florida. Journal of Mammalogy, 13:256- 259. Auffenberg, W., and W.W. Milstead 1965. Reptiles in the Quaternary of North America. In H.E. Wright, Jr., and D.G. Frey, editors, The Quaternary of the United States, pages 557-568. Princeton; Princeton University Press. Baker, R.J., and H.H. Genoways 1978. Zoogeography of Antillean Bats. In FB. Gill, editor, Zoogeography in the Caribbean. Academy of Natural Sciences of Philadelphia, Special Publication, 13:53-97. Baskin, J.A. 1978. Bensonomys, Calomys, and the Origin of the Phyllotine Group of Neotropical Cricetines (Rodentia: Cricetidae). Journal of Mammal­ ogy, 59:125-135. 1986. The Late Miocene Radiation of Neotropical Sigmodontine Rodents in North America. Contributions in Geology, University of Wyo­ ming, Special Paper, 3:287-303. 1989a. Comments on New World Tertiary Procyonidae (Mammalia: Car- nivora). Journal of Vertebrate Paleontology. 9:110-117. 1989b. The Initial Origin and Diversification of the Neotropical Sigmodon- tinae (Rodentia: Muridae): A Perspective from the North American Fossil Record. In Fifth International Theriological Congress, Rome, 22-29 August 1989, Abstracts of Papers and Posters, 1:263-264. 1995. The Giant Flightless Bird Titanis Walleri (Aves: Phorusrhacidae) from the Pleistocene Coastal Plain of South Texas. Journal of Verte­ brate Paleontology, 15:842-844. Becker, J.J. 1985. A Late Pleistocene (Wisconsinan) Avifauna from West Palm Beach, Florida. Bulletin of the British Ornithologists' Club. 105(1 ):37-40. Belwood, J.J. 1981. Wagner's Mastiff Bat, Eumops Glaucinus Floridanus (Molossidae), in Southwest Florida. Journal of Mammalogy, 62:411-413. Berta, A. 1985. The Status of Smilodon in North and South America. Contributions in Science, Natural History Museum of Los Angeles County, 370:1-15. Bloom, A.L. 1983. Sea Level and Coastal Morphology of the United States through the Late Wisconsinan Glacial Maximum. In S.C. Porter, editor, Quater­ nary Environments of the United States, I: The Late Pleistocene, pages 215-229. Minneapolis: University of Minnesota Press. Brodkorb, P. 1957. New Passerine Birds from the Pleistocene of Reddick Florida. Jour­ nal of Paleontology, 31:129-138. 1959. The Pleistocene Avifauna of Arredondo, Florida. Bulletin of the Florida State Museum, Biological Sciences, 4(9):269-291. 1963. A Giant Flightless Bird from the Pleistocene of Florida. Auk, 80:111-115. Buckley, J., and E.H. Willis 1972. Isotopes' Radiocarbon Measurements, IX. Radiocarbon, 14:114— 139. Carleton, M.C. 1980. Phylogenetic Relationships in Neotomine-Peromyscine Rodents (Muroidea) and a Reappraisal of the Dichotomy within New World Cricetinae. Museum of Zoology, University of Michigan, Miscella­ neous Publications, 157:1-146. Carr, R.S. 1987. Early Man in South Florida. Archaeology, November/December: 62-63. Cartelle, C , and G. De Iuliis 1995. Eremotherium laurillardi: The Panamerican Late Pleistocene Mega- theriid Sloth. Journal of Vertebrate Paleontology, 15:830-841. Chandler, R.M. 1994. The Wing of Titanis walleri (Aves: Phorusrhacidae) from the Late Blancan of Florida. Bulletin of the Florida Museum of Natural His­ tory, Biological Sciences, 36(6): 175-180. Clausen, C.J., A.D. Cohen, C. Emiliani, J.A. Holman, and J.J. Stipp 1979. Little Salt Spring, Florida: A Unique Underwater Site. Science, 203: 609-614. Converse, H.H., Jr. 1973. A Pleistocene Vertebrate Fauna from Palm Beach County, Florida. Plaster Jacket, 21:1-2. Czaplewski, N.J. 1997. Chiroptera. In R.F. Kay, R.H. Madden, R.L. Cifelli, and J.J. Flynn, editors, Vertebrate Paleobiology in the Neotropics: The Miocene Fauna of La Venta, Colombia, pages 410-431. Washington, D .C: Smithsonian Institution Press. Domning, D.P. 1982. Evolution of Manatees: A Speculative History. Journal of Paleon­ tology, 56:599-619. Downing, K.F., and R. White 1995. The Cingulates (Xenarthra) of Leisey Shell Pit IA (Irvingtonian), Hillsborough County, Florida. In R.C. Hulbert, Jr., G.S. Morgan, and S.D. Webb, editors, Paleontology and Geology of the Leisey Shell Pits, Early Pleistocene of Florida. Bulletin of the Florida Mu­ seum of Natural History. 37(2)12:375-396. Dunbar, J.S., S.D. Webb, and D. Cring 1989. Culturally and Naturally Modified Bones from a Paleoindian Site in the Aucilla River, North Florida. In R. Bonnichsen and M.H. Sorg, editors, Bone Modification, pages 473^497. Orono, Maine: Univer­ sity of Maine, Center for the Study of the First Americans, Institute for Quaternary Studies. Eger, J.L. 1977. Systematics of the Genus Eumops (Chiroptera: Molossidae). Life Sciences Contributions, Royal Ontario Museum, 110:1-69. Emslie, S.D. 1995. An Early Irvingtonian Avifauna from Leisey Shell Pit, Florida. In R.C. Hulbert, Jr., G.S. Morgan, and S.D. Webb, editors, Paleontol­ ogy and Geology of the Leisey Shell Pits, Early Pleistocene of Flor­ ida. Bulletin of the Florida Museum of Natural History, 37(1)10:299-344. 1996. A Fossil Scrub-Jay Supports a Recent Systematic Decision. Condor, 98:675-680. 1998. Avian Community, Climate, and Sea-Level Changes in the Plio- Pleistocene of the Florida Peninsula. Ornithological Monographs, 50: 113 pages. Emslie, S.D., and G.S. Morgan 1995. Taphonomy of a Late Pleistocene Carnivore Den, Dade County Florida. In D.W. Steadman and J.I. Mead, editors, Late Quaternary Environments and Deep History: A Tribute to Paul S. Martin. The Mammoth Site of Hot Springs, South Dakota, Scientific Papers, 3:65-83. Eshelman, R.E., and G.S. Morgan 1985. Tobagan Recent Mammals, Fossil Vertebrates, and Their Zoogeo- graphical Implications. Research Reports, National Geographic So­ ciety, 21:137-143. 36 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY Frank, P.A. 1997a. First Record of Artibeus jamaicensis Leach (1821) from the United States. Florida Scientist, 60:37-39. 1997b. First Record of Molossus molossus tropidorhynchus Gray (1839) from the United States. Journal of Mammalogy, 78:103-105. Frazier, M.K. 1981. A Revision of the Fossil Erethizontidae of North America. Bulletin of the Florida State Museum, Biological Sciences, 27(1): 1-76. Gazin, CL . 1956. Exploration for the Remains of Giant Ground Sloths in Panama. An­ nual Report of the Smithsonian Institution, pages 341-354. Gillette, D.D. 1976. A New Species of Small Cat from the Late Quaternary of Southeast- em United States. Journal of Mammalogy'. 57:664-676. Graham, R.W., and E.L. Lundelius, Jr. 1984. Coevolutionary Disequilibrium and Pleistocene Extinctions. In P.S. Martin and R.G. Klein, editors, Quaternary Extinctions: A Prehis­ toric Revolution, pages 223-249. Tucson: University of Arizona Press. Guilday, J.E., H.W. Hamilton, E. Anderson, and P.W. Parmalee 1978. The Baker's Bluff Cave Deposit, Tennessee, and the Late Pleis­ tocene Faunal Gradient. Bulletin of the Carnegie Museum of Natu­ ral History, 11:1-67. Guilday, J.E., P.W. Parmalee, and H.W. Hamilton 1977. The Clark's Cave Bone Deposit and the Late Pleistocene Paleoecol­ ogy of the Central Appalachian Mountains. Bulletin of the Carnegie Museum of Natural History, 2:1-87. Gut, H.J. 1959. A Pleistocene Vampire Bat from Florida. Journal of Mammalogy, 40:534-538. Hall, E.R. 1981. The Mammals of North America. Second edition, 2 volumes, xv + 1181 pages. New York: John Wiley and Sons. Hamon, J.H. 1964. Osteology and Paleontology of the Passerine Birds of the Reddick, Florida, Pleistocene. Geological Bulletin, Florida Geological Sur­ vey, 44:1-210. Harmon, R.S., R.M. Mitterer, N. Kriausakul, L.S. Land, H.P. Schwarcz, P. Gar­ rett, G.J. Larson, H.L. Vacher, and M. Rowe 1983. U-Series and Amino-Acid Racemization Geochronology of Ber­ muda: Implications for Eustatic Sea-Level Fluctuation over the Past 250,000 Years. Palaeogeography, Palaeoclimalology, Palaeoecol- ogy, 44:41-70. Hershkovitz, P. 1958. A Geographic Classification of Neotropical Mammals. Fieldiana, Zoology, 36(6):581-620. Hibbard, C.W. 1960. An Interpretation of Pliocene and Pleistocene Climates in North America. 62nd Annual Report of the Michigan Academy of Science. Arts and Letters, pages 5-30. Hirschfeld, S.E. 1968. Vertebrate Fauna of Nichol's Hammock, a Natural Trap. Quarterly Journal of the Florida Academy of Sciences, 31:177-189. Hirschfeld, S.E., and S.D. Webb 1968. Plio-Pleistocene Megalonychid Sloths of North America. Bulletin of the Florida State Museum, Biological Sciences, 12(5):213-296. Holman, J.A. 1959. Birds and Mammals from the Pleistocene of Williston, Florida. Bul­ letin of the Florida Stale Museum, Biological Sciences, 5( 1 ):24. Holman, J.A., and C.J. Clausen 1984. Fossil Vertebrates Associated with Paleo-Indian Artifact at Little Salt Spring, Florida. Journal of Vertebrate Paleontology, 4: 146-154. Hulbert, R.C, Jr. 1995. The Giant Tapir, Tapirus haysii, from Leisey Shell Pit 1A and Other Florida Irvingtonian Localities. In R.C. Hulbert, Jr., G.S. Morgan, and S.D. Webb, editors, Paleontology and Geology of the Leisey Shell Pits, Early Pleistocene of Florida. Bulletin of the Florida Mu­ seum of Natural History, 37(2)16:515-551. 1997. A New Late Pliocene Porcupine (Rodentia: Erethizontidae) from Florida. Journal of Vertebrate Paleontology, 17:623-626. 2001. The Fossil Vertebrates of Florida. 350 pages. Gainesville, Florida: University of Florida Press. Humphrey, S.R. 1974. Zoogeography of the Nine-Banded Armadillo (Dasypus novemcinc­ tus) in the United States. Bioscience, 24(8):457-462. Jacobs, L.L., and E.H. Lindsay 1981. Prosigmodon oroscoi, a New Sigmodont Rodent from the Late Ter­ tiary of Mexico. Journal cf Paleontology, 55:425-430. 1984. Holarctic Radiation of Neogene Muroid Rodents and the Origin of South American Cricetids. Journal of Vertebrate Paleontology, 4: 265-272. Jennings, W.L. 1958. The Ecological Distribution of Bats in Florida. 126 pages, doctoral dissertation. University of Florida, Gainesville. Karrow, P.F., G.S. Morgan, R.W. Portell, E.H. Simons, and K. Auffenberg 1996. Middle Pleistocene (Early Rancholabrean) Vertebrates and Associ­ ated Marine and Non-Marine Invertebrates from Oldsmar, Pinellas County, Florida. In K.M. Stewart and K.L. Seymour, editors, Palae- oecology and Palaeoenvironmenls of Late Cenozoic Mammals: Tributes to the Career ofC.S. (Rufus) Churcher, pages 97-133. Tor­ onto: University of Toronto Press. Kenyon, K.W. 1977. Caribbean Monk Seal Extinct. Journal of Mammalogy, 58:97-98. Koopman, K.F. 1958. A Fossil Vampire Bat from Cuba. Breviora. 90:1—4. 1971. The Systematic and Historical Status of the Florida Eumops (Chi­ roptera, Molossidae). American Museum Novitates, 2478: 6 pages. Koopman, K.F., and P.S. Martin 1959. Subfossil Mammals from the Gomez Farias Region and the Tropical Gradient of Eastern Mexico. Journal of Mammalogy, 40:1-12. Kurten, B., and E. Anderson 1980. Pleistocene Mammals of North America. 443 pages. New York: Co­ lumbia University Press. Kurten, B., and L. Werdelin 1990. Relationships between North and South American Smilodon. Jour­ nal of Vertebrate Paleontology. 10:158-169. Layne, J.N. 1974. The Land Mammals of South Florida. In P.J. Gleason, editor, Envi­ ronments of South Florida: Present and Past. Miami Geological So­ ciety Memoir, 2:386^413. Lazell, J.D. 1989. Wildlife of the Florida Keys: A Natural History. 253 pages. Wash­ ington, D.C: Island Press. Lazell, J.D., Jr., and K.F. Koopman 1985. Notes on Bats of Florida's Lower Keys. Florida Scientist, 48:37-41. Long, R.W. 1974. Origin of the Vascular Flora of South Florida. In P.J. Gleason, editor, Environments of South Florida: Present and Past. Miami Geological Society Memoir, 2:28-36. Lundelius, E.L., Jr., C S . Churcher, T. Downs, CR. Harington, E.H. Lindsay, G.E. Schultz, H.A. Semken, S.D. Webb, and R.J. Zakrzewski 1987. The North American Quaternary Sequence. In M.O. Woodburne, editor, Cenozoic Mammals of North America: Geochronology and Biostratigraphy, pages 211-235. Berkeley: University of California Press. MacPhee, R.D.E., and M.A. Iturralde-Vinent 1994. First Tertiary Land Mammal from Greater Antilles: An Early Mi­ ocene Sloth (Xenarthra, Megalonychidae) from Cuba. American Museum Novitates, 3094: 13 pages. NUMBER 93 37 Martin, R.A. 1972. Synopsis of Late Pliocene and Pleistocene Bats of North America. American Midland Naturalist, 87:326-335. 1974. Fossil Mammals from the Coleman IIA Fauna, Sumter County. In S.D. Webb, editor, Pleistocene Mammals of Florida, pages 35-99. Gainesville, Florida: University Presses of Florida. 1977. Late Pleistocene Eumops from Florida. Bulletin of the New Jersey Academy of Science, 22:18-19. 1978. A New Late Pleistocene Conepatus and Associated Vertebrate Fauna from Florida. Journal of Paleontology. 52:1079-1085. 1979. Fossil History of the Rodent Genus Sigmodon. Evolutionary Mono­ graphs, 2:1-36. McDonald, H.G. 1990. Understanding the Paleoecology of Fossil Vertebrates: Contribu­ tions of Submerged Sites. In W.C. Jaap, editor, Diving for sci­ ence...1990. Proceedings of the American Academy of Underwater Sciences Tenth Annual Scientific Diving Symposium, October 4-7, 1990, St. Petersburg. Florida, pages 273-292. 1995. Gravigrade Xenarthrans from the Early Pleistocene Leisey Shell Pit IA, Hillsborough County, Florida. In R.C. Hulbert, Jr., G.S. Mor­ gan, and S.D. Webb, editors, Paleontology and Geology of the Lei­ sey Shell Pits, Early Pleistocene of Florida. Bulletin of the Florida Museum of Natural History, 37(2)11:345-373. McNab, B.K. 1973. Energetics and the Distribution of Vampires. Journal of Mammal­ ogy, 54:131-144. Meylan, P.A. 1982. The Squamate Reptiles of the Inglis IA Fauna (Irvingtonian: Citrus County, Florida). Bulletin of the Florida State Museum, Biological Sciences, 27(3): 1-85. 1995. Pleistocene Amphibians and Reptiles from the Leisey Shell Pit, Hillsborough County, Florida. In R.C. Hulbert, Jr., G.S. Morgan, and S.D. Webb, editors, Paleontology and Geology of the Leisey Shell Pits, Early Pleistocene of Florida. Bulletin of the Florida Mu­ seum of Natural History. 37(1)9:273-297. Montellano-Ballesteras, M., and O. Carranza-Castaneda 1986. Descripcion de un milodontido del Blancano temprano de la Mesa Central de Mexico. Universidad Nacional Autonoma de Mexico. In­ stitute de Geologia. Revista. 6:193-203. Morgan, G.S. 1985. Fossil Bats (Mammalia: Chiroptera) from the Late Pleistocene and Holocene Vero Fauna, Indian River County, Florida. Brimleyana, 11:97-117. 1989a. Fossil Chiroptera and Rodentia from the Bahamas, and the Histori­ cal Biogeography of the Bahamian Mammal Fauna. In CA. Woods, editor, Biogeography of the West Indies: Past, Present, and Future, pages 685-740. Gainesville, Florida: Sandhill Crane Press. 1989b. New Bats from the Oligocene and Miocene of Florida and the Ori­ gins of the Neotropical Chiropteran Fauna. Journal of Vertebrate Paleontology, supplement, 9(3):33A. 1991. Neotropical Chiroptera from the Pliocene and Pleistocene of Flor­ ida. Bulletin of the American Museum of Natural History, 206: 176-213. 1994a. Late Quaternary Fossil Vertebrates from the Cayman Islands. In M.A. Brunt and J.E. Davies, editors, The Cayman Islands: Natural History and Biogeography, pages 465-508. Dordrecht, Netherlands: Kluwer Academic Publishers. 1994b. Miocene and Pliocene Marine Mammal Faunas from the Bone Val­ ley Formation of Central Florida. In A. Berta and T.A. Demere, edi­ tors, Contributions in Marine Mammal Paleontology Honoring Frank C. Whitmore, Jr. Proceedings of the San Diego Society of Natural History, 29:239-268. 2001. Patterns of Extinction in West Indian Bats. In C A . Woods, editor, Biogeography of the West Indies: New Patterns and Perspectives, pages 367-405. Boca Raton, Florida: CRC Press. Morgan, G.S., and R.C. Hulbert, Jr. 1995. Overview of the Geology and Vertebrate Biochronology of the Lei­ sey Shell Pit Local Fauna, Hillsborough County, Florida. In R.C. Hulbert, Jr., G.S. Morgan, and S.D. Webb, editors, Paleontology and Geology of the Leisey Shell Pits, Early Pleistocene of Florida. Bul­ letin of the Florida Museum of Natural History, 37(1)1:1-92. Morgan, G.S., O.J. Linares, and C E . Ray 1988. New Species of Fossil Vampire Bats (Mammalia: Chiroptera: Des- modontidae) from Florida and Venezuela. Proceedings of the Bio­ logical Society of Washington, 101:912-928. Morgan, G.S., and R.W. Portell 1996. The Tucker Borrow Pit: Paleontology and Stratigraphy of a Plio- Pleistocene Fossil Site in Brevard County, Florida. Papers in Flor­ ida Paleontology, 7:1-25. Morgan, G.S., and R.B. Ridgway 1987. Late Pliocene (Late Blancan) Vertebrates from the St. Petersburg Times Site, Pinellas County, Florida, with a Brief Review of Florida Blancan Faunas. Papers in Florida Paleontology, 1:1-22. Morgan, G.S., and K.L. Seymour 1997. Fossil History of the Panther (Puma concolor) and the Cheetah-Like Cat (Miracinonyx inexpectatus) in Florida. Bulletin of the Florida Museum of Natural History, 40(2): 177-219. Morgan, G.S., and J.A. White 1995. Small Mammals (Insectivora, Lagomorpha, and Rodentia) from the Early Pleistocene (Early Irvingtonian) Leisey Shell Pit Local Fauna, Hillsborough County, Florida. In R.C. Hulbert, Jr., G.S. Morgan, and S.D. Webb, editors, Paleontology and Geology of the Leisey Shell Pits, Early Pleistocene of Florida. Bulletin of the Florida Mu­ seum of Natural History. 37(2)13:397-461. Morgan, G.S., and CA. Woods 1986. Extinction and the Zoogeography of West Indian Land Mammals. Biological Journal of the Linnean Society, 28:167-203. Ober, L.D. 1978. The Monkey Jungle, a Late Pleistocene Fossil Site in Southern Flor­ ida. Plaster Jacket, 28:1-13. Olsen, S.J. 1960. Additional Remains of Florida's Pleistocene Vampire. Journal of Mammalogy. 41:458^462. Osmond, J.K., J.R. Carpenter, and H.L. Windom 1965. Th230/U234 Age of Pleistocene Corals and Oolites of Florida. Jour­ nal of Geophysical Research, 70:1843-1847. Paula Couto, C de 1967. Pleistocene Edentates of the West Indies. American Museum Novi­ tates, 2304: 55 pages. Ray, C E . 1958. Additions to the Pleistocene Mammalian Fauna from Melbourne, Florida. Bulletin of the Museum of Comparative Zoology, 119(7): 421-449. Ray, C.E., O.J. Linares, and G.S. Morgan 1988. Paleontology. In A.M. Greenhall and U. Schmidt, editors, Natural History of Vampire Bats, pages 19-30. Boca Raton, Florida: CRC Press. Ray, C E , S.J. Olsen, and H.J. Gut 1963. Three Mammals New to the Pleistocene Fauna of Florida, and a Re­ consideration of Five Earlier Records. Journal of Mammalogy, 44: 373-395. Rice, D.W. 1957. Life History and Ecology of Myotis austroriparius in Florida. Jour­ nal of Mammalogy. 38:15-32. Robertson, J.S., Jr. 1974. Fossil Bison of Florida. In S.D. Webb, editor, Pleistocene Mammals of Florida, pages 214-246. Gainesville, Florida: University Presses of Florida. Robertson, W.B., Jr., and J.A. Kushlan 1974. The Southern Florida Avifauna. In P.J. Gleason, editor, Environ- 38 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY ments of South Florida: Present and Past. Miami Geological Society Memoir, 2:414-452. Savage, D.E. 1951. Late Cenozoic Vertebrates of the San Francisco Bay Region. Uni­ versity of California Publications, Bulletin of the Department of Geological Sciences, 28:215-314. Silva Taboada, G. 1974. Fossil Chiroptera from Cave Deposits in Central Cuba, with De­ scription of Two New Species (Genera Pteronotus and Mormoops), and the First West Indian Record of Mormoops megalophylla. Acta Zoologica Cracoviensia, 19(3):33—73. Smith, J.D. 1972. Systematics of the Chiropteran Family Mormoopidae. University of Kansas Museum of Natural History, Miscellaneous Publication, 56: 132 pages. Waldrop, J.S. 1974. The Scimitar Cat, Homotherium serum, from the Florida Late Pleis­ tocene. In S.D. Webb, editor, Pleistocene Mammals of Florida, pages 154-157. Gainesville, Florida: University Presses of Florida. Watts, W.A., and B.C.S. Hansen 1988. Environments of Florida in the Late Wisconsin and Holocene. In B.A. Purdy, editor, Wet Site Archaeology, pages 307-323. Caldwell, New Jersey: Telford Press. Webb, S.D. 1974a. Chronology of Florida Pleistocene Mammals. In S.D. Webb, editor, Pleistocene Mammals of Florida, pages 5-31. Gainesville, Florida: University Presses of Florida. 1974b. Pleistocene Llamas of Florida, with a Brief Review of the Lamini. In S.D. Webb, editor, Pleistocene Mammals of Florida, pages 170-213. Gainesville, Florida: University Presses of Florida. 1976. Mammalian Faunal Dynamics of the Great American Interchange. Paleobiology, 2:220-234. 1985. Late Cenozoic Mammal Dispersals between the Americas. In F.G. Stehli and S.D. Webb, editors, The Great American Biotic Inter­ change, pages 357-386. New York: Plenum Press. 1989. Osteology and Relationships of Thinobadistes segnis, the First Myl- odont Sloth in North America. In K. Redford and J.F. Eisenberg, ed­ itors, Advances in Neotropical Mammalogy, pages 469-532 . Gainesville, Florida: Sandhill Crane Press. Webb, S.D., and S.C Perrigo 1984. Late Cenozoic Vertebrates from Honduras and El Salvador. Journal of Vertebrate Paleontology, 4:237-254. 1985. New Megalonychid Sloths from El Salvador. In G.G. Montgomery, editor, The Evolution and Ecology of Armadillos, Sloths, and Vermi- linguas, pages 113-120. Washington, D.C: Smithsonian Institution Press. Webb, S.D., and F.G. Stehli 1995. Selenodont Artiodactyla (Camelidae and Cervidae) from the Leisey Shell Pits, Hillsborough County, Florida. In R.C. Hulbert, Jr., G.S. Morgan, and S.D. Webb, editors, Paleontology and Geology of the Leisey Shell Pits, Early Pleistocene of Florida. Bulletin of the Flor­ ida Museum of Natural History, 37(2)19:621-643. Webb, S.D., and K.T. Wilkins 1984. Historical Biogeography of Florida Pleistocene Mammals. In H.H. Genoways and M.R. Dawson, editors, Contributions in Quaternary Vertebrate Paleontology: A Volume in Memorial to John E. Guilday. Special Publication, Carnegie Museum of Natural History, 8:370- 383. Weigel, R.D. 1962. Fossil Vertebrates of Vero, Florida. Florida Geological Survey, Spe­ cial Publication, 10:1-59. Werdelin, L. 1985. Small Pleistocene Felines of North America. Journal of Vertebrate Paleontology, 5:194-210. Wilkins, K.T. 1983. Pleistocene Mammals from the Rock Springs Local Fauna, Central Florida. Brimleyana, 9:69-82. 1984. Evolutionary Trends in Florida Pleistocene Pocket Gophers (Genus Geomys), with Description of a New Species. Journal of Vertebrate Paleontology, 3:166-181. Wilson, D.W., and D.M. Reeder, editors 1992. Mammal Species of the World: A Taxonomic and Geographic Refer­ ence. 1206 pages. Washington, D.C: Smithsonian Institution Press. Winans, M.C. 1989. A Quantitative Study of North American Fossil Species of the Ge­ nus Equus. In D.R. Prothero and R.M. Schoch, editors, The Evolu­ tion of Perissodactyls, pages 262-297. New York: Oxford Uni­ versity Press. Wing, E.S. 1992. West Indian Monk Seal, Monachus tropicalis. In S.R. Humphrey, editor, Rare and Endangered Biota of Florida, 1: Mammals, pages 35-40. Gainesville, Florida: University Presses of Florida. The Basicranium of the Giant Wombat Phascolonus gigas Owen (Vombatidae: Marsupialia) and Its Significance in Phylogeny Richard H. Tedford ABSTRACT The availability of two skulls of the giant wombat Phascolonus gigas Owen, from Pleistocene deposits of Lake Callabonna, South Australia, allows characters from the basicranium to contribute to the phylogenetic resolution of the Vombatidae. In agreement with previous work, this study confirms that Phascolonus is a part of the clade containing the living hairy-nosed wombat, Lasiorhinus, and other large fossil wombats. Phascolonus retains some plesio- morphies seen only in the Miocene wombats (large, bilobed pre­ molars) but also exhibits a number of autapomorphies of the skull and dentition, the most striking of which are the great breadth of the upper incisors, the peculiar arcuate ectotympanic, and the enclosure of the facial canal within the mastoid process of the peri­ otic. Introduction In 1913, E.C. Stirling announced the discovery at Lake Cal­ labonna, South Australia, of the first associated upper and lower dentitions and most of the major skeletal elements of Phascolonus gigas (Owen, 1858). This association demon­ strated that the mandibular remains first described by Owen in 1858, as Phascolomys gigas (made the genotypic species of Phascolonus Owen in 1872), pertained to the same taxon as the peculiar, flattened, central upper incisors described as Sceparn- odon stephensii by Ramsay in 1880. In 1970, nearly sixty years after Stirling's announcement, a joint Smithsonian Institution- American Museum of Natural History-South Australian Mu­ seum expedition to Lake Callabonna discovered two partial skeletons of Phascolonus gigas at Site 4 in the northern part of the fossil field (Wells and Tedford, 1995:110, fig. 2). The skel­ etons included two crushed, but nearly complete, adult skulls Richard H. Tedford, Department of Vertebrate Paleontology1, Ameri­ can Museum of Natural History, Central Park West at 79th Street, New York, New York 10024. now housed at the National Museum of Natural History (the former United States National Museum), USNM 214696 and 215139, of which the former supplied most of the basicranial morphology described below. Shortly after the field shipment was received at the Smithso­ nian, these specimens were skillfully prepared by Frank Pearce, who had been a member of the field party. He used a wax-impregnation technique, which avoided problems of dry­ ing and fragmentation by salt crystal growth that so plagued the preparation of Callabonna fossils owing to their long burial within the saline water table. Study and photographic illustra­ tion of material prepared in this way are best effected under ul­ traviolet light, as the wax-impregnated bone fluoresces strongly in contrast to the matrix. This study of the basicranial morphology is intended to help clarify the phylogenetic relationship of Phascolonus. Compari­ sons are limited to living wombats because the only compara­ ble cranial material available from fossil species is from forms closely related to those living. In addition to the living wombats, Lasiorhinus and Vomba- tus, the following outgroup taxa were examined: Ilaria illu- midens Woodburne and Tedford (AMNH 102651), casts of the Alcoota diprotodontids and the diprotodontid Ngapakaldia ted- fordi Stirton, and a cranium of Thylacoleo carnifex Owen from Victoria Cave (SAM P16681). ACKNOWLEDGMENTS.—This paper is dedicated to Clayton Ray who not only lovingly collected the principal specimen used in this study but also throughout the sandstorms and flies of a successful though trying field season in Australia in 1970, contributed more than his share of enthusiasm and good cheer in the best "she'll be right" spirit of Australia. I also owe a debt of gratitude to my other companions in the field, namely, Bob Emry, Frank Pearce, and Neville Pledge, who also made that trip memorable, and to our institutions and the National Sci­ ence Foundation (grant number GB 18273x1). The work would never have been so productive without the superb logistic skills 39 40 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY of Paul Lawson of the South Australian Museum, who kept a consistent stream of vital supplies arriving on time and who later expedited their shipment to the United States. We are eter­ nally grateful to Mick and Audrey Sheehan of Moolawatana Station, who put up with our intrusion. Lorraine Meeker and Chester Tarka of the American Museum of Natural History provided the innovative ultraviolet/white-light stereophotos that illustrate this paper. Frank Pearce developed the equally in­ novative wax-impregnation technique that enabled the preser­ vation and preparation of the well-preserved, but frustratingly fragile, Callabonna fossils. ABBREVIATIONS.—The following abbreviations are used: AMNH American Museum of Natural History, New York NMNH National Museum of Natural History, Smithsonian Institution, Washington, D.C. SAM South Australian Museum, Adelaide USNM Collections of the National Museum of Natural History, Smith­ sonian Institution, which include the collections of the former United States National Museum Description of the Basicranium FIGURES 1,2 EXOCCIPITAL.—This element of the most posterior part of the basicranium includes the foramen magnum, the occipital condyles, the condyloid foramina, the posterior rim of the pos­ terior lacerate foramen, and the foramina for the transverse si­ nus within the lateral wall of the foramen magnum. The occipi­ tal condyles are more salient than in living wombats and are laterally separated from the body of the exoccipital by grooves. As in Vombatus, there is an occipital foramen just lateral to the condyles that appears to be open and probably communicates with the transverse sinus. The exoccipital forms only the me­ dial base of the paroccipital process, whereas in both living wombat species the entire process is formed by the exoccipital. BASIOCCIPITAL.—The precise limits of this element are diffi­ cult to discern because of its fusion with adjacent basicranial bones. The basioccipital bears a low median keel that separates two shallow depressions for insertion of the rectus capitus mus­ cles. No foramina pierce this bone. BASISPHENOID.—The precise limits of this bone are also dif­ ficult to determine owing to the advanced ontogenetic age of the specimens available. The anterior end probably lies at the level of the posterior end of the palatine bar, which closes the palatine fenestra. The basioccipital is penetrated by the carotid foramen, which enters the body of the basioccipital near the posterior limit of the bone. The transverse canal appears to be absent or tiny in Phascolonus, as it is in Lasiorhinus, rather than large, as in Vombatus. PTERYGOID.—The pterygoid wing ascends from the poste­ rior end of the palate and passes backward to the level of the foramen ovale. It forms the floor of the opening of the carotid canal as in Vombatus, rather than passing lateral to the canal as in Lasiorhinus. ALISPHENOID.—As in other vombatiforms, the extension of the squamosal into the anterior part of the tympanic space has largely limited the alisphenoid to the medial side of the tempo­ ral opening. In the living wombats, the alisphenoid forms the anterior and ventral rim of the foramen ovale, which lies on the suture between these two cranial elements. In Phascolonus, damage in this area limits observation to the left side of USNM 214696, where the foramen is situated astride the squamosal- alisphenoid suture and opens into the braincase on the lateral side of the median lacerate foramen, as in both species of living wombats. SQUAMOSAL.—Phascolonus lacks the prominent squamosal tympanic wings seen in Vombatus. Instead, the hypotympanic sinus extends deeply anterolaterally into the squamosal, as in Lasiorhinus, without the ventral processes underlying the ecto- tympanic as seen in both living wombats. Also, there is no ex­ tensive hollowing of the epitympanic fossa of the squamosal, as in Lasiorhinus. The morphology more nearly resembles that in Vombatus. The posterior part of the epitympanic fossa is more deeply excavated than in the living genera, forming a chamber into which the external auditory meatus opens. One or two postglenoid foramina lie deep within the fossa above the dorsal process of the ectotympanic, as in Vombatus. A single large subsquamosal foramen lies just above the posterior root of the zygoma. In contrast, living wombats, especially Vomba­ tus, have a suite of small foramina in this position. The squa­ mosal forms the lateral rim of the mastoid process, and its dis­ tribution resembles that of Lasiorhinus. ECTOTYMPANIC.—This bone has a unique form in Phascolo­ nus; rather than being tubular, as in most diprotodont marsupi­ als, it is flattened into a body that bridges the external opening of the tympanum between the squamosal and the mastoid pro­ cess of the periotic. In ventral view, the ectotympanic forms an anteriorly concave, curved structure. It is united with the mas­ toid process by a large squamous suture but apparently only touches the squamosal tympanic wing. The pointed anterome- dial process of the ectotympanic underlies the squamosal wing. The crista tympanica surrounds the slit-like porus acousticus on the anteromedial surface of the ectotympanic. The external auditory meatus is a very small dorsoventrally oriented slit, deeply contained within the epitympanic fossa. The convex posterior surface of the ectotympanic is cupped within a com­ parably curved surface of the mastoid medial to the tip of the mastoid process and partially separated from the mastoid by a narrow space. PERIOTIC—As in some other vombatiforms (thylacoleonids, ilariids, diprotodontids), the paroccipital process in Phascolo­ nus is formed mainly from the mastoid process of the periotic. In the living wombats, the parocciptial process is formed en­ tirely by the exoccipital. A prominent stylomastoid foramen is enclosed by the mastoid process and opens onto the ventral sur­ face behind the ectotympanic. In living wombats, the facial nerve passes, by way of a canal on the ectotympanic-mastoid suture, ventral to the promontorium posteriorly and medially, NUMBER 93 41 FIGURE 1.—Ventral view of the basicranium of Phascolonus gigas, USNM 214696, taken in ultraviolet (A) and composite white and ultraviolet light (B). The associated outline at the same scale (C) identifies the structures discussed in the text. (al=alisphenoid; bo=basioccipital; bs=basisphenoid; caf=carotid foramen; cf=condyloid foramina; et=ectotympanic; ex=exoccipital; frn=fora- men magnum; fo=foramen ovale; hs=hypotympanic sinus; ju=jugal; m=mas- toid; mp=mastoid process; of=occipital foramen; pal=palatine; pet=petrosal; plf=posterior lacerate foramen; pop=paroccipital process; pt=pterygoid; sf= stylomastoid foramen; sq=squamosal.) 5 cm and exits posterior to the external auditory meatus. A similar lateral course for the facial nerve also seems to be the case in other vombatiforms. The mastoid process of the periotic is lat­ erally produced into a thin flange, backed by the squamosal. This plate-like structure is similar in form and composition to that seen in other vombatiforms. At its posteriormost extent, medial to the stylomastoid foramen, it bears a fossa for recep­ tion of the hyoid apparatus. Thepf petrosal is remarkably small and deeply sunken into the hypotympanic space to the level of the crista tympani. In its anteroposteriorly elongate form it closely resembles the petro­ sal of living wombats. In the living wombats, the petrosal lies more superficially in the hypotympanic space, not far above the level of the basicranial bones. Although the petrosals in both skulls of Phascolonus appear to be displaced, they are confined to a relatively narrow space behind the squamosal hypotym­ panic sinus. In reviewing the salient points of the basicranium of Phas­ colonus, two aspects appear to be unique to this taxon among known vombatiforms. The first is the form of the ectotym­ panic. The second is the facial nerve, which exits from a canal in the mastoid, rather than from a groove or canal at the ecto- 42 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY caf FIGURE 2.—Oblique view of the posterior part of the right side of the cranium of Phascolonus gigas, USNM 214696, taken in composite white and ultraviolet light (A) and detailed view (B). The associated outline at the same scale identi­ fies the structures discussed in the text. The rectangle encloses the detailed view. (eam=external auditory meatus; pa=parietal; pgf=postglenoid foramina; so= supraoccipital. Other abbreviations as in Figure 1.) 5 cm tympanic-mastoid suture as in all other diprotodontians. Other­ wise, the basicranium closely resembles living wombats; it lacks only the ventral inflation of the squamosal hypotympanic sinus and the dorsal position of the petrosal in the narrow space between the basioccipital and the ectotympanic. Discussion The distinctive suite of synapomorphies that characterize the Vombatidae, especially the gliriform incisive battery and the strong modification of the skeleton for fossorial existence, has long justified the hypothesis of wombat monophyly. Possession of a squamosal tympanic process has clearly grouped the Vom­ batidae with the other vombatoid families, all of which are ex­ tinct (Marshall et al., 1990). With the inclusion of the fossil record, the recognition of their sister group has been less uni­ versally acknowledged. In one of the earliest cladistic analyses, Archer (1984, fig. 310) placed the wombats in an unresolved trichotomy with "diprotodontoids" (Diprotodontidae + Palorchestidae) and the NUMBER 93 43 Thyl Vom FIGURE 3.—Comparison of the resolution of vombatimorph phylogeny of Mar­ shall et al. (1990), based upon cranial and dental features (top), versus that of Munson (1992), based upon postcranial features (bottom). Thylacoleonidae based on cranial and dental features. Later, Marshall et al. (1990), on the basis of shared loss of the upper canine and of elongation and increased ankylosis of the man­ dibular symphysis, placed the vombatids specifically as sister taxa of the diprotodontids + palorchestids. Other vombatoids were arranged in pectinate fashion below these terminal sister taxa, with the marsupial-lions being the least-derived relative to the suite of cranial and dental characters used. Munson's (1992) study of vombatiform (phascolarctids + vombatoids of Marshall et al., 1990) relationships focused on Ngapakaldia (Diprotodontidae) and Ilaria (Ilariidae). Munson (1992) considered 100 postcranial characters of the families of vombatiforms and successive outgroups, including several spe­ cies within the phalangeriforms, peramelids, dasyurids, and mi- crobiotheres. The analysis was generalized to the family level, and a parsimony algorithm (PHYLIP 2.8 of Felsinstein, 1987) yielded a single tree (no statistics were given) that grouped Vombatidae with Ilariidae as the crown group of a pectinate cladogram in which the Diprotodontidae, Palorchestidae, and Thylacoleonidae were successive taxa included in the Vombati- formes. When this tree is compared with cranial-dental results of Marshall et al. (1990), the only discordance among equiva­ lent taxa is the placement of the ilariids and the relationships among the diprotodontoid taxa (Figure 3). These conclusions represent conflicts that pit a hypothetical dental transformation series against modification of the skele­ ton for digging. Because the ilariids are primitive in possessing a selenodont dentition and the living wombats retain this pat­ tern in the unworn teeth, it seems likely that the postcranial skeleton contains the more phylogenetically valuable signal of phylogeny. Munson (1992) used the skeleton of Phascolonus gigas in her analysis, but she did not find that it diverged mark­ edly from that of living wombats, except in size and features related to size. Phascolonus gigas thus makes no particular contribution to the questions of within-wombat relationships in the postcranial skeleton. Nonetheless, divergent views have been advanced re­ garding the phylogenetic position of Phascolonus within the Vombatidae. Dawson (1981) used cladistic arguments to inves­ tigate the phylogeny of the giant wombats, Phascolonus gigas, Ramsayia magna Owen, and "Phascolomys" medius Owen. Using the distribution of five dental and two cranial characters, she specifically examined the relationships of these three to the two living species. The "Palorchestidae" (now allocated to the Diprotodontidae) were used as an outgroup to establish the po­ larity of these features. The result was a single cladogram (Fig­ ure 4A) (Hypothesis A herein) that grouped Phascolonus with Vombatus and Ramsayia with Lasiorhinus (including "Phasco- lomys" medius, following Marcus, 1976). In 1983, using new material that more clearly showed the palatal features, Dawson returned to the subject of wombat phylogeny and the phylogenetic position of L. medius. In this revised analysis (Dawson, 1983) she used two more characters of the anterior palate; these continued to group "Phascolomys" medius with Lasiorhinus but shifted the position of Phascolo­ nus from a sister taxon of Vombatus (involving four parallel­ isms, as in Figure 4A) to a sister taxon of the Lasiorhinus group, which included Ramsayia and L. medius (two parallel­ isms and two reversals, as shown in Figure 4B). Subsequently, Archer (1984) presented a cladogram (Figure 5) of wombat phylogeny that included the Pleistocene Warendja and the Miocene Rhizophascolonus as successive outgroups. Determination of character polarity was assisted by consideration of other "diprotodontians." This view was sup­ ported by 17 characters, including those considered by Dawson (1983), and favored the arrangement of Phascolonus + "P." medius as a sister taxon of Lasiorhinus and Ramsayia and more distantly Vombatus. A recent analysis by Murray (1998) used a suite of 28 cranial and dental characters in an ostensibly cladistic analysis (Figure 6), which essentially reinforced Archer's (1984) view. Accept­ ing the monophyly of the Vombatidae as having been well sub- 44 RZ SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY WA VO RM LA PM PG FIGURE 4.—Alternative phylogenies for the Vombatidae proposed by Dawson: A, Hypothesis A (1981); B, Hypothesis B (1983). Both are based upon cranial and dental features. (LA=Lasiorhinus; PG-Phascolonus gigas; PM-"Phas- colomys" medius; RM-Ramsayia magna; VO= Vombatus.) stantiated by the data reviewed by Murray, we examine a smaller suite of characters (including 10 of those used by Mur­ ray) to analyze the phylogeny of wombat species. The charac­ ters used by Murray are not explicitly polarized, but they can be if other vombatiforms—diprotodontids, palorchestids, ilari­ ids, and thylacoleonids—are used as outgroups. The following characters and their polarities were used: (1) The highly modi­ fied II of Phascolonus is "transversely implanted" (sensu Mur­ ray, 1998); in the primitive condition, the II is obliquely im­ planted, as seen in Vombatus. (2) The II in Phascolonus is autapomorphic in its extreme breadth compared with its dors­ oventral depth; the enamel is restricted to the anterior face. (3) The P3 has a labial and a lingual groove in the primitive form of this tooth, but in wombats both grooves may be reduced or lacking. Both conditions are usually accompanied by reduction in the size of the tooth relative to the size of the molar rows. Phascolonus retains both grooves. (4) A small P3, shorter than FIGURE 5.—Archer's (1984) phylogeny of the Vombatidae, based upon cranial and dental characters. (Taxa abbreviated as in Figure 4, with RZ (Rhizophas- colonus) and WA (Warendja) added.) Ml, characterizes all wombats except Phascolonus, in which the P3 is nearly as long as the Ml , and possibly the Miocene wombat Rhizophascolonus, which has rooted teeth. (5) An il that is deeper than it is broad is the primitive form in the vom­ batiforms. A flattened lower incisor that either is broader than it is deep or is equidimensional appears to be derived (contrary to the polarity assignment of Dawson, 1981, 1983). (6) The di- astemal palate widens anteriorly in derived taxa, whereas the primitive anteriorly narrowing or parallel-sided condition is seen in Vombatus and other vombatiforms. (7) The incisive fo- RZ UD WA VO RM LA PM PG FIGURE 6.—Murray's (1998) phylogeny of the Vombatidae, based upon cranial and dental characters. (UD=undescribed Riversleigh (Miocene) form; other abbreviations of taxa as in Figures 4 and 5.) NUMBER 93 45 TABLE 1.—Distribution of character states among taxa of Vombatidae discussed in the text. Character 1. 11 transversely implanted 2. 11 broad 3. P3 lab/ling grooves shallow or lacking 4. P3 small 5. il broad 6. Palate wide anteriorly 7. Incisive foramen in deep fossa 8. Incisive foramen posterior 9. Palate emarginated 10. Upper tooth rows converge 11. Fossa superficial, masseter dorsal 12. Median premaxillary process 13. Masseteric foramen large 14. Squamosal epitympanic wings inflated 15. Hypotympanic sinus large 16. Epitympanic fossa small Other vombati- morphs 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 Warendja 0 0 1 1 0 0 0 0 0 0 0 0 1 1 ? 1 Vombatus 0 0 1 1 1 0 0 0 0 1 0 0 1 1 0 1 Lasiorhinus 1 0 1 1 0 0 1 1 1 0 ? 1 0 7 ? ? Ramsayia magna 1 0 1 1 0 1 0 0 0 0 1 1 0 1 1 0 "Phascolomys" medius 1 0 1 1 0 1 1 1 1 0 1 0 0 ? ? ? Phascolonus gigas 1 1 0 0 0 1 1 1 1 1 1 1 0 0 1 1 ramina lie in a concavity in the diastemal palate in Lasiorhinus and the giant wombats. The primitive anterior palate of Vomba­ tus is nearly flat. (8) In the giant wombats, the incisive foram­ ina lie posteriorly, just anterior to P3, in a deep pocket at the rear of the diastema. This posterior position of the foramina correspondingly displaces the premaxillary-maxillary suture to a position just anterior to the cheek-tooth row. This is a highly derived condition that is not clearly foreshadowed in Lasiorhi­ nus. It is present in both Phascolonus and Ramsayia, even though the latter lacks the diastemal flare associated with the wide incisors that is so prominent in the former. (9) In the giant wombats, the strong emargination of the palate between the cheek teeth forms a trough that passes dorsally into the deep di­ astemal trench. (10) The upper cheek teeth are strongly conver­ gent anteriorly, particularly across the anterior part of Ml. Al­ though the cheek-tooth rows of Lasiorhinus do converge, the condition in Vombatus and Phascolonus is extreme in its nar­ rowing palate. (11) The fossa for the superficial masseter ante­ rior to the orbit rises to the dorsal surface of the skull in Lasi­ orhinus, Phascolonus, and "Phascolomys" medius. This condition gives rise to a strong notch in front of the zygomatic root in these taxa. The masseter origin lies mostly below the or­ bit in Vombatus and other vombatiforms. (12) A median pre­ maxillary process is well developed in Phascolonus and Ram­ sayia and is present as a low spine at the midline dorsal surface of the premaxillaries in Lasiorhinus. This feature is shared only with Diprotodon among other vombatiforms. (13) In Vombatus and Warendja the masseteric foramen is enlarged. This fora­ men pierces the masseteric fossa and opens in the vicinity of the mandibular foramen on the medial side of the ascending ra­ mus of the mandible. It is small or lacking in other wombats and in other vombatiforms. (14) The squamosal epitympanic processes, or "wings," are inflated and terminate in pointed processes in the living wombats. This derived condition con­ trasts with the lack of inflation in Phascolonus and other vom­ batiforms. (15) The hypotympanic sinus extends anterolaterally into the zygomatic root of the squamosal in Lasiorhinus and Phascolonus. This is a derived condition relative to Vombatus and other vombatiforms. (16) The epitympanic fossa lies above the ectotympanic and lateral to the hypotympanic sinus and is extensive in Lasiorhinus. This condition is widely present in other vombatiforms and therefore seems primitive for that group as a whole. In Vombatus and Phascolonus this structure is shallow, and the division between the posterior segment, into which the external auditory meatus opens, and the larger ante­ rior segment is marked by a low ridge. These characters are distributed as shown in Table 1. Analy­ sis using the maximum parsimony program of Hennig86 (Far­ ris, 1988) and the distribution of synapomorphies found by ClaDos (Nixon, 1992) resulted in three trees of equivalent length (23 steps) and statistics (Consistency Index=69 and Rescaled Consistency Index=66) (Figure 7). All three trees confirm essentially the same phylogenetic hypotheses and dif­ fer only in the resolution of relationships among the giant wombats. They also agree with Dawson (1983, Hypothesis B), Archer (1984, fig. 64), and Murray (1998) in grouping the giant wombats with Lasiorhinus. Support for this relationship on all trees lies with three synapomorphies: the transversely im­ planted II (1); the dorsal extension of the origin of the superfi­ cial masseter (11); and a large hypotympanic sinus (15). The program assumed that the last two features would be found in Ramsayia. The more-resolved cladograms assume either that the wide anterior palate (6) was acquired twice in this clade or that it was lost in Ramsayia. In fact, however, the structure of the anterior palate in Lasiorhinus, although wider than in Vom­ batus, may not be homologous with the condition in the giant forms, in which this state is reinforced by the depth and posi­ tion of the incisive foramina. A number of reversals to the 46 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY r- VB 3 4 14 16 I I I I i - VB 3 4 14 16 I I I ! VB 3 4 14 16 I I I I 1 1112 16 I I I I 6 16 i-0-O-L4 1 0 7 8 9 14 I • I o- 1 1 1 0 v-RM *i 2 3 4 10 1 0 0 1 i - WA 6 10 1-0- i 1 - D IU Hr-0- VO 16 r + M 1 6 11 12 16 I I I I I 1 1 1 1 1 7 8 9 14 • I I 0- 0 %-PM 2 3 4 10 H-OHHhPG 1 0 0 1 i - WA 6 10 1 HHr - VO 1 1 1 6 11 16 I I I I 12 16 I-0-0-L4 1 0 v- PM 7 8 9 14 1 1 1 0 r-Q-RM 0 2 3 4 10 L -HHHhPG r- WA 6 10 Hh 1 1 - O IU H H h VO FIGURE 7 (left).—Alternative cladograms of relationships of vombatid taxa discussed in the text. These trees have identical statistics (length 23, Consis­ tency Index=69, Rescaled Consistency Index=66) but differ in the resolution of relationships of the giant wombats. Cladograms were generated by the parsi­ mony algorithm of Hennig86 (Farris, 1988), and the distribution of character states used by ClaDos (Nixon, 1992). (VB=other vombatiforms; other abbrevi­ ations of taxa as in Figures 4 and 5.) primitive state (five, nearly one-third of the data, in the best-re­ solved arrangements) must be assumed in grouping Lasiorhi­ nus with the giant wombats, an arrangement that reinforces the phyletic isolation of this clade from Vombatus. Phascolonus, with its large, bilobed P3, remains primitive in some respects. Its converging upper tooth rows may be, like its uniquely wide II, an autapomorphy. Additional unique features are the plate­ like ectotympanic (primitive) and enclosed stylomastoid fora­ men, both of which serve to further isolate this taxon phyloge- netically. Conclusions Acquisition of two crania of the giant wombat Phascolonus gigas from the Pleistocene of Lake Callabonna by a joint Smithsonian Institution-American Museum of Natural History- South Australian Museum expedition in 1970 allowed descrip­ tion of the basicranium and a reevaluation of previous attempts to place this taxon in wombat phylogeny. The basicranium re­ veals a unique combination of features; some of these are syna­ pomorphies with other wombats, whereas others appear to be autapomorphies of Phascolonus. The new data, and those ad­ vanced by previous studies, can be used in a cladistic analysis of phylogeny to reinforce previous conclusions that Phascolo­ nus and the other giant wombats, Ramsayia and "Phasco- lomys" medius, form a clade with the living hairy-nosed wombat, Lasiorhinus. They are united with their sister taxa— the living naked-nosed wombat, Vombatus, and the fossil Warendja—by a single synapomorphy of a mandible bearing a large masseteric foramen. The well-established monophyletic wombat clade is based upon four characters noted herein and is reinforced by many others in the skeleton, soft anatomy, and biochemistry noted by previous authors (summarized by Mur­ ray, 1998). NUMBER 93 47 Literature Cited Archer, M. 1984. The Australian Marsupial Radiation. In M. Archer and G. Clayton, editors, Vertebrate Zoology and Evolution in Australasia: Animals in Space and Time, pages 633-808. Carlisle, Western Australia: Hesperian Press. Dawson, L. 1981. The Status of the Taxa of Extinct Giant Wombats (Vombatidae: Marsupialia) and a Consideration of Vombatid Phylogeny. Austra­ lian Mammalogy, 4:65-79. 1983. On the Uncertain Generic Status and Phylogenetic Relationships of the Large Extinct Vombatid Species Phascolomys medius Owen, 1872 (Marsupialia: Vombatidae). Australian Mammalogy, 6:5-13. Farris, J.S. 1988. Hennig86, Version 1.5. New York: Port Jefferson Station. Felsinstein, J. 1987. PHYLIP Manual, Version 2.8. Berkeley: University Herbarium, University of California. Marcus, L.F. 1976. The Bingara Fauna: A Pleistocene Vertebrate Fauna from Murchi- son County, New South Wales, Australia. University of California Publications in Geological Sciences, 114: 145 pages. Marshall, L.G., J.D. Case, and M.D. Woodburne 1990. Phylogenetic Relationships of the Families of Marsupials. In H.H. Genoways, editor, Current Mammalogy, 2:433-505. New York: Ple­ num Press. Munson, C.J. 1992. Postcranial Description of Ilaria and Ngapakaldia (Vombatiformes, Marsupialia) and the Phylogeny of the Vombatiforms Based on Postcranial Morphology. University of California Publications in Zoology, 125:1-99. Murray, P.F. 1998. Palaeontology and Paleobiology of Wombats. In R.T. Wells and P.A. Pridmore, editors, Wombats, pages 1-33. Chipping Norton, New South Wales: Surrey Beatty and Sons in association with the Royal Zoological Society of South Australia. Nixon, K.C. 1992. ClaDos, Version 1.2. Ithaca, New York: Cornell University. Owen, R. 1858. Odontology. In The Encyclopedia Britannica, or Dictionary of Arts, Sciences and General Literature. Eighth edition, 16:447, 450, text figure 80. 1872. On the Fossil Mammals of Australia, Part VII; Genus Phascolomys: Species Exceeding the Existing Ones in Size. Philosophical Trans­ actions of the Royal Society of London, 162:241-258, plates 32-40. Ramsay, E.P. 1880. [Untitled.] The Sydney Morning Herald, 13286, 30 October 1880, page 5, column 4. Stirling, E.C. 1913. On the Identity of Phascolomys (Phascolonus) gigas, Owen, and Sceparnodon ramsayi, Owen, with a Description of Some Parts of the Skeleton. Memoirs of the Royal Society of South Australia, 1:127-178, plates 40-58. Wells, R.T., and R.H. Tedford 1995. Sthenurus (Macropodidae: Marsupialia) from the Pleistocene of Lake Callabonna, South Australia. Bulletin of the American Mu­ seum of Natural History, 225: 112 pages. Vassallia maxima Castellanos, 1946 (Mammalia: Xenarthra: Pampatheriidae), from Puerta del Corral Quemado (Late Miocene to Early Pliocene), Catamarca Province, Argentina Gerardo De Iuliis and A. Gordon Edmund A B S T R A C T A specimen in the Field Museum of Natural History is the first record of a pre-Pleistocene South American pampathere that pre­ serves the skull, dentary, and numerous osteoderms of a single individual. We confidently assign the specimen, FMNH P14424, to Vassallia maxima Castellanos after comparing it with the man­ dible and osteoderms of the type of this species and with other material in South American museums. FMNH P14424 is from fos­ siliferous deposits near Puerta del Corral Quemado, Catamarca Province, Argentina. Field notes are ambiguous with regard to its precise stratigraphic level, and therefore age, but the section from which it was recovered includes both the Huayquerian and the Montehermosan. The range of osteodermal variation of V. maxima FMNH P14424 encompasses that of Plaina intermedia Ameghino and P. subintermedia Rovereto, which suggests that these genera and spe­ cies may not justifiably be distinguished. If this is confirmed, then Vassallia is the valid genus and V. intermedia (Ameghino, 1888) is the valid specific name for the species; however, until the status of these taxa is unambiguously demonstrated, we maintain V. max- Introduction The pampatheres represent a distinct lineage of cingulates (Hulbert and Morgan, 1993) and are known with certainty from the late Miocene to late Pleistocene of South and North Amer­ ica (Scillato-Yane, 1980). Traditionally, they have been consid- Gerardo De Iuliis, Department of Zoology, University of Toronto, 25 Harbord Street, Toronto, Ontario M5S 3G5, Canada, and Faculty of Community Services and Health Sciences, George Brown College of Applied Arts and Technology, 200 King Street East, Toronto, Ontario M5A 1J5, Canada. A. Gordon Edmund, Department of Geology, Earth Sciences Centre, University of Toronto, 22 Russell Street, Toronto, On­ tario M5S1A1, Canada. ered to be intermediate in general morphology between dasy- podids and glyptodonts, although Simpson (1930) rejected this concept. Pampatheres bear pectoral and pelvic bucklers, formed from sutured osteoderms; the bucklers are separated by only three transverse movable or imbricating bands, whereas dasypodids usually possess more than three movable bands. Furthermore, each osteoderm bears only a single keratinized scute in pampatheres, whereas more than one scute usually contributes to the keratinized covering of an osteoderm in other cingulates. Edmund (1987) listed additional pampathere char­ acters. Pampatheres have long been regarded as distinct from both dasypodids and glyptodonts, although Ameghino (1889), fol­ lowed by various authors (e.g., Castellanos, 1927), considered them probably ancestral to the glyptodonts but more closely re­ lated to the true armadillos. Patterson and Pascual (1972) and Engelmann (1985), however, believed that they were more closely related to glyptodonts, and the latter author formally placed pampatheres with glyptodonts and eutatine armadillos in a group termed the Glyptodonta. For much of the last century, researchers (e.g., Paula Couto, 1954; Hoffstetter, 1958; Robertson, 1976; Scillato-Yane, 1980; Cartelle and Bohorquez, 1985; Patterson et al., 1989; Cartelle et al., 1991) have usually followed Lydekker (1894) in classify­ ing the Pampatheriinae (=Chlamydotheriinae) within the Dasy- podidae. Earlier workers, such as Moreno and Mercerat (1891) and Ameghino (1889), recognized pampatheres as the family Pampatheriidae (=Chlamydofheriidae or Chlamydotheridae). Bordas (1939), exceptionally, raised them to superfamilial rank as the Chlamydotherioidea. More recently, there has been a tendency among researchers (e.g., Edmund, 1985a, 1987; Downing and White, 1995; Edmund and Theodor, 1997) to consider them a separate family. We follow this, pending a comprehensive phylogenetic analysis of cingulates. 49 50 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY The Pleistocene pampatheres are represented by eight spe­ cies distributed in the genera Pampatherium and Holmesina (including the North American Blancan H. floridanus Robert­ son; see also Morgan and Hulbert, 1995). They are reasonably well represented by osteoderms and other skeletal elements, and their taxonomy, based largely upon osteoderms, is re­ viewed by Edmund (1996). Except for some of these Pleis­ tocene species, particularly H. septentrionalis Simpson and H. floridanus, pampatheres are not well defined. Little is known or published of their osteological variation at the specific or generic levels, although much of the research has focused on osteoderms. For example, Edmund (1987) and Hulbert and Morgan (1993) documented changes in the size and morphol­ ogy of osteoderms and various postcranial elements; and Ed­ mund (1996) recognized Pleistocene species solely on the ba­ sis of osteodermal characters. Among the reasons that workers have relied on osteoderms is that these elements are fre­ quently recovered. Unfortunately, however, they are rarely found in association with skull or postcranial remains (Ed­ mund, 1996). Other than H. floridanus from the late Blancan, pre-Pleis- tocene pampatheres are South American and are known mainly from Argentina and Colombia. Four genera (see Edmund and Theodor, 1997) are currently recognized: Kraglievichia, Plaina, Scirrotherium, and Vassallia (excluding the fragmen­ tary remains from the Patagonian Eocene assigned to Machly- dotherium; see Edmund, 1985a). Ameghino (1883) described K. paranense from isolated osteoderms. Castellanos (1927) amplified the description and included unassociated skull and postcranial material from the type area. Plaina and Vassallia are based upon sparse remains, mostly isolated osteoderms. Among the difficulties with relying on osteoderms from differ­ ent localities is that intraspecific or age variation, or even dif­ ferences between osteoderms of a single individual, may be ac­ corded taxonomic importance. Ultimately one must learn for each taxon the range of osteodermal variation. That goal is complicated by the fact that the armor of pampatheres "consists of specialized osteoderms covering the head, body, tail, and limbs" that vary dramatically in size, shape, and ornamenta­ tion, depending upon the type of osteoderm and its precise lo­ cation (Edmund, 1985a:6). Identification or erection of new taxa based upon small samples of osteoderms is unreliable. The specimen described herein, FMNH P14424, is assigned to V. maxima Castellanos on the basis of its dentary and osteo­ derms. It is the first record of a pre-Pleistocene South Ameri­ can pampathere that preserves nearly completely the skull and dentary of a single individual, along with a large number and variety of almost certainly associated osteoderms of the same individual. Thus, it expands the knowledge of variation in pam­ patheres and permits a considerably more complete description and understanding of pre-Pleistocene genera. FMNH P14424 was assigned to Plaina sp. by Marshall and Patterson (1981) and Marshall et al. (1983). Patterson et al. (1989) referred it tentatively to P. subintermedia Rovereto, and Edmund (1985a) referred it to Plaina sp. but incorrectly cited FMNH P14414 as its catalog number. ABBREVIATIONS.—The following museum abbreviations are used: AMNH American Museum of Natural History, New York FMNH Field Museum of Natural History, Chicago IFG Instituto de Fisiografia y Geologia, Rosario, Santa Fe, Argentina INSP Instituto Nacional Superior del Profesorado, Parana, Entre Rios, Argentina MACN Museo Argentino de Ciencias Naturales "Bernardino Rivadavia," Buenos Aires MLP Museo de La Plata, La Plata, Argentina MMP Museo Municipal de Ciencias Naturales "Lorenzo Scaglia," Mar del Plata, Argentina MNA Museum of Northern Arizona, Flagstaff ROM Royal Ontario Museum, Toronto UF Florida Museum of Natural History, Gainesville, Florida In the description of material, the following abbreviations are used: D L M or m Mt Ph R digit left molariform upper or lower, respectively (for convenience molar- iform is used for all teeth even though not all are such) metatarsal phalanx right. ACKNOWLEDGMENTS.—We thank the numerous curators of the various institutions for allowing us to examine specimens in their care. We appreciate the efforts of D.D. Gillette (MNA), S.F. Vizcaino (MLP), and S.D. Webb (UF) in reviewing the manuscript; their comments and suggestions significantly im­ proved the quality of this paper. We benefited greatly from dis­ cussions with G.J. Scillato-Yane (MLP), who freely shared his insights on and knowledge of pampatheres. We are grateful to W.F. Simpson (FMNH) for his research on the stratigraphic po­ sition of the specimen; to E. Knapp and L. Yuik (Department of Zoology, University of Toronto), for their help with the photog­ raphy; to M.J. Lo Presti, V.A. Stock, D. Stonehouse, and N. Taccogna for accommodating the schedule of GDI; and to P. Gisondi for various logistic support during many years. This paper was funded in part by National Science and Engineering Research Council Grants A1760 to C S . Churcher and A1810 to A.G. Edmund. TAXONOMIC HISTORY Castellanos (1927, 1937) created the generic names Krag­ lievichia, Vassallia, and Plaina for pampatheres of the late Mi­ ocene and early Pliocene of Argentina using materials previ­ ously assigned by others to Chlamydotherium (=Pampathe- rium; see Edmund, 1985b). Kraglievichia is based upon the posterior part of a right dentary, which Ameghino (1883) used to erect C. paranense, from the left bank of the Rio Parana, En­ tre Rios Province, Argentina. The deposits in the Argentinian Mesopotamia were commonly considered as "Mesopotamian" NUMBER 93 51 but are now assigned to the Huayquerian or Montehermosan on the basis of geographic position (Marshall et al., 1983). Cas- tella-nos (1927:2; 1937:15) considered Ameghino's specimen a "metatypo" and stated that it was housed in MLP. Scillato-Yane (1980, 1982) and Mones (1986:231), however, listed the type of C. paranense as "INSP ?"; the latter author indicated that the type is lost and that MLP has casts of the type (MLP M-112, M-227, M-228). Using comparisons with the posterior end of a right dentary (i.e., Ameghino's metatypo, according to Castellanos, 1927, 1937), Castellanos (1927, 1937) referred additional remains to K. paranense, including a nearly complete, but edentulous, skull (MACN 2617), a left palatal fragment (MACN 8943), a large part of a left dentary (MACN 8968), and other remains. These were recovered from the "Mesopotamiense de la For- macion Entrerriana de la margen izquierda del rio Parana" (Castellanos, 1927:2). Castellanos (1937) provided more pre­ cise locality information for MACN 2617, which indicates, ac­ cording to criteria suggested by Marshall et al. (1983), that the skull is of Huayquerian age. Kraglievich (1934) assigned an os­ teoderm from the Chapadmalal region to Kraglievichia. Pas­ cual et al. (1966) reported K. paranense from the Monte Her­ moso Formation and the Chapadmalal Formation in Buenos Aires Province. Castellanos (1927) transferred C. intermedium Ameghino, 1888, and C. subintermedium Rovereto, 1914, to Kraglievichia (vide infra). Vassallia Castellanos, 1927, is based upon the species C. minutum from the "Araucanian" of Catamarca. Moreno and Mercerat (1891) had described this species from a right dentary and three questionably associated osteoderms, which were il­ lustrated by Lydekker (1894, pl. 33: figs. 1,1a, and figs. 2, 3a, respectively). Scillato-Yane (1980, 1982) and Mones (1986) listed the dentary, MLP 29-IV-15-4, as the type of the species. Marshall and Patterson (1981) suggested that the remains were recovered from the Andalhuala Formation, to which Marshall et al. (1983) assigned a Huayquerian age. Castellanos (1946:6) described the larger V. maxima from re­ mains recovered from various localities "del Araucanense me­ dio y superior" in the Valle de Yocavil (= Santa Maria; Marshall and Patterson, 1981), Tucuman and Catamarca Provinces, Ar­ gentina. The type, IFG 500, preserves the anterior part of a right dentary and a few imbricating osteoderms "de la margen derecha del arroyo de la Totora, al N.W. de 'Los Nacimientos,' Valle de Yocavil, provincia de Tucuman" (Castellanos, 1946:6). These remains are reasonably considered Huayque­ rian by Marshall et al. (1983). It is worth noting that some of the referred osteoderms are from Loma Rica, near Andalhuala, probably very close to the deposits yielding V. minuta. Castellanos (1937) erected Plaina to include P. intermedia and P. subintermedia. Ameghino (1888) based the former spe­ cies upon isolated osteoderms from Monte Hermoso. Mones (1986) listed the type as MACN?, suggesting that it is lost, al­ though the title of the article in which Ameghino (1887) first mentioned the osteoderms suggests that the type was in MLP. Edmund, in 1986, located a single buckler osteoderm, MACN A7799, labeled by F. Ameghino, which is possibly the type. Ameghino (1895) incorrectly identified " C " minutum as a syn­ onym of this species (Castellanos, 1937). Anaya and MacFad- den (1995) assigned pampathere remains from the Pliocene In- chasi fauna of Bolivia to P. intermedia. Rovereto (1914) erected the species P. subintermedia on the basis of various osteoderms, MACN 8473 (Mones, 1986), and described and figured them as from the Huayquerias region of Mendoza Province, Argentina. Pascual and Odreman Rivas (1973) and Marshall et al. (1983) noted that these remains (i.e., those listed by Rovereto, 1914:214-224) are actually from the Tunuyan Formation and are of Montehermosan age. Plaina was listed as a Montehermosan mammal by Marshall et al. (1983). Castellanos (1958) typified P. brocherense Castellanos on various osteoderms (IFG 769) from Los Reartes Valley, Cordoba Province, Argentina. The fauna from Los Reartes and Nono Valleys are now believed, however, to "represent a mix­ ture of Montehermosan and Ensenadan taxa" (Marshall et al., 1984:26). Robertson (1976) proposed that Plaina be synony­ mized with Kraglievichia. FMNH P14424 is assigned to Vassallia maxima on the basis of its similarity in size and morphology to a partial dentary and osteoderms of the type of this species, IFG 500. Vassallia max­ ima is approximately twice the linear size of V. minuta, al­ though the osteoderms of these species are similarly orna­ mented. The dentary of V. maxima is considerably more robust, with a relatively larger angular process, than that of V minuta. GEOLOGY FMNH P14424 was recovered from the fossiliferous depos­ its near the Puerta del Corral Quemado, Catamarca Province during the 1926 FMNH field expedition to South America un­ der the supervision of E.S. Riggs (Marshall and Patterson, 1981). These authors listed FMNH P14424 as being from units 15-23 of the section, but W.F. Simpson (pers. comm., 1995) of the FMNH stated that the original field notes suggest less pre­ cision and indicate only that these remains may have been re­ covered from the section between units 16 and 32. This is es­ sentially the entire fossiliferous section of the Corral Quemado section and includes the upper part of Stahlecker's Araucan­ ense and his Corral Quemado Formation (Riggs and Patterson, 1939), which are currently considered Huayquerian (late Mi­ ocene) and Montehermosan (early Pliocene), respectively. Marshall and Patterson (1981:71) listed FMNH P14424 as including "skull, paired mandibles, ...leg scutes, and foot bones" among its skeletal elements. These elements and a few additional osteoderms were originally, and for a considerable time, labeled with this catalog number; their field number is 334. W.F. Simpson's investigation of original FMNH docu­ ments, however, strongly suggests that numerous additional os­ teoderms and the axis, which had remained uncataloged, were part of field number 334, and thus that these additional ele- 52 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY . . r y, •*' I / - ' 9 FIGURE 1.—Lateral view of the skull of Vassallia maxima FMNH P14424. (Scale bar=50 mm.) ments, which are identical in preservation, were also found in association and probably belong to the same individual. These element were recently labeled as FMNH P14424. FMNH P15302 includes three osteoderms of similar orna­ mentation to that of FMNH P14424 but different preservation. They were recovered from the same locality, but the horizon was unrecorded (Marshall and Patterson, 1981), and they are not associated with FMNH P14424. Systematics Infraorder CINGULATA Illiger, 1811 Family PAMPATHERIIDAE Paula Couto, 1954 Genus Vassallia Castellanos, 1946 Vassallia maxima Castellanos, 1946 DIAGNOSIS.—Medium-sized pampathere, with marked pala­ tal flexion, prominently inflated frontals, strongly tapered ros­ trum, and robust zygomatic arches. Sagittal crest weak, but more prominent than in other pampatheres. Temporal fossa large. Lacrimal large, with elongated dorsal portion. Cranial length relative to facial length approximately intermediate be­ tween condition in Kraglievichia, in which cranium is rela­ tively longest, and Holmesina and Pampatherium. Hard palate TABLE 1.—Skull measurements (in mm) for FMNH P14424. Skull feature Greatest width across zygomatic processes of squamosal Greatest width across anterior part of zygomatics Greatest width across pterygoid processes Occipital width Bicondylar width Postorbital constriction Width across postorbital processes Maximum width across palate (between M6s) Tooth-row length Length of temporal fenestra and orbit Occipital height Maximum height of zygomatic Measurement 69x2=138 64x2=128 51 97 75 52 79 28 140 49 59 42 extending farther posteriorly than in Kraglievichia, but not as far as in Holmesina and Pampatherium. Dentary robust, with relatively short but deep horizontal ramus and prominent angu­ lar process, in comparison with those of Holmesina and Pam­ patherium. Axis possibly, but not certainly, fused to other cer­ vical vertebrae (if so, this complex is morphologically distinct from cervical tube of other pampatheres and dasypodids); ele­ ment compressed anteroposteriorly and very large in relation to size of skull; axis apparently bearing transverse processes. DESCRIPTION.—Skull: The skull of FMNH P14424 is nearly complete and only slightly distorted (Figures 1, 2). Mea­ surements of the skull and upper dentition are presented in Ta­ bles 1 and 2, respectively. The skull lacks the posterior part of the left zygomatic arch, the posterior one-third of the margin of the sagittal crest, and most of the molariform teeth. Its longitu­ dinal axis veers anteriorly to the right because the left temporal region is damaged, but the right side is virtually intact. The dorsal midline is clearly indicated by the nasal suture, which is TABLE 2.—Dental measurements for FMNH P14424. Transverse widths for M4-M9 and m5-m9 include widths of mesial and distal lobes on the left and right sides, respectively, of the slash. (*=alveolar.) Tooth Ml* M2* M3* M4* M5* M6 M7* M8 M9 m2 m3* m4* m5* m6 m7 m8 m9 Medio distal length 6.8 8.0 8.5 14.5 18.5 19.0 17.5 16.7 13.7 7.2 7.4 11.5 18.5 19.5 17.0 15.7 12.8 Transverse width 4.5 5.5 6.1 5.5/6.6 7.9/8.0 8.4/8.6 8.0/8.5 7.7/7.5 6.0/7.0 5.1 5.1 5.5 5.7/7.0 8.3/7.0 7.8/7.0 6.2/6.2 5.3/5.1 Minimum width between alveoli 5.0 8.5 13.0 16.3 19.6 22.0 26.3 25.7 25.4 _ _ _ _ _ _ _ - NUMBER 93 53 FIGURE 2.—Skull of Vassallia maxima FMNH P14424: a, dorsal view; b, ventral view. (Scale bar=50 mm.) open at its anterior one-third, and the prominent sagittal crest. Dorsoventral distortion is not apparent. In lateral view (Figure 1) the outline resembles those of Holmesina (ROM 3881) and Pampatherium (Bordas, 1939, figs. 1, 2, pis. 1, 2; Winge, 1915, pis. 9, 10), compared with the lower, elongated, and more strongly tapered skull of Kraglievi­ chia (MACN 2617; Castellanos, 1937, fig. 8). The palate of FMNH P14424 is longitudinally concave anterior to the alveo­ lar septum between M7 and M8, owing to its anteroventral in­ clination anterior to the distal margin of the M5 alveolus. Pala­ tal flexion is less in Holmesina and Pampatherium and is slight in Kraglievichia. The nasals are broad and long but widen slightly posteriorly (Figure 2a). Their anterior tips appear to turn slightly ventrally, but this may be the result of the considerable fracturing and re­ construction in this area of FMNH P14424. The naso-frontal suture is nearly rectilinear and extends posteromedially at an angle of approximately 42° to the midline. In Holmesina the lateral part of the suture passes medially and then curves abruptly posteromedially to the midline, so that the posterome- 54 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY dial tips of the nasals taper markedly posteriorly. Furthermore, the nasals in Holmesina are narrower posteriorly and widen an­ teriorly. The lateral margin of the external naris is formed by the premaxilla, which is deeply embayed between its dorsal and ventral rami. The lateral surface of the rostrum is raised and rugose from the middle of M2 to the distal end of M4, pre­ sumably for the attachment of the buccinator. Posterior to this, the surface of the maxilla is depressed between the middle of M5 and the infraorbital canal, which lies over the distal one- half of M6. The rostrum tapers more strongly in FMNH P14424 than in known crania of Holmesina, Pampatherium, and Kraglievi­ chia. This appearance may be partly caused by the strongly in­ flated frontals, although in anterior view the narial opening is considerably flattened relative to that of Holmesina. Anterior to the temporal lines, the frontals and adjacent parts of the nasals, lacrimals, and maxillae form a low dome over the orbital re­ gion, approximately between the distal end of M5 and the mid­ dle of M8 (Figure 1). The frontals in Kraglievichia and Pam­ patherium are considerably less domed; Holmesina is approximately intermediate between these and FMNH P14424. Anterior to and between the temporal lines, the frontals of FMNH P14424 are nearly flat, whereas they are depressed in Holmesina. The frontals continue posteriorly past the postorbital con­ striction and meet the parietals along a nearly transverse suture approximately at the level of the anterior margin of the poste­ rior zygomatic root (Figure 2a). The frontal forms the anterior part of the temporal fossa and the posterior part of the orbit. The parietals bear nearly all of the sagittal crest, which is ap­ parently more prominent and raised in FMNH P14424 than in other pampatheres. They meet the occipitals dorsally and later­ ally at the posterior margin of the temporal fossae, and they are excluded from the nuchal crest. The parietals and squamosals meet in a gently sinuous suture and form the posterior part of the temporal fossa. The surface of the fossa, including all of the parietal and adjacent part of the squamosal, is penetrated by nu­ merous, large vascular foramina. The nuchal crests, formed mostly from the occipitals, are prominent and more robust than in Holmesina. In Holmesina and Pampatherium, however, the crests extend farther posteri­ orly, so that the dorsolateral surfaces of the occiput are con­ cave, whereas in FMNH P14424 they are nearly flat. In lateral view the crests slope more markedly posterodorsally in Holme­ sina and Pampatherium than in FMNH P14424 and Kraglievi­ chia. The crest becomes less robust ventrally as it extends into the mastoid process. The squamosal and occipital meet for about 10 mm on the posteroventral surface of the fossa. Ventral to this, a small, rect­ angular expansion of the squamosal onto the occiput forms a small part of the nuchal crest (Figure 1). The ventral part of this expansion forms the dorsal margin of the mastoid foramen. A similar, though larger and ventrolaterally more elongated, ex­ pansion occurs in Dasypus (e.g., ROM R676, R948, R2019). In dorsal view the temporal fossa is relatively larger in FMNH P14424, particularly anteromedially, than in other known pampatheres. The size of the fossa is reflected in the form of the temporal lines and the width across the frontal re­ gion. In FMNH P14424 the temporal line extends nearly medi­ ally from the postorbital process and then curves abruptly pos­ teromedially into the sagittal crest (Figure 1). In Holmesina it curves posteromedially in a gradual curve. The frontal region is wider in Holmesina, and it is reflected in dorsal view in the form of the margin of the anterolateral skull wall, which is con­ vex, or bulging, and elongated between the postorbital process and the postorbital constriction. The condition in Pampathe­ rium resembles that of Holmesina. In FMNH P14424 and Kra­ glievichia this wall is concave and shorter. The postorbital constrictions in FMNH P14424, Holmesina, Kraglievichia, and Pampatherium are expressed about equally, but the positions of the constrictions differ. In dorsal view the braincase of FMNH P14424 is longer relative to the facial re­ gion. The length of the cranium is nearly 92 mm, measured parasagittal^ from the postorbital constriction to the nuchal crest. An equivalent distance measured anteriorly from the pos­ torbital constriction almost reaches the mesial part of M5, which is well anterior to the anterior zygomatic root. A similar condition occurs in Kraglievichia, although the equivalent length may reach even farther anteriorly. In Holmesina the equivalent length reaches only to near the alveolar septum be­ tween M6 and M7, as is also the condition in Pampatherium. The zygomatic arch is more robust in FMNH P14424 than in Holmesina and Pampatherium (the arch is not preserved in specimens of Kraglievichia), so that the ventral margin of the orbit, formed by the descending process of the maxilla, the maxillary process of the zygomatic, and the lacrimal is shal­ lower. The ventral margin of the arch is sculptured and anterolater- ally wider in FMNH P14424, whereas in Holmesina it is rela­ tively smooth and thin, and it tapers ventrally. The anteroven­ tral surface of the arch of FMNH P14424, as in Holmesina and Pampatherium, bears a prominent, expanded, and rugose tuber­ osity, probably for the origin of the superficial masseter muscle (Figures 1, 2b). In FMNH P14424 the tuberosity is more robust and crescentic in ventral view, curving anteromedially. In Holmesina and Pampatherium the tuberosity is oriented nearly anteroposteriorly. In Holmesina the zygomatic extends poste­ rior to the tuberosity as a thinner plate. The arch of FMNH P14424 is also more robust posteriorly. For example, the an­ teroposterior length of the posterior root is almost equal in FMNH P14424 and H. occidentalis Hoffstetter (ROM 3881), although the latter is considerably larger. The more robust char­ acteristics of the arch, larger temporal fossae, and more promi­ nent sagittal crest of FMNH P14424 suggest a larger, more powerful masseteric musculature than in other pampatheres. The relationship between form and function of the masticatory system in pampatheres was discussed by Vizcaino et al. (1998) and De Iuliis et al. (2000). NUMBER 93 55 An additional difference is that the squamoso-zygomatic su­ ture occurs more anteriorly in Holmesina and in Pampathe­ rium, so that the squamosal and zygomatic contribute almost equally to the arch in lateral view. In FMNH P14424 the squa­ mosal makes a smaller contribution to the arch. It is worth not­ ing that the posterodorsal tip of the zygomatic, at the squa­ moso-zygomatic suture, projects dorsally beyond the zygo­ matic process of the squamosal, as occurs also in other pam­ patheres for which this character is known. This is not apparent in Figure 1, because the tip of the zygomatic is broken, but it is preserved on the left side of the skull. The lacrimal is large and contributes ventrally to the anterior root of the zygomatic process (Figures 1, 2a). Its tapered dorsal margin extends farther dorsally and is longer dorsally in FMNH P14424 than in Holmesina and Kraglievichia. The lac­ rimal foramen is quite small in FMNH P14424, as is that of Kraglievichia. Dorsal to the foramen, the lacrimal bears a raised, rugose, and nearly circular protuberance in FMNH P14424. The homologous surface in Holmesina is somewhat rugose, but a protuberance is not developed. The ventral mar­ gin of the lacrimal extends posteroventrally across the lateral surface of the anterior zygomatic root, as occurs in the dasypo- dine Dasypus (ROM R2019). The premaxillo-maxillary suture is not visible, although vari­ ous cracks are present anteroventrally on the palate (Figure 2b). Presumably it ran between the alveoli of M1 and M2 and per­ haps contributed to the mesial part of the M2 alveolus. Edmund (1985b) stated that this was the general condition in pam­ patheres, and he rejected the supposed presence (as by Winge, 1915; Castellanos, 1927) of Ml and M2 in the premaxilla of Pampatherium. The maxillae form most of the hard palate (Figure 2b). Right and left Ml alveoli are very close together, so that their lingual walls are separated by about 2 mm and largely cover the medial processes of the premaxillae. In Krag­ lievichia, Holmesina, and Pampatherium the Mis are farther apart. The tooth rows diverge posteriorly to the distal walls of the alveoli of the M6s, which are separated by approximately 28 mm; between the M7s and M9s they are then nearly parallel or converge very slightly posteriorly. The maxilla, zygomatic, and lacrimal meet lateral to the in­ fraorbital canal and the anterior surface of the zygomatic root. The maxillo-zygomatic suture extends nearly ventrally to the masseteric tuberosity (Figure 1). Medially these bones meet along an almost vertical suture at about the level of the deepest part of the orbit, but ventrally this turns anteroventrally onto the tuberosity. The course of the suture on the tuberosity is un­ clear, but it probably resembles that of Dasypus, in which both bones contribute to the tuberosity. The maxilla, lacrimal, and frontal meet at the large ethmoid foramen, about 17 mm posterior to the lacrimal foramen. The maxillo-frontal suture, well preserved in FMNH P14424, undu­ lates gently posteriorly. The maxilla has a large lateral expo­ sure. It extends posteriorly to the level of the sphenopalatine foramen, where it meets the small orbitosphenoid, and then turns abruptly ventrally into the margin of the foramen. This re­ lationship is odd, yet the sutures are sufficiently distinct. Usu­ ally, as in Dasypus and Choloepus, the frontal contacts the an­ terior margin of the orbitosphenoid, which has a broader lateral exposure and contains various foramina, such as the optic canal and orbital fissure. The palatine meets the ventral margin of the orbitosphenoid and forms the posterior and ventral margins of the sphenopalatine foramen; the maxilla forms its anterior mar­ gin. This usual condition is indeed present in Holmesina (ROM 3881); ventral to the postorbital process the fronto-maxillary suture descends posteroventral ly into the anterodorsal margin of the sphenopalatine foramen. The condition in FMNH P14424, if correctly interpreted, is therefore not characteristic of pampatheres and may be unique to this taxon or perhaps to this individual. Confirmation of this pattern by additional spec­ imens and detailed study of the region in Kraglievichia (MACN 2617) are desirable. In FMNH P14424 the orbitosphenoid is small and apparently consists of an anterodorsal and a posteroventral wing. The former has a short suture with the frontal. In lateral view, the orbitosphenoid is mostly covered by the curved, crest-like, an­ terior margin of the alisphenoid, which ends ventrally at the sphenopalatine foramen. The surface medial to the margin of the alisphenoid is depressed and contains various foramina, but their morphology is partly obscured by matrix. Some of the matrix has been removed, but the osseous partitions are deli­ cate and preclude further preparation. A dorsal, elongated groove almost surely represents the optic canal. Its dorsal margin is formed by the margin of the alisphe­ noid posteriorly and a crest of the frontal (which is confluent with that of the alisphenoid) anteriorly. Ventrally the margin of the optic canal is formed by a nearly horizontal and raised ridge that nearly coincides with the suture between the maxilla and the anterodorsal wing of the orbitosphenoid. Ventral to this ridge the depressed surface bears at least two passages: a large, apparently single, canal leading posteromedi­ ally and the smaller sphenopalatine foramen, leading ventro- medially. The former possibly represents the confluent orbital fissure and foramen rotundum. This morphology is apparently the same as in other pampatheres. The sphenopalatine foramen is single in FMNH P14424, but on the left side of Holmesina (ROM 3881) a second smaller foramen lies dorsal to the main foramen and becomes confluent with it. The sphenopalatine is thus closely associated with the orbital foramen in pam­ patheres, whereas it is farther anterior in dasypodines. The lateral suture between the maxilla and palatine is un­ clear, but it may be represented by a weak ridge between the ventral margin of the sphenopalatine foramen and a fine groove lying ventrally and some 5 mm posteriorly to the distal margin of the M9 alveolus. A second fine groove 3^4 mm posterior to the first may represent the palatopterygoid suture. On the pal­ ate, the suture between the palatine and maxilla is distinct only anteriorly. The suture strongly resembles that of Kraglievichia and is unlike that in Holmesina and Dasypus, in which the su- 56 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY ture curves gradually posterolaterally from the midline. In FMNH P14424 and Kraglievichia the suture extends anterolat- erally from the ventral midline, approximately between the dis­ tal and mesial ends of M8. The suture then turns abruptly pos­ teriorly and lies very near the margins of the alveoli. At the ventral midline the palatines and posterior parts of the maxillae form a median crest between the middle of M6 and the posterior margin of the hard palate in FMNH P14424. The crest is most prominent on the maxilla, approximately between the distal one-half of M6 and mesial one-half of M7. In Holme­ sina the crest is shorter and less prominent. It arises anteriorly from the maxillae at about the middle of M7 as a very weak ridge, is most prominent posteriorly, and ends about 15 mm an­ terior to the margin of the hard palate. The condition in Kra­ glievichia cannot be clearly distinguished from Castellanos's (1937, fig. 7) illustration, but it apparently resembles more closely that of Holmesina. In FMNH P14424 the mastoid has a wide occipital exposure. It is approximately Y-shaped, with the mastoid foramen lying at the knee of the Y. Its medial branch rises more dorsally than the lateral, and the stem of the Y is oriented ventrolaterally in occipital view. The mastoid in Holmesina is also Y-shaped, but the lateral arm is higher and the stem curves markedly ventro- medially. The paroccipital process is more robust in Holmesina, partic­ ularly ventrally. That of FMNH P14424 extends anteroven­ trally at about the same angle as in Holmesina but tapers ven­ trally to a blunt process (the right process is distorted so that its ventral end is displaced anterodorsally). In Holmesina the pa­ roccipital process is expanded and elongated, with its long axis oriented anteromedially. The paroccipital process of FMNH P14124 lies adjacent to the lateral margin of the occipital condyle. It projects ventrally a short distance past the levels of the condyle and mastoid pro­ cess as a blunt process (Patterson et al., 1989). A deep, narrow cleft lies between the paroccipital and mastoid processes. In Holmesina the paroccipital process extends laterally, rather than ventrally, at about the level of the dorsal one-half of the occipital condyle. It contacts the root of the mastoid process, thereby enclosing a short, anteroposteriorly oriented canal that lies a short distance ventral to the mastoid foramen on the oc­ ciput. The occipital condyles project slightly more ventrally and posteriorly relative to the mastoid process in FMNH P14424 than in Holmesina and Pampatherium. In these genera, how­ ever, the basicranium lies higher relatively to the pterygoid processes and alveolar margins. In FMNH P14424 the condyles lie slightly above the plane of the alveolar margins. In Kraglievichia they are apparently even lower. The pterygoid bone cannot be identified with confidence in this specimen and may have been lost postmortem, as is common in other pam­ pathere specimens. The basicranium is relatively short and wide in Holmesina and Pampatherium but is relatively elongated in FMNH P14424 and Kraglievichia. The auditory region of FMNH p 14424 was described by Patterson et al. (1989). Ventrally the pterygoid process is expanded and rugose and bears three or four prominent ridges. This condition also occurs in Holmesina and Pampatherium, but the processes apparently considerably less expanded and rugose. The position of the posterior end of the hard palate differs in these taxa. In Kraglievichia it ends slightly posterior to M9; in FMNH P14424, slightly farther posteriorly, approximately midway between the distal surface of M9 and the posterior end of the pterygoid process; in Hol­ mesina, about two-thirds of the way between these structures; and in Pampatherium, near the posterior part of the pterygoid process. Mandible; The mandibles are of similar form in all known pampatheres (Figure 3). FMNH P14424 includes a right den­ tary, missing only its anterior portion and m3-m5. Measure­ ments of the lower dentition and dentary are presented in Ta­ bles 2 and 3, respectively. The angular region is expanded and bears near its margin various tuberosities laterally and ridges medially for insertion of the masseteric musculature. The den­ tary of FMNH P14424 differs from that of Holmesina (ROM 4954, 4955, 4956, AMNH 26856), Pampatherium (Bordas, 1939, fig. 2; Winge, 1915, pis. 9, 10), and Kraglievichia (MACN 8968) in that the horizontal ramus is less elongated relative to the ascending ramus and angular region, so that it appears short and robust (Figure 3a). The horizontal ramus is relatively deep, however, and the ventral margin below the symphysis inclines more abruptly. The masseteric fossa is barely indicated in FMNH P14424, in contrast to the small but distinct fossa in Holmesina and Pampatherium. The medial surface of the dentary differs in minor ways from that described by Castellanos (1946). For example, the scar line for the mylohyoid lies somewhat farther dorsally and is more prominent in FMNH P14424. Also, the various depressed re­ gions dorsal and ventral to it (e.g., fovea submaxillaris, fovea sublingualis) are apparently deeper and their surfaces are more heavily scarred. It is probable, however, that these features are not taxonomically significant. These features vary considerably in two dentaries of subequal size (ROM 4954, 4955) of H. oc- cidentalis from a single locality. The ascending ramus rises steeply, as indicated by James (1957). A prominent, curved crest lies on the dorsolateral sur­ face of the coronoid process (Figure 3a). The crest is highest posteriorly and diminishes gradually as it extends anteroven­ trally to become confluent with the anterior edge of the pro­ cess. The crest possibly marks the separation between the in­ sertions of the temporal and masseteric musculatures, although Smith and Redford (1990) indicated that a similar crest in Eu- phractus sexcinctus Linnaeus is associated only with the tem­ poral. The anterior margin of the ascending ramus begins to rise just lateral to the mesial part of M8, although this tooth is sometimes entirely visible (e.g., K. paranense, MACN 8968). It slopes rapidly dorsally in a regular curve and in lateral view NUMBER 93 57 FIGURE 3.—Right dentary of Vassallia maxima FMNH P14424: a, lateral view; b, occlusal view. (Scale bar=50 mm.) covers most of the distal one-half of M8. This morphology is present in all pampatheres. Castellanos (1927, 1937) claimed, even though M9 is broken at the alveolar margin, that in V. minuta the anterior edge of the ramus lies somewhat posteri­ orly farther and therefore left part of this tooth uncovered. This is apparently incorrect, however. Although it is not clear from Castellanos's illustrations (1927, fig. 5; 1937, fig. 5), a cast (ROM 29580) of its type, MLP 29-IV-15-4, indicates that the position of the ascending ramus is precisely as in other pam­ patheres. It is evident from the position of M9 that its exposed part would have to have been extraordinarily high for the ra­ mus to have left part of it uncovered. Dentition: Only three upper teeth are preserved in situ: the right M6, nearly complete; the left M8, broken at the alveolar margin; and M9, broken below the alveolar margin. The shapes of the teeth in section therefore are interpreted largely from the form of the alveoli (Figure 2b). M1-M3 are not lobate. Ml is nearly circular; M2 and M3 are slightly and successively more elliptical. M4 is reniform, with a relatively weak labial sulcus. M5 is strongly bilobate, with a deep labial sulcus. Its lingual margin is not regularly curved but undulates slightly owing to a weak anterior groove, which is inferred from the correspondingly weak ridge of the alveolar margin. M6 and M7 are also strongly bilobate. Their lingual margins bear two weak grooves, and the alveoli bear two weak ridges. M8 and M9 are similar, but the mesial and distal lobes are relatively narrower transversely. M5 is the largest tooth, judging from its alveolus, but M6 was probably almost as large, followed in size by M7 and M8. M9 and M4 are approx­ imately equal in size, followed by M3, M2, and Ml. The upper teeth of other pampatheres differ in various de­ tails. M2 is subreniform in Holmesina and probably in Pam­ patherium, with the long axis oriented markedly obliquely. M3 is elliptical, and M4 is subreniform and longer than M2. In M2 TABLE 3.—Dentary measurements (in mm) for FMNH P14424. Dentary feature Ramus height at m5 Ramus height at m6 Ramus height at m7 Ramus height at m8 Ramus height at m9 Height coronoid process to angular process Height coronoid process to ventral margin of dentary Height condyle to angular process Height condyle to ventral margin of dentary Condyle, transverse length Condyle, anteroposterior width Symphysis length Symphysis thickness Measurement 37.0 43.3 47.5 48.0 50.0 80.0 132.0 47.0 96.4 31.2 11.0 54.0 16.0 58 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY and M4, however, the sulcus lies on the lingual surface, rather than on the labial as in FMNH P14424. In Kraglievichia, M1-M4 are nearly elliptical, with the long axis of M2 oriented nearly as in Holmesina. M5-M9 in these other pampatheres are similar to those of FMNH P14424. The dentary of FMNH P14424 preserves m2 and m6-m9, the alveoli for m3-m5, and the distal one-half of the ml alveo­ lus (Figure 3b). The ml alveolus is insufficiently preserved to permit recognition of the shape of the tooth; however, it is as­ sumed to be circular in section, as indicated by Castellanos (1927, 1937). Tooth m2 is elliptical, with the long axis oriented obliquely, and its distolingual surface is flattened, as described by Castellanos (1946) for V maxima. The alveolus of m3 is el­ liptical. This is in contrast to the shape of the right m3 de­ scribed by Castellanos (1946, figs. 5, 6, 7b), in which the disto­ lingual surface is concave and the distal part of the tooth is keeled in section. It is worth noting, however, that the pre­ served lingual one-half of the left alveolus of Castellanos's specimen (IFG 500) does not indicate a concave surface. In FMNH P14424, m4 is elongated and reniform or possibly slightly bilobate, with weak sulci present on the lingual and la­ bial surfaces. Castellanos (1946) described and figured the cor­ responding tooth in IFG 500 as prominently keeled. The pre­ served one-half of the left alveolus of m4 in Castellanos's specimen suggests, however, that the lingual surface was prob­ ably semi-elliptical. The differences between left and right teeth in IFG 500 suggest that the features described by Castell­ anos (1946) for the right m3 and m4 reflect individual varia­ tion. m5 of FMNH P14424 is strongly bilobate, with a deep la­ bial sulcus; m6-m8 are also markedly bilobate, each with a deep and flattened labial sulcus; m9 is weakly bilobate, with mesial and distal lobes subequal in width. The alveolus of m5 is slightly larger than that of m6, which is followed in size by m7, m8, m9, m4, m3, m2, and presumably ml . In size and form of the teeth, FMNH P14424 essentially follows Kraglievi­ chia and Pampatherium (Castellanos, 1927, 1937). It differs from Holmesina (Simpson, 1930) in that m4 in the latter is reniform, with a slightly angular distal surface, and from V. minuta (Castellanos, 1927, 1937) in that m4 is subelliptical. Hoffstetter (1958) characterized the teeth of cingulates as composed of a small central portion of osteodentine sur­ rounded by compact dentine, which forms most of the tooth, and finally by a layer of cementum. He indicated that in some lineages, most notably glyptodonts, the osteodentine may be harder and structurally resembles vasodentine. This condition is present in the three molariform teeth (M6, m7, m8) of FMNH P14424 for which the occlusal surfaces are preserved. The central osteodentine is a narrow, elongate strip that widens slightly mesially and distally. It is raised in relief (as is the pe­ riphery of the tooth), is convex lingually, and strongly resem­ bles the form illustrated by Hoffstetter (1958, fig. 22b) for ml and m2 of the glyptodont Eucinepeltus petesatus Ameghino. The osteodentine is not well preserved in Holmesina septentri- onalis (Edmund, 1985a, figs. 5, 6) or H. occidentalis (ROM 3881, 4964), suggesting that it was considerably softer. The central portion of the tooth tends to split longitudinally (see Ed­ mund, 1985a, fig. 7), creating a cavity (presumably filled in life by osteodentine) that often divides the tooth in half. In some teeth, single or multiple fractures radiate from the mesial or distal end of the cavity toward the periphery of the tooth. In Kraglievichia paranense (MACN 8968, 8943) and Pampathe­ rium typum Ameghino (MLP 81-X-30-1) osteodentine is present and stands in relief. Cervical Vertebrae: Perhaps the most peculiar part of FMNH P14424 is the axis, which may be either separate or fused with additional elements (Figure 4). The form of this ele­ ment is unexpected in a pampathere or dasypodid. Although possibly it is not part of the individual assigned to FMNH P14424, the state of its preservation and the nature of the ad­ herent matrix are identical with those of the remaining suite of elements constituting FMNH P14424. Furthermore, we are un­ able to reconcile its morphology with that of any known mam­ mal. It is generally known that in dasypodids and other pam­ patheres the axis is fused to two or three succeeding cervical vertebrae to form a cervical tube. Utaetus is the only dasypodid genus, to our knowledge, in which the axis is free (Simpson, 1948; Hoffstetter, 1958). In Holmesina, for example, the tube includes the axis and cervical vertebrae 3-5, with intervertebral foramina clearly delimiting each vertebra. The length of the tube clearly reflects the presence of multiple elements. In dor­ sal view, it is constricted posterior to the lateral facets for the atlas, forming what may be termed a neck. The slender, trans­ verse processes of cervical vertebrae 3-5, fused lateral to the intervertebral foramina, extend lateroventrally and end in ex­ panded, somewhat rugose tuberosities. The tube in dasypo- dines (e.g., Dasypus, ROM R948) is extremely similar. The element in FMNH P14424 clearly includes the axis, with its high, overhanging neural spine, dens, and three anterior facets for the atlas (Figure 4a). These facets are smaller than those in Holmesina, and the dens is relatively shorter. The ele­ ment is extremely large and robust for the size of the skull and is strongly compressed anteroposteriorly (Figure 4b). It is ap­ proximately as long, posterior to the facets, as the axis in the cervical tube of Holmesina ROM 3856. Two very long and ro­ bust processes extend laterally on either side immediately pos­ terior to the articular facets, i.e., the position of the neck. Inter­ vertebral foramina are absent from the dorsolateral surface of these processes, in contrast to the condition in ROM 3856, but one large foramen and two very small foramina anterior to it open from the ventral surface of the process; two large, sub- equal foramina lie ventrally in ROM 3856. This element possibly represents the axis only. In dasypodids and other pampatheres, however, the axis, although fused to other cervical vertebrae, does not bear transverse processes,, a condition that also occurs in the free axis of Pilosa (sloths and vermilinguas). It more closely resembles the fused axis and cervical vertebrae 3-5 of the glyptodont Propalaeohoplopho- NUMBER 93 59 FIGURE 4.—Axis (or axis complex) of Vassallia maxima FMNH P14424: a, anterior view; b, lateral view. (Scale bar=50 mm.) rus australis Ameghino (Scott, 1903, pl. 25: fig. 1) in the shape, compression, and relationship of the transverse pro­ cesses to the articular facets. In this species, however, the fac­ ets are oriented differently and three intervertebral foramina identify the cervicals. The element in FMNH P14424 differs considerably from the tube of Glyptotherium arizonae Gidley (Gillette and Ray, 1981, fig. 22). The transverse processes of FMNH P14424 resemble those of the free axis of Manis penta- dactyla Linnaeus (ROM R589), although they are less ex­ panded laterally in the latter. Manus: FMNH P14424 preserves the R unciform, L and R Ph 1 of D II, L Ph 2 of D III, L Ph 1 of D IV, and L Ph 1 of D V. The elements were compared with casts (ROM 5000) of those of H. septentrionalis, UF 9336. Those of FMNH P14424 are smaller, but they differ mostly in minor ways from those of Holmesina. The unciform is missing a small mediopalmar por­ tion. It is relatively deeper proximodistally than that of Holme­ sina. Ph 1 of D II differs markedly. It is considerably shorter prox- imomedially, and the articular surface for Mc II is flatter. In Holmesina, this surface is extended by mediolateral and medio­ palmar articular shelves, probably for sesamoids, which are separated by a deep notch. The shelves are rudimentary, and the notch is replaced by a sulcus in FMNH P14424. Distally the articular condyles in FMNH P14424 are reduced and flat­ tened, and the median palmar protuberance of the articular sur­ face is more distal relative to the atrophied condyles. The artic­ ular surfaces of Ph 1 suggest relatively restricted movement. In proximal view the articular surface of Ph 2 of D III has a less excavated palmar margin because the palmomedian sulcus is less prominent; however, the dorsal margin is more convex, and its median portion projects farther distally than in Holme­ sina. Ph 1 of D IV closely resembles that of Holmesina, but its proximal and distal articular surfaces have less relief. Ph 1 of D V is relatively elongated and thus more gracile. Its shaft is less triangular in section, and its dorsal surface is more rounded. In proximal view, Ph 1 of Holmesina is distinctly triangular, whereas that of FMNH P14424 has two short and two long margins. Pes: Nearly all of the pes is represented in FMNH P14424, including L astragalus, calcaneum, navicular, cuboid, entocu- neiform, mesocuneiform, and ectocuneiform; L Mt I, L and R Mt II, R Mt III, L Mt IV; L and R proximal phalanges, R Ph 2 and ungual phalanx of digit III, L ungual phalanx of digit IV, L and R phalanges 1 and 2, and R ungual phalanx of digit V; L plantar sesamoid. As with the manus, the pedal elements are smaller than those of H. septentrionalis (compared with casts, ROM 5000, of UF 9336). The neck of the astragalus is not as well defined as that of Holmesina, and the neck and the facet for the navicular taper more strongly anteriorly. The trochlea tali is relatively larger because its anterior margin extends farther distovolarly. Also, this margin is not as prominently notched. The calcaneum lacks the tuber. The facet for the cuboid is relatively larger and wider, but in other respects the calcaneum strongly resembles that of Holmesina. The navicular differs mainly in that the bony rug­ osities are less prominent and the facet for the cuboid is elon­ gate compared with that of Holmesina. The cuboid is more ro­ bust because it is relatively compressed dorsovolarly. Its calcaneal facet is wider dorsovolarly, and its lateral portion ap­ proaches the volar sesamoid facet much more closely than in Holmesina. The cuneiforms are very similar to those in Holme­ sina; however, the proximal and distal surfaces are subequal in dorsovolar length, whereas in Holmesina the distal surface is elongated. Mt I is relatively compressed dorsovolarly, and the keel of the distal trochlear surface is poorly defined. Mt I of Holmesina is relatively expanded mediovolarly. Mt II and Mt III are more 60 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY T* >'*"?£ ?3Jft •C%j*Lfr&k'* >?7Tv' %$*r --•.vii FIGURE 5.—Dorsal views of osteoderms of Vassallia maxima FMNH P14424: a-c, osteoderms probably from the anterior buckler; d,e, osteoderms from the margin of the buckler;/ osteoderm probably from a limb; g, imbri­ cating osteoderm. (Scale bar=50 mm.) gracile, and their shafts are relatively narrower mediolaterally just distal to the proximal articular surfaces. This also occurs in Mt IV, and the median keel of the distal trochlear surface is nearly semicircular, whereas in Holmesina it is truncated. Ph 1 of D 1 is relatively elongate proximodistally. Ph 2 of D III is nearly as long as that of Holmesina but is markedly com­ pressed mediolaterally. The ungual phalanx of D III is rela­ tively elongate and less spatulate than that of Holmesina. The ungual phalanx of D IV is relatively elongated, less spatulate, and less dorsovolarly flattened. The bony projections on the pe­ rimeter of the proximal articular surfaces are markedly less prominent, particularly laterally. Ph 1 of D V is relatively com­ pressed mediolaterally, and its distal articular surface has less relief. Ph 2 of D V is more robust, apparently because it is rela­ tively thicker dorsovolarly. The bony core of the ungual is sim­ ilar in shape, but there is greater development of bony projec­ tions posterovolarly and laterally in Holmesina. The plantar sesamoid bears two articular facets dorsally, which meet along a straight line at an angle of approximately 120° One facet, probably for the cuboid, is elongate, hexago­ nal, and nearly flat. The other is somewhat larger and articu­ lated with the navicular; it is partially concave, but it becomes convex on two of its edges. The remainder of the bone is rug­ ose and bears a prominent almond-shaped protuberance on the side nearest the cuboid facet. In Holmesina this swelling is less prominent and is nearly centered on the rugose surface. Osteoderms: The suite of osteoderms associated with the skeletal and dental material make FMNH P14424 a pivotal specimen in the study of late Miocene and early Pliocene pam­ patheres, especially those assigned to the genera Vassallia (Fig­ ure 5), Plaina, and Kraglievichia, and to the Pleistocene genus Pampatherium. The feeble ornamentation of the osteoderms of Vassallia dis­ tinguishes the genus from Kraglievichia, although their size ranges overlap (Table 4). Some isolated buckler osteoderms as­ signed to Kraglievichia, however, have an ornamentation not unlike that of some in the series of FMNH P14424, raising doubts regarding the identification of isolated osteoderms. Plaina intermedia (MACN A7799) and P. subintermedia (MACN 8473) are based upon inadequate osteodermal material that is similar in size and morphology to that of Vassallia sp., and their generic validity is questionable, as suggested by Ed­ mund in Patterson et al. (1989). The material assigned to Plaina brocherense (IFG 769; Castellanos, 1958) is larger than that of any species of Vassallia so far described. The ornamen­ tation of the osteoderms is not unlike that of FMNH P14424, and other series referred to Vassallia, except that it is larger and the buckler osteoderms have wide margins made up of large follicular pits. Vassallia minuta, as typified by MLP 69-VIII- 25-7, strongly resembles V. maxima morphologically but is smaller (Table 4). NUMBER 93 61 TABLE 4.—Length, width, and thickness of non-marginal buckler osteoderms of pampatheres. (Mx Mn = minimum; Me = mean.) Species Length Width Thickness Specimen Mx Mn Me Mx Mn Me Mx Mn Me Vassallia maxima V. maxima V. maxima V. maxima V. maxima V. minuta Plaina subintermedia P. intermedia P. intermedia P. brocherense Pampatherium typum P. typum P. typum P. typum Kraglievichia paranense FMNH P14424 IFG 502-515' PVL 32052 MLP29-X-10-123 MLP52-IX-30-2I4 MLP69-XII-26-175 MACN 84736 MACN A-77997 MLP69-VIII-25-18 IFG 7699 IFG HO10 MLP69-VIII-22-1111 MMP216S12 MLP69-VIII-25-713 MACN 1 4 29 29 45 33 49 39 43 31 33 31 36 22 34 22 - 40 43 37 50 44 51 30 49 28 37 32 39 39 40 37 39 28 30 40 47 39 36 36 40 34 23 46 23 36 26 36 30 24 20 26 12 - 26 39 28 29 34 28 43 34 35 27 43 44 37 40 23 39 32 25 35 26 34 31 33 30 33 22 20 33 38 30 28 29 33 32 17 24 7 8.7 55 6.1 4 6.2 6.3 6.4 4.8 6.7 7.3 8.5 8.1 6.7 7.3 7.1 6.6 'Series including type of Vassallia maxima in Castellanos (1946). identified as V. maxima by Scillato-Yane (1982); Andalhuala Formation, Caspinchango, Valle de Santa Maria, Tucuman Province, Huayquerian. identified as V. maxima by Scillato-Yane (1982); San Fernando, Catamarca Province. 4 V maxima according to museum label, Tiopunco, Valle de Santa Maria, Tucuman Province. 5 Identified as V. minuta by Scillato-Yane (1982); "Araucanense," Catamarca Province. 6Plaina subintermedia, holotype of Rovereto (1914); Montehermosan (Marshall and Patterson, 1981); we consider this morphologically inseparable from osteoderms of V maxima. 1 Plaina intermedia, labeled by Ameghino as Chlamydotherium intermedium; Monte Hermoso, Buenos Aires Province. 8Identified by Scillato-Yane (1982), P. subintermedia according to museum label; Monte Hermoso, Buenos Aires Province, no stratigraphic data. 9Plaina brocherense, type series of Castellanos (1958); Valle de los Reartes, Cordoba Province, Formacion Brocherense (Castellanos, 1958) cf. Huayquerian (Marshall et al., 1983). 10Pampatherium typum according to museum label; Rio Cararana, Santa Fe Province, referred to upper Pleis­ tocene by A.G. Edmund. "Identified by Scillato-Yane (1982); no stratigraphic or locality data. 12Pampatherium typum according to museum label; Ensenadan. 13 Identified as P. typum by Scillato-Yane (1982); no stratigraphic or locality data. 14Kraglievichia paranense, includes a series of MACN specimens (MACN 1637, 2639, 3951, 4056, 4076, 4693, 4700, 4758, 5053, 8944) according to Scillato-Yane, and MACN 4858, 12278 according to museum label; all are from Entre Rios, Parana Province. Vassallia is generally believed to be the precursor of Pam­ patherium, and indeed it has essentially identical morphology but is larger in size. Usually members of the the genus Vassal­ lia, especially V. maxima, have been confined to the late Mi­ ocene and much of the Pliocene. Some suites of osteoderms supposedly of Pleistocene age, however, have been assigned to P. typum and have typical Vassallia morphology. Their sizes are equal to, or sometimes smaller than, those of V. maxima, the latter including the type specimen and a series of osteoderms confidently identified as being of the Huayquerian age (Table 4). Vassallia maxima and smaller individuals of P. typum may therefore be confused, at least on the basis of osteoderms. Osteoderms of the type and referred specimens of V. maxima (e.g., IFG 505-528; Castellanos, 1946) agree well with those of FMNH P14424, except that the latter are somewhat thicker (Table 4). This may be an age-related variation. A typical non- marginal buckler osteoderm of P14424 has most of the hair fol­ licles confined to the anterior end, in a band up to 5 mm wide. In some specimens the follicles continue down the sides in a very narrow margin. The marginal band (Edmund 1985b) is ap­ proximately 5-8 mm wide and is smooth and slightly rounded. It is separated from the central figure by a sulcus usually less than 1 mm deep and approximately 5-8 mm wide. The sulcus is usually horseshoe shaped, with the central figure reaching the submarginal band posteriorly. The central figure is highly variable, usually a flat, rounded, almost impalpable boss, but in some specimens it is slightly raised, tapering posteriorly to join the submarginal band. Some osteoderms of FMNH P14424 lack a sulcus or central figure and instead have a distinct central depression approxi­ mately 2 mm deep and 10 mm in diameter. In others the sulcus is represented by two narrow furrows that are subparallel to the central figure and are barely united at the anterior end. Within the range of variability of the osteoderms of FMNH P14424, 62 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY however, one can find examples of the surface morphology given in the literature for all species of Plaina and Vassallia. With regard to the illustrations of the type description of V. maxima (Castellanos, 1946), one should be cautious of the half-tone illustrations. Castellanos, who had impaired vision, highlighted the central figures with a soft pencil. In fact, these figures are barely palpable, and the markings were intended only for Castellanos's personal reference. In the sample of osteoderms of FMNH P14424, there are nu­ merous specialized osteoderms from the nuchal notch, the mar­ gins of the buckler, and the rows adjacent to the imbricating bands. These are similar to those illustrated by Holmes and Simpson (1931) and by Edmund (1985a), except that they bear little surface ornamentation. In the large sample represented by P14424, one would expect to find numerous rectangular buckler osteoderms, representing the posterior buckler. Instead, there are only a few, suggesting that this area had been lost either before becoming fossilized or by subsequent erosion. No part of the skeleton underlying this area was recovered. Most suites of randomly collected osteo­ derms of Vassallia sp. or Pampatherium sp. contain approxi­ mately 50% rectangular buckler osteoderms (Edmund, 1996). Specimen P14424 includes 176 nearly complete osteoderms, distributed as follows: Non-marginal buckler 68 Posterior margin of pectoral buckler 15 Anterior margin of pelvic buckler 5 Nuchal notch 6 Lateral margins 6 Buckler, imbricating 20 Tail and possibly leg, imbricating 36 Small, non-imbricating 20 Discussion and Conclusions FMNH P14424 is from the Puerta del Corral Quemado, Cat­ amarca Province, but the field notes are sufficiently vague that its age may be Huayquerian or Montehermosan. The type spec­ imen of Vassallia maxima was recovered from the Valle de Santa Maria, Tucuman Province. Castellanos (1946) consid­ ered these strata as middle and upper Araucanian; Marshall et al. (1983) suggested that they are probably of Huayquerian age. These localities are reasonably close to each other, and both are well removed from deposits typically yielding remains of Krag-lievichia, Pampatherium, and Holmesina. Although the localities are not identical, their proximity supports the possi­ bility that FMNH P14424 belongs to V maxima. The morphological and mensural differences between the left dentary of FMNH P14424 and the partial right dentary of the type of V. maxima (IFG 500) are few and minor. The osteo­ derms of FMNH P14424 agree well morphologically and men- surally with those of the type series of V. maxima. Morphologi­ cally, the osteoderms also strongly resemble those of the Huayquerian (late Miocene) V. minuta and the Pleistocene Pampatherium. On the basis of comparable morphological and, apparently, stratigraphic criteria, there is no obvious justifica­ tion for separating FMNH P14424 from V maxima; we there­ fore assign it to this species. The only identifiable vertebral material of FMNH P14424 is the axis or axis complex. Although it was apparently associated in the field with the remaining and typically pampathere mate­ rial of FMNH P14424, its morphology renders an association questionable. The extensive suite of osteoderms of FMNH P14424 demon­ strates a moderate degree of variation, suggesting that the iden­ tification of material, and more importantly the erection of taxa, on the basis of small series of isolated osteoderms is es­ sentially futile. The range of variation in FMNH P14424 also encompasses the descriptions of Plaina intermedia and P. sub- intermedia, and indeed of some suites identified as Pampathe­ rium typum. This suggests that the validity of P. intermedia and P. subintermedia, which were described from inadequate mate­ rial, is probably unjustified, and that the genus Plaina should perhaps be considered a synonym of Vassallia. The size and morphology of the type specimens for these species fall within the range of the osteoderms of FMNH P14424, suggesting that P. intermedia, P. subintermedia, and V maxima are conspecific. That this is probably the case for the species of Plaina has al- redy been suggested by Paula Couto (1979), who indicated that Plaina was a monotypic genus with P. intermedia being the valid name. Although direct comparison between FMNH P14424 and the Bolivian material assigned to P. intermedia by Anaya and MacFadden (1979) was not possible, the resem­ blance of the remains illustrated and described by those authors to FMNH P14424 seems sufficiently close to support a hypoth­ esis of conspecifity of P. intermedia and V. maxima. If this is demonstrated to be true, then the epithet intermedia is the earli­ est available name. Its type is apparently lost, and the descrip­ tion is vague but is probably sufficient for its time of publica­ tion. The valid name for the species would thus be V. intermedia (Ameghino 1888), and the epithets subintermedia and maxima would fall as junior synonyms. We do not propose this formally at this time, however, because of the paucity of data for the two earlier species. The thickness of the osteo­ derms of P. brocherense suggests that the latter may be a dis­ tinct species, but the morphology of the osteoderms suggests that it also may possibly be referred to Vassallia. NUMBER 93 63 Literature Cited Ameghino, F. 1883. Sobre una coleccion de mamiferos fdsiles del piso mesopotamico de la formacion patagonica recogidos en las barrancas del Parana por el Prof. Pedro Scalabrini. Boletin de la Academia Nacional de Cien­ cias en Cdrdoba, 5:101-116. 1887. Apuntes preliminares sobre algunos mamiferos extinguidos del yacimiento de "Monte Hermoso" existentes en el "Museo de La Plata." Boletin del Museo La Plata, 1:1-20. 1888. Lista de las especies de mamiferos fosiles del Mioceno superior de Monte-Hermoso, hasta ahora conocidas. 21 pages. Buenos Aires: P.E. Coni. 1889. Contribucion al conocimiento de los mamiferos de la Republica Ar­ gentina. Adas de la Academia Nacional de Ciencias en Cdrdoba, 6:1-1027. 1895. Sur les edentes fossiles de 1'Argentine. Revista del Jardin Zoologico, 3(4):97-192. Anaya, F., and B.J. MacFadden 1995. Pliocene Mammals from Inchasi, Bolivia: The Endemic Fauna Just before the Great American Interchange. Bulletin of the Florida Mu­ seum of Natural History, 39(3):87-140. Bordas, A.F. 1939. Craneometria y region auditiva de Chlamytherium typum Ame­ ghino. Physis, 14:447-460. Cartelle, C , and G. Bohorquez 1985. Pampatherium paulacoutoi, una nova especie de tatu gigante da Ba- hia, Brasil (Edentata, Dasypodidae). Revista Brasileira de Zoologia, 2:229-254. Cartelle, C.A., B.G. Camara, and P.I.L. de Prado 1991. Estudo comparativo dos esqueletos da mao e pe de Pampatherium humboldti (Lund 1839) e Holmesina paulacoutoi (Cartelle & Bo­ horquez 1985); Edentata Pampatheriinae. Anais do XI Congresso Brasileiro de Paleontologia, 2:621-634. Castellanos, A. 1927. Breves notas sobre los clamidoterios. Publicacion del Centro de Estudiantes de Ingenieria de Rosario, 8 pages. 1937. Anotaciones sobre las lfneas filogeneticas de los clamiterios. Publi- caciones del Instituto de Fisiografia y Geologia, Universidad Na­ cional del Litoral (Rosario, Argentina), Serie Tecnico-Cientifica, 8: 35 pages. 1946. Una nueva especie de clamiterio, Vassallia maxima n. sp. Publica- ciones del Instituto de Fisiografia y Geologia, Universidad Nacio­ nal del Litoral (Rosario, Argentina), Serie Tecnico-Cientifica, 26: 47 pages. 1958. Nota preliminar sobre nuevos restos de mamiferos fosiles en el Bro­ cherense de Valle de los Reartes (Provincia de Cordoba, Argentina). XX Congreso Geologico Internacional, Mexico, section 7, pages 217-233, figures 1-7. De Iuliis, G., M.S. Bargo, and S.F. Vizcaino 2000. Variation in Skull Morphology and Mastication in the Fossil Giant Armadillos Pampatherium spp. and Allied Genera (Mammalia: Xe- nartha: Pampatheriidae), with Comments in Their Systematics and Distribution. Journal of Vertebrate Paleontology, 20(4):743-754. Downing, K.F., and R.S. White 1995. The Cingulates (Xenarthra) of the Leisey Shell Pit Local Fauna (Irv­ ingtonian), Hillsborough County, Florida. Bulletin of the Florida Museum of Natural History, 37:375-396. Edmund, A.G. 1985a. The Fossil Giant Armadillos of North America (Pampatheriinae, Xenarthra=Edentata). In G.G. Montgomery, editor, The Evolution and Ecology of Armadillos, Sloths, and Vermilinguas, pages 83-94. Washington, D.C: Smithsonian Institution Press. 1985b. The Armor of Fossil Giant Armadillos (Pampatheriidae, Xenarthra, Mammalia). Pearce-Sellards Series, Texas Memorial Museum, 40: 20 pages. 1987. Evolution of the Genus Holmesina (Pampatheriidae, Mammalia) in Florida, with Remarks on Taxonomy and Distribution. Pearce-Sell­ ards Series, Texas Memorial Museum, 45: 20 pages. 1996. A Review of Pleistocene Giant Armadillos (Pampatheriidae, Xenar­ thra, Mammalia). In K.M. Stewart and K.L. Seymour, editors, Palaeoecology and Palaeoenvironments of Late Cenozoic Mam­ mals: Tributes to the Career ofC.S. (Rufus) Churcher, pages 300-321. Toronto: University of Toronto Press. Edmund, A.G., and J.M. Theodor 1997. A Giant New Pampatheriid Armadillo. In R.F. Kay, R.H. Madden, R.L. Cifelli, and J.J. Flynn, editors, Vertebrate Paleontology in the Neotropics: The Miocene Fauna of La Venta, Colombia, pages 227-232. Washington, D.C: Smithsonian Institution Press. Engelmann, G.F. 1985. The Phylogeny of the Xenarthra. In G.G. Montgomery, editor, The Evolution and Ecology of Armadillos, Sloths, and Vermilinguas, pages 51-64. Washington, D.C: Smithsonian Institution Press. Gillette, D.D., and C E . Ray 1981. Glyptodonts of North America. Smithsonian Contributions to Pale­ obiology, 40:1-255. Hoffstetter, R. 1958. Xenarthra. In J. Piveteau, editor, Trade de Paleontologie, 6:534- 636. Paris: Masson. Holmes, W.W., and G.G. Simpson 1931. Pleistocene Exploration and Fossil Edentates in Florida. Bulletin of the American Museum of Natural History, 59:383—418. Hulbert, R.C, Jr., and G.S. Morgan 1993. Quantitative and Qualitative Evolution in the Giant Armadillo Holmesina (Edentata: Pampatheriidae) in Florida. In R.A. Martin and A.D. Bamosky, editors, Morphological Change in Quaternary Mammals of North America, pages 134—177. New York: Columbia University Press. James, G.T. 1957. An Edentate from the Pleistocene of Texas. Journal of Paleontol­ ogy, 31:796-808. Kraglievich, L. 1934. La antiguedad pliocena de las faunas de Monte Hermoso y Chapad­ malal, deducidas de su comparacion con las que le precedieron y sucedieron. 133 pages. Montevideo: El Siglo Ilustrado. Lydekker, R. 1894. Contributions to a Knowledge of the Fossil Vertebrates of Argen­ tina, Part 11(2): The Extinct Edentates of Argentina. Anales del Mu­ seo de La Plata, Paleontologia Argentina, III: 118 pages. Marshall, L.G., A. Berta, R. Hoffstetter, R. Pascual, O.A. Reig, M. Bombin, and A. Mones 1984. Mammals and Stratigraphy: Geochronology of the Continental Mammal-Bearing Quaternary of South America. Palaeovertebrata (Montpellier), Memoire Extraordinaire, 1984: 76 pages. Marshall, L.G., R. Hoffstetter, and R. Pascual 1983. Mammals and Stratigraphy: Geochronology of the Continental Mammal-Bearing Tertiary of South America. Palaeovertebrata (Montpellier), Memoire Extraordinaire, 1983: 93 pages. Marshall, L.G., and B. Patterson 1981. Geology and Geochronology of the Mammal-Bearing Tertiary of the Valle de Santa Maria and Rio Corral Quemado, Catamarca Prov­ ince, Argentina. Fieldiana, new series (Geology), 9:1-80. Mones, A. 1986. Palaeovertebrata Sudamericana, Catalogo Sistematico de los Verte­ brados Fosiles de America del Sur, Parte I: Lista Preliminar y Bib- liografia. Courier Forschungsinstitut Senckenberg, 82:1-625. 64 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY Moreno, F.P., and A. Mercerat 1891. Paleontologia. Revista del Museo de La Plata, 1:222-236. Morgan, G.S., and R.C Hulbert, Jr. 1995. Overview of the Geology and Vertebrate Biochronology of the Lei­ sey Shell Pit Local Fauna, Hillsborough County, Florida. Bulletin of the Florida Museum of Natural History, 37:1-92. Pascual, R., and 0 . Odreman Rivas 1973. Las unidades estratigraficas del terciario portadores de mamiferos, su distribution y sus relaciones con las acontecimentos diastroficos. Adas del Quinto Congreso Geologico Argentino, 3:293-338. Pascual, R., E.J. Ortega Hinojosa, D. Gondar, and E.P. Tonni 1966. Vertebrata. In A.V. Borrello, editor, Paleontografia Bonaerense, 4 202 pages. La Plata: Comision de Investigaciones Cientificas. Patterson, B., and R. Pascual 1972. The Fossil Mammal Fauna of South America. In A. Keast, F.C. Erk, and B. Glass, editors, Evolution, Mammals, and Southern Conti­ nents, pages 247-309. Albany: State University of New York Press. Patterson, B., W. Segall, and W.D. Turnbull 1989. The Ear Region in Xenarthrans (= Edentata: Mammalia), Part I: Cin­ gulates. Fieldiana, new series (Geology), 18:1—46. Paula Couto, C. de 1954. Sobre um gliptodonte do Uruguai e um tatii fossil do Brasil. Notas Preliminares e Estudos, Divisao de Geologia e Mineralogia, 80: 1-10. 1979. Tratado de Paleomastozoologia. 590 pages. Rio de Janeiro: Aca­ demia Brasileira de Ciencias. Riggs, E.S., and B. Patterson 1939. Stratigraphy of Late Miocene and Pliocene Deposits of the Province of Catamarca (Argentina) with Notes on the Faunae. Physis, 14: 143-162. Robertson, J.S. 1976. Latest Pliocene Mammals from Haile 15A, Alachua County, Flor­ ida. Bulletin of the Florida State Museum, Biological Sciences, 20: 111-186. Rovereto, C. 1914. Los estratos araucanos y sus fosiles. Anales del Museo Nacional de Historia Natural (Buenos Aires), 25: 250 pages. Scillato-Yane, G.J. 1980. Catalogo de los Dasypodidae fosiles (Mammalia, Edentata) de la Republica Argentina. Adas del Segundo Congreso Argentino de Pa­ leontologia y Bioestratigrafia y Primero Congreso Latinoamericano de Paleontologia, Buenos Aires, 1978, 3:7-36. 1982. Los Dasypodidae (Mammalia, Edentata) del Plioceno y Pleistoceno de Argentina. 159 pages. Doctoral dissertation, Facultad de Ciencias Naturales y Museo, Universidad Nacional de La Plata. La Plata, Ar­ gentina. Scott, W.B. 1903. Mammalia of the Santa Cruz Beds, Part I: Edentata; Part II: Glyp- todontia and Gravigrada. Reports of the Princeton University Expe­ ditions to Patagonia, 1896-1899, Palaeontology, 5:107-227. Simpson, G.G. 1930. Holmesina septentrionalis. Extinct Giant Armadillo of Florida. American Museum Novitates, 442: 10 pages. 1948. The Beginning of the Age of Mammals in South America, Part 1: Introduction; Systematics: Marsupialia, Edentata, Condylarthra, Litopterna and Notioprogonia. Bulletin of the American Museum of Natural History, 91(1): 1-232. Smith, K.K., and K.H. Redford 1990. The Anatomy and Function of the Feeding Apparatus in Two Arma­ dillos (Dasypoda): Anatomy Is Not Destiny. Journal of Zoology (London), 222:27-47. Vizcaino, S.F., G. De Iuliis, and M.S. Bargo 1998. Skull Shape, Masticatory Apparatus, and Diet of Vassallia and Holmesina (Mammalia: Xenartha: Pampatheriidae): When Anat­ omy Constrains Destiny. Journal of Mammalian Evolution, 5(4): 291-322. Winge, H. 1915. Jordfunde og nulevende Gumlere (Edentata) fra Lagoa Santa, Minas Geraes, Brazilien. E Museo Lundii, 3(2): 321 pages, 42 plates. Additional Records of the Giant Beaver, Castoroides, from the Mid-South: Alabama, Tennessee, and South Carolina Paul W. Parmalee and Russell Wm. Graham ABSTRACT Four previously unreported records of Castoroides provide sup­ portive evidence that the giant beaver probably occurred through­ out the southeastern United States, especially along the middle stretch of the Tennessee River. A distal section of an upper right incisor and an incisor fragment of the extinct Pleistocene giant beaver, Castoroides, were recovered from Bell Cave, Colbert County, Alabama. Cave ACb-3, also in Colbert County and con­ taining an extensive deposit of late Pleistocene megafauna, yielded a single incisor enamel fragment. A fragment of a left ilium of this beaver was found in a dry stream bed in Ruby Falls Cave, Lookout Mountain, Hamilton County, Tennessee. These four specimens are referred to Castoroides sp. A relatively complete skull of Castoroides has been recovered from the Cooper River, near Strawberry Hill, Charleston County, South Carolina. The cranial characters of this specimen make it referable to Castoroides leiseyorum Morgan and White, 1995, which was described from the Irvingtonian Leisey Shell Pit, Hills­ borough County, Florida. The taxonomy of Castoroides from the southeastern United States is uncertain, and at least two different interpretations are possible. Introduction The extinct giant beaver, Castoroides Foster, 1838, was the largest rodent known in North America during the Pleistocene, reaching a length of about 2.5 m and a weight between 150 and 200 kg (Kurten and Anderson, 1980). The animal possessed huge convex incisors that in adult individuals extended 75 to 100 mm beyond the gum line. Longitudinal grooves and ridges on the exterior enamel make even small fragments diagnostic. Paul W. Parmalee, Frank H. McClung Museum, University of Tennes­ see, Knoxville, Tennessee 37996. Russell Wm. Graham, Earth Sci­ ences, Denver Museum of Natural History, 2001 Colorado Boulevard, Denver, Colorado 80205. The rounded and blunt tips of the incisors, along with certain features of the skull and relative proportions of postcranial ele­ ments, have led several researchers (e.g., Barbour, 1931; Stir­ ton, 1965) to conclude that Castoroides, unlike the modern beaver, Castor canadensis Kuhl 1820, would not have been ef­ fective at felling trees. Considered to have been an inhabitant of lakes and ponds bordered by swamps, the giant beaver was probably more similar in habits to the muskrat, Ondatra zibe- thicus (Linnaeus, 1766), than to C. canadensis. "The teeth were used in cutting off and grinding up the coarse swamp vegeta­ tion on which the giant beaver fed" (Kurten and Anderson, 1980:236). Cahn (1932) summarized distribution records of the giant beaver, based upon reported specimens by state, known at that time. Nearly 50 years later Kurten and Anderson (1980) noted that it had been reported from 30 local faunas as well as from hundreds of isolated sites. Although its remains have been found as far north as Alaska and as far south as Florida, and from Nebraska to the East Coast, Castoroides apparently oc­ curred most abundantly in the region immediately south of the Great Lakes (Faunmap Working Group, 1994). Relatively abundant remains of Castoroides also have been recorded in Florida (Martin, 1969, 1975; Morgan and White, 1995). There are, however, few records of this extinct beaver (Faunmap Working Group, 1994) for the mid-South (~33°N-36°N) east of the Mississippi River. The most recent record of Castoroides from Alabama, and apparently the first for the state, is of an in­ cisor fragment from the Bogue Chitto Creek site, a late Pleis­ tocene vertebrate assemblage in the coastal plain west of Selma, Dallas County (McCarroll and Dobie, 1994). A large portion of a right lower jaw with full dentition from Shelby County, Tennessee, was first described by Wyman (1850) and was referred to as the "Memphis specimen" by Cahn (1932:234). Parmalee et al. (1976) reported isolated cheek teeth from two caves in east Tennessee: Baker Bluff Cave (Sul­ livan County) and an unnamed cave along the Clinch River 65 66 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY FIGURE 1.—Map showing the location of Castoroides localities mentioned in the text: 1 =Baker Bluff Cave, Sul­ livan County, Tennessee; 2=Clinch River Unnamed Cave, Roane County, Tennessee; 3=Ruby Falls Cave, Hamilton County, Tennessee; 4="Memphis specimen," Shelby County, Tennessee; 5 = Bell Cave, Colbert County, Alabama; 6=Cave ACb-3, Colbert County, Alabama; 7=Bogue Chitto Creek, Dallas County, Alabama; 8=Strawberry Hill, Charleston County, South Carolina; 9=Edisto Island, Charleston County, South Carolina; 10=Leisey Shell Pit, Hillsborough County, Florida. (Roane County). Castoroides cf. ohioensis also has been re­ ported from Edisto Island, a barrier island located 34 km south­ west of Charleston, South Carolina (Roth and Laerm, 1980). We herein report four new records for the mid-South region (Figure 1): one from Tennessee (Hamilton County), two from Alabama (Colbert County), and one from South Carolina (Charleston County). ACKNOWLEDGMENTS.—Special appreciation is extended to Kent Ballew for bringing to our attention bone deposits he dis­ covered in Ruby Falls Cave and for collecting samples for study. We thank Jack Steiner and Ronnie Burk for granting per­ mission to examine the vertebrate material and to publish this record of Castoroides from Ruby Falls Cave. Gorden L. Bell and James P. Lamb, directors of the Bell and ACb-3 cave exca­ vations, and Susan Henson, Discovery 2000 (formerly known as Red Mountain Museum), are acknowledged with gratitude for the loan of the giant beaver specimens from these caves and for permission to publish the records. We acknowledge with thanks Patricia P. Adams for assisting with the collection of specimens in Ruby Falls Cave. We appreciate the assistance of Maria Wilson and Malinda Aiello for typing drafts of the manuscript. We thank Jeffrey Saunders for providing access to the Castoroides innominate from the Hopwood Farm site. We extend our gratitude to Ruth Holmes Whitehead for bringing to our attention the skull from Strawberry Hill. We also thank James Knight for access to the Strawberry Hill specimen and for permission to describe it. We appreciate the efforts of W. Miles Wright and Marlin Roos for preparation of the photo­ graphs. We thank R. Bruce McMillan for his valuable com­ ments on an earlier draft of the manuscript and we appreciate reviews by William Korth and Fred Grady. Material During 1991-1993, Kent Ballew from Hixon, Tennessee, was able to investigate several extensive unexplored passage- NUMBER 93 67 FIGURE 2.—Section of left ilium of Castoroides sp. from Ruby Falls Cave, Hamilton County, Tennessee. ways forming a complex multilevel system of corridors in Ruby Falls, a commercially operated cave in Lookout Moun­ tain, Chattanooga, Hamilton County, Tennessee. This cave, overlooking the Tennessee River, has been inhabited by ani­ mals since late Pleistocene times. Many of the more accessible passageways have been greatly modified by human activity in historic times, including mining for saltpeter during the Civil War, and more recently by enlarging and straightening pas­ sages to accommodate visitors. Animal remains occur in many of the passageway floors and streambeds; a varied extant fauna is represented, as well as ex­ tinct Pleistocene taxa, such as jaguar (Panthera onca (Lin­ naeus, 1758)) and tapir (Tapirus (Brunnich, 1772)). In the fall of 1993, Mr. Ballew recovered a 63.5 mm long section of a left ilium broken just anterior of the acetabulum (Figure 2). It was found lying on the surface of a dry streambed, but the original site of deposition could not be determined. The specimen was compared with a complete Castoroides innominate from the Hopwood Farm site, Montgomery County, Illinois (King and Saunders, 1986), housed in the paleontological collections of the Illinois State Museum, Springfield, Illinois. Although B FIGURE 3.—Castoroides sp. from Bell Cave, Colbert County, Alabama: A, upper right incisor (RMM 4000); B, incisor fragment (RMM 5547); C, incisor fragment (RMM 6223). slightly smaller and less rugose, the Ruby Falls specimen can be assigned to Castoroides with certainty. The size differences in these two specimens may be attributed to sexual dimor­ phism, ontogenetic development, or geographic variation. During the summers of 1984 and 1987, paleontologists from the Red Mountain Museum (RMM) dug six excavation units (1 x 1 m) into the floor of Bell Cave, located in the south bluff of the Tennessee River (Tennessee River Mile 248.2 (km 397.1) ~11 km west of Tuscumbia, Colbert County, Alabama). A spe­ cies-rich vertebrate assemblage was recovered, which included remains offish, turtles, birds (Parmalee, 1992), and mammals. Many of the extant species, such as fisher (Martes pennanti Erxleben, 1777) and caribou (Rangifer tarandus (Linnaeus, 1758)) (Churcher et al., 1989), have boreal affinities and no longer occur in the area. In addition to extant mammals, extinct species including tapir (Tapirus sp.) and long-nosed peccary (Mylohyus nasutus Leidy, 1869) also were represented. Re­ mains of the giant beaver from Bell Cave consisted of a 195.0 mm external section of an upper right incisor (RMM 4000) and a 44.0 x 10.5 mm incisor fragment (RMM 5547) (Figure 3A,B). Radiocarbon dates indicate that the accumulation of vertebrate remains occurred between ca. 25,000 and 11,500 radiocarbon years before present (RCYBP). Lively et al. (1992) reported Uranium (U)-series dates from travertines associated with a late Pleistocene megafauna recov­ ered in a cave designated as ACb-3 (Colbert County, Alabama). This cave is -70 m longx 15 m wide and has several huge open rooms. It is situated -0.5 km west of Little Bear Creek (Tennes­ see River Mile 249.5 (km 399.2)). ACb-3 was excavated by pa­ leontologists from RMM during the summers of 1985 and 1987. Like Bell Cave, it contained remains of both extant and extinct taxa; included in the latter group were Tapirus sp., M. nasutus, sabertooth (Smilodon fatalis (Leidy, 1868)), beautiful armadillo (Dasypus bellus (Simpson, 1930)), and Jefferson's ground sloth (Megalonyx jejfersonii (Desmarest, 1822)). Jeffer­ son's ground sloth was represented by a minimum of six indi­ viduals, including old adults and infants. Castoroides was rep­ resented in the assemblage by a single incisor enamel fragment (RMM 6223) measuring 36.5 x 22.0 mm (Figure 3c). According to the U-series dates, ACb-3 "was accumulating small verte­ brate remains as early as 228,000 B.P. and was visited by large vertebrates from about 170,000 years to at least 115,000 years B.P. and probably later" (Lively et al., 1992:1). A skull of Castoroides (Figure 4) was recovered by divers from the Cooper River, near Strawberry Hill, Charleston County, South Carolina. It has been reposited under the number SC75.33.1 in the South Carolina State Museum, Columbia, South Carolina. The skull is relatively complete, but both zygo­ matics (Figure 4A) and both incisors (Figure 4B) have been broken. In addition, the right maxilla and the right palatine bones have been broken just posterior to the alveolus for P4, but the left maxilla is complete with alveoli for the left P4-M3 (Figure 4c). All of the molar and premolar teeth are missing. The left palatine bone is broken and missing. Most of the su- SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY FIGURE 4.—Cranium (SC75.33.1) of Castoroides leiseyorum from Cooper River, Strawberry Hill, South Caro­ lina: A, dorsal view; B, right lateral view; c, ventral view. NUMBER 93 69 TABLE 1 .—Measurements (in mm) of the cranium (SC75.33.1) of Castoroides leiseyorum from Cooper River, Strawberry Hill, South Carolina. Cranial feature 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. Greatest length from anterior end of premaxillary to posterior end of occipital condyles Basal length from anterior end of premaxillary to inferior notch between occipi­ tal condyles Greatest length from anterior end of premaxillary to posterior edge of palatine wing of interpterygoid at dorsal choana Greatest length from anterior end of premaxillary to anterior edge of incisive fo- Tcimcn Greatest length from anterior end of nasals to dorsomedial notch of lamboidal Greatest length from anterior end of premaxillary to posterior edge of lamboidal crest Chord from posterior edge of incisor alveolus to anterior edge of P4 alveolus Length of alveolar tooth row from anterior edge of P4 alveolus to posterior edge of M3 alveolus Length of alveolar molar row from anterior edge of Ml alveolus to posterior edge of M3 alveolus Greatest mediolateral diameter of 11 alveolus Greatest anterodorsal diameter of 11 alveolus Width from lateral anterior edges of alveoli for P4s Greatest width across rostrum from lateral edges of the alveoli for the incisors Greatest width of anterior nares at premaxillo-nasal sutures Greatest height of anterior nares from nasal suture to premaxillary suture Greatest width across rostrum at the maxillo-premaxillary suture at anteroven­ tral edge of the infraorbital foramen Least width of postorbital constriction at fronto-parietal sutures Greatest width across postorbital processes Least width of postzygomatic constriction across squamosals and parietals Greatest width between outer edges of mastoid processes Greatest width between outer edges of parocciptal processes Greatest diameter of foramen magnum at medial edges of occipital condyles Greatest height of foramen magnum Height of occiput from base of basioccipital to middle of lamboidal crest Height of rostrum above incisive foramen Unsided 289.4 275.2 238.3 83.5 248.6 - 112.9 - - - - 39.6 74.3 41.5 42.7 80.9 66.1 85.9 81.6 150.6 115.6 28.0 19.0 68.5 98.0 Right side _ - - - - 284.8 _ 67.4 - 33.6 31.2 - - - - - - - - - - - - - - Left side _ - - - - 283.8 _ - 49.8 33.4 31.8 - - - - - - - - - - - - - - tures in the skull have not fused completely, although the size of the skull indicates that the individual was an adult. Cranial measurements are given in Table 1. Morgan and White (1995) described a new species, Castor­ oides leiseyorum, from the Irvingtonian Leisey Shell Pit, Hills­ borough County, Florida. Castoroides leiseyorum is biometri- cally and morphologically similar to C. ohioensis (Morgan and White, 1995). Castoroides leiseyorum is distinguished from C. ohioensis, in part, however, by the absence of the palatine wing of the interpterygoid fossa and the mesopterygoid fossa (Mor­ gan and White, 1995:416). Specifically, Castoroides species are unique in possessing two separate openings for the poste­ rior internal nares (see Stirton, 1965:280, fig. 3 for a detailed description). These two openings are referred to as the dorsal and ventral choanae. A large, deep, ovate fossa in the basisphe­ noid just posterior of the dorsal choana is known as the mesop­ terygoid fossa, and it is one of the most unusual characters of C. ohioensis (Stirton, 1965:280). The mesopterygoid fossa is clearly apparent in figured specimens of C. ohioensis from New York (Stirton, 1965, fig. 3) and Iowa (Hay, 1914, pl. 71: fig. 1). Castoroides leiseyorum has both ventral and dorsal choanae but lacks the mesopterygoid fossa. Instead, the basisphenoid bears a slightly concave, elongated groove along the midline that connects anteriorly to the dorsal choana (Morgan and White, 1995:416). Although the posterior palatine region of the South Carolina specimen (SC75.33.1) is broken, the basisphe­ noid and dorsal choana are well preserved. The South Carolina specimen, like the Florida specimens of C. leiseyorum, lacks a deep, ovate mesopterygoid fossa, but it has a slight concave groove extending into the dorsal choana (Figure 4c). Morgan and White (1995:420) also stated that "there is a slight ridge along the midline of the basioccipital in the Leisey crania and the portion of the basioccipital lateral to this ridge is only slightly concave." The same is true for the basioccipital of the South Carolina specimen (Figure 4c). In "typical" speci­ mens of C. ohioensis, the median ridge is higher and the lateral fossae are well developed (more concave) (Morgan and White, 1995:420). In addition, the lamboidal crests in C. leiseyorum form a dis­ tinct V-shaped outline, whereas in C. ohioensis the lamboidal crest is sharper and more vertical, and it meets the sagittal crest 70 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY at nearly a right angle (Morgan and White, 1995). The lamboi­ dal crest of the South Carolina specimen is V-shaped like that of C. leiseyorum (Figure 4A). Because the South Carolina skull shares diagnostic characteristics with C. leiseyorum Morgan and White, 1995, it is assigned to this taxon. Discussion Because of its size, the distinctive occlusal pattern of the cheek teeth, the unique longitudinally grooved enamel on the anterior and labial surfaces of the huge incisors, and its former extensive range, the giant beaver is one of the most intriguing species constituting the late Pleistocene megafauna of eastern North America. It is thought to have been more like the musk- rat in its habits than the modern beaver, inhabiting river bottom lakes, embayments, and associated bogs and swamplands where it probably fed on coarse marsh vegetation. Kurten and Anderson (1980:236) stated that "there is no evidence that the giant beaver built dams or felled trees." Castoroides ohioensis appears to have reached its greatest abundance in areas adja­ cent to the Great Lakes, especially the region that is now Mich­ igan, Illinois, Indiana, and Ohio. The animal apparently was able to adapt well to a wide variety of environments with aquatic habitats, as evidenced by its extensive temporal (Blan­ can to Rancholabrean) and geographic (Alaska to Florida and Nebraska to East Coast) ranges. Just as for the majority of the megafauna species, reasons for extinction of the giant beaver at the end of the Pleistocene are unclear (e.g., Martin and Klein, 1984). In the northern latitudes a combination of a gradual increase in the annual mean temper­ ature, resulting in replacement of a spruce/pine/hemlock forest with a deciduous one, and reduction of backwater marsh and swamp habitat may have brought about the animal's demise. The natural reorganization of biological communities at the end of the Pleistocene may have resulted in the destruction of habitats for many mammal species (Graham and Lundelius, 1984). Cahn (1932) noted that a radical change would not have been required for climatic, biotic, or hydrographic conditions to have a profound effect on a highly specialized form such as Castoroides. Competition for habitat with C. canadensis as a factor, as has been suggested by Cahn (1932) and others, is questionable in light of apparent differences between the two relative to foods, feeding behavior, and general adaptations to aquatic habitat. Furthermore, these two taxa appear to have co­ existed throughout much of the Pleistocene (Kurten and Ander­ son, 1980). Remains of Castoroides sp. from Ruby Falls, Bell, and ACb- 3 caves provide new records of this extinct beaver for the mid- South and establish its former presence along the middle stretch of the Tennessee River. During the late Pleistocene, floodplain lakes and marshes bordering the middle (northwest­ ern Georgia, northern Alabama) and lower stretches of the Ten­ nessee River were probably extensive. These environments probably persisted into the Holocene. Parmalee (1993:81), in reporting an avian assemblage from Smith Bottom Cave, Lau­ derdale County, Alabama, (approximately opposite Bell and ACb-3 caves), a primarily Archaic aboriginal deposit 8950± 950 RCYBP, commented that "the predominance of remains of ducks such as the mallard (Anas platyrhynchos), teal and closely related species suggests considerable expanses of back­ water sloughs, embayments and floodplain lakes within close vicinity to the site." The taxonomy of Castoroides in the southeastern United States is not clear. Morgan and White (1995:420) indicated that the Leisey specimens may represent the early stages in the evo­ lution of Castoroides, especially with regard to the develop­ ment of the basicranial region of the skull. The Leisey fauna correlates most closely with the late early Irvingtonian, 1.6 to 1.0 million years ago (Morgan and Hulbert, 1995). The age of the South Carolina specimen is unknown, but it is quite likely late Pleistocene. It is, therefore, possible that the basicranial features of C. leiseyorum are actually geographic variants within C. ohioensis rather than characters diagnostic of a chro- nospecies. The lack of a mesopterygoid fossa may, conse­ quently, be characteristic of a southeastern phenon of Castoroi­ des. This hypothesis could be tested, if the South Carolina specimen could be dated by radiocarbon methods (e.g., Stafford et al., 1991). It is interesting to note that all specimens of Castoroides documented to have a mesopterygoid fossa, a high median ridge and well-developed lateral fossa on the ba­ sioccipital, and a strong lamboidal crest that joins the sagittal crest at right angles (typical C. ohioensis features) come from northern states (e.g., New York, Illinois, Iowa). Martin (1969) described, on the basis of dental parameters, an extinct subspecies of C. ohioensis, C. o. dilophidus, from Florida. Castoroides ohioensis dilophidus has been found throughout Florida and may persist from the Blancan into the late Rancholabrean (Martin, 1969, 1975). Morgan and White (1995:421), however, noted that the presumed Blancan speci­ mens of C. o. dilophidus are from the Santa Fe IB fauna, which may be a heterochronic mixture of Blancan and Irvingtonian vertebrates. The diagnostic character of C. o. dilophidus is the occurrence of two isolated lophs (ids) in place of a single ante­ rior loop in the upper third molar and the lower fourth premolar (Martin, 1969:1035). Martin (1969) also assigned some postcranial elements to this subspecies, but cranial material of this taxon apparently was not known. In their later review, Morgan and White (1995: 420-421) indicated that other than the two Leisey specimens, there are no crania of Castoroides from Florida with a basicra­ nial region. Different diagnostic criteria have been used for the various Florida taxa, so it is difficult to compare C. leiseyorum (cranial characters) with C. o. dilophidus (dental characters). The only lower fourth premolar from Leisey lacks the isolated lophid characteristic of C. o. dilophidus (Morgan and White, 1995:416). Martin (1969:1035), however, indicated that not all upper third molars and lower fourth premolars of C. o. dilophi­ dus have the diagnostic isolated lophs (ids). In the original NUMBER 93 71 samples he studied, Martin (1969:1035) found that only 83% (n=6) and 29% (n=7) of the lower fourth premolars and upper third molars, respectively, had the diagnostic feature. Absence of the diagnostic feature in the only lower fourth premolar from Leisey therefore does not significantly distinguish the Leisey sample of C. leiseyorum from C. o. dilophidus. Two different interpretations of current knowledge of the taxonomy and evolutionary history of Castoroides in the Southeast are possible. On the one hand, it may be that the more typical C. ohioensis morphology (sensu stricto Stirton, 1965) is characteristic of more northern populations, and that C. leiseyorum and C. o. dilophidus are the same southern geo­ graphic variant of C. ohioensis. Under this scenario, C. leisey­ orum represents cranial characters and C. o. dilophidus typifies dental features of this southeastern phenon. If this proves to be the case, then C. leiseyorum should be synonymized with C. ohioensis, and, depending upon additional taxonomic evalua­ tions, either C. o. dilophidus or C. dilophidus would be the ap­ propriate name. On the other hand, if C. leiseyorum is shown to be distinct from C. o. dilophidus, then it appears that two dif­ ferent phena (C. leiseyorum and C. o. dilophidus) may have persisted throughout the southeastern United States (at least Florida and South Carolina) for most of the Pleistocene. Fur­ ther studies of associated cranial and dental material are re­ quired before either of these interpretations can be refuted. Literature Cited Barbour, E.H. 1931. The Giant Beaver, Castoroides, and the Common Beaver, Castor, in Nebraska. Nebraska State Museum Bulletin, 20(1): 171 — 186. Cahn, A.R. 1932. Records and Distribution of the Fossil Beaver, Castoroides ohioen­ sis. Journal of Mammalogy. 13(3):229—241. Churcher, C.S., P.W. Parmalee, G.L. Bell, and J.P. Lamb 1989. Caribou from the Late Pleistocene of Northwestern Alabama. Cana­ dian Journal of Zoology, 67:1210-1216. Faunmap Working Group 1994. FAUNMAP: A Database Documenting Late Quaternary Distribu­ tions of Mammal Species in the United States. Illinois State Mu­ seum Scientific Papers, 25(1-2): 690 pages. Graham, R.W., and E.L. Lundelius, Jr. 1984. Coevolutionary Disequilibrium and Pleistocene Extinctions. In P.S. Martin and R.G. Klein, editors, Quaternary Extinctions: A Prehistoric Revolution, pages 223-249. Tucson: University of Arizona Press. Hay, O.P. 1914. Pleistocene Mammals of Iowa. Iowa Geological Survey Annual Re­ port, 1912:22. King, J.E., and J.J. Saunders 1986. Geochelone in Illinois and the Illinoian-Sangamonian Vegetation of the Type Region. Quaternary Research, 25(l):89-99. Kurten, B., and E. Anderson 1980. Pleistocene Mammals of North America. 442 pages. New York: Co­ lumbia University Press. Lively, R.S., G.L. Bell, and J.P. Lamb, Jr. 1992. Uranium-Series Dates from Travertines Associated with a Late Plei­ stocene Megafauna in ACb-3, Alabama. Southeastern Geology, 33(l) : l -8. Martin, P.S., and R.G. Klein, editors 1984. Quaternary Extinctions: A Prehistoric Revolution. 892 pages. Tuc­ son: University of Arizona Press. Martin, R.A. 1969. Taxonomy of the Giant Beaver Castoroides from Florida. Journal of Paleontology, 43(4): 1033-1041. 1975. Giant Pleistocene Beavers and the Waccasassa River, Levy County, Florida. Bulletin of the New Jersey Academy of Science, 20( 1): 26-30. McCarroll, S.M., and J.L. Dobie 1994. Additional Pleistocene Mammals from Bogue Chitto Creek, Dallas County, Alabama. Journal of the Alabama Academy of Science, 65(1): 16-27. Morgan, G.S., and R.C. Hulbert, Jr. 1995. Overview of the Geology and Vertebrate Biochronology of the Lei­ sey Shell Pit Local Fauna, Hillsborough County, Florida. Bulletin of the Florida Museum of Natural History, 37(1)1:1-92. Morgan, G.S., and J.A. White 1995. Small Mammals (Insectivora, Lagomorpha, and Rodentia) from the Early Pleistocene (Irvingtonian) Leisey Shell Pit Local Fauna, Hills­ borough County, Florida. Bulletin of the Florida Museum of Natural History, 37(11)13:397-461. Parmalee, P.W. 1992. A Late Pleistocene Avifauna from Northwestern Alabama. Natural History Museum of Los Angeles County, Science Series, 36: 307-318. 1993. An Archaeological Avian Assemblage from Northwestern Alabama. ArchaeoZoology, 5(2):77-92. Parmalee, P.W., A.E. Bogan, and J.E. Guilday 1976. First Records of the Giant Beaver (Castoroides ohioensis) from Eastern Tennessee. Journal of the Tennessee Academy of Science. 51(3):87-88. Roth, J., and J. Laerm 1980. A Late Pleistocene Vertebrate Assemblage from Edisto Island, South Carolina. Brimleyana, 3:1-29. Stafford, T.W., Jr., P.E. Hare, L. Currie, A.J.T. Jull, and D.J. Donahue 1991. Accelerator Radiocarbon Dating at the Molecular Level. Journal of Archaeological Science. 18:35-72. Stirton, R.A. 1965. Cranial Morphology of Castoroides. In A.G. Jhingran et al., editors, Dr. D.N. Wadia Commemorative Volume, pages 273-285. Calcutta: Mining, Geological and Metallurgical Institute of India. Wyman, J. 1850. Remarks on Finding Bones of Megalonyx, Castor and Castoroides at Memphis, Tennessee. Proceedings of the Boston Society of Natu­ ral History, 3:281. The Northernmost Occurrence of the Pleistocene Vampire Bat Desmodus stocki Jones (Chiroptera: Phyllostomatidae: Desmodontinae) in Eastern North America Frederick Grady, Joaquin Arroyo-Cabrales, and E. Ray Garton ABSTRACT Four bones of the extinct vampire bat Desmodus stocki Jones were recovered from New Trout Cave, Pendleton County, West Virginia. Three of the four elements were located in a level 30 cm below a level radio-carbon dated to 29,400±1700 years before present (BP); the fourth was located in a younger layer but is sus­ pected to have been redeposited. This is the first record of Desmo­ dus stocki from the central Appalachians. Introduction The Pleistocene vampire bat Desmodus stocki Jones, 1958, occurs in several localities in the southwestern and southeast­ ern United States, northern Mexico (Ray et al., 1988), and as far south as the valley of Mexico (Alvarez, 1972). Most of the known localities are outside the range of the extant vampire bat, Desmodus rotundus, and these two vampire bats have been considered allopatric species (Alvarez, 1972). The occurrence of the extinct Desmodus stocki to the north of the common vampire bat's present range has attracted the at­ tention of paleontologists (Ray et al., 1988). The extant vam­ pire bats are intolerant of cold temperatures, and their northern­ most occurrence is limited by the 10°C winter isotherm (McNab, 1973). Morgan (1991) offered two explanations for Frederick Grady, Department of Paleobiology, National Museum of Natural History, Smithsonian Institution, Washington, D.C. 20560- 0121, United States. Joaquin Arroyo-Cabrales, Museum of Texas Tech University, Lubbock, Texas 79409-3191, United States, and Subdirec- cion de Servicios Academicos, I.N.A.H., Col. Centro, Distrito Federal 06060, Mexico. E. Ray Garton, P.O. Box 200, Barrackville, West Vir­ ginia 26559, United States. the occurrences of Desmodus stocki north of the present range of Desmodus rotundus: (1) the larger D. stocki could withstand somewhat cooler temperatures, and (2) climatic conditions were more equable during the late Pleistocene, with warmer winters and cooler summers (Graham and Mead, 1987:371). Identification of Pleistocene vampire bat remains in West Vir­ ginia is therefore important for our understanding of these highly specialized mammals, as well as of the historical bioge­ ography of the Appalachian region. ACKNOWLEDGMENTS.—Clayton Ray first identified the Des­ modus remains from New Trout Cave and encouraged the au­ thors to write this paper. Many cavers assisted in collecting and processing 2000 kg of cave earth. Thomas Stafford made the new carbon-14 dates available. Two anonymous reviewers pro­ vided many useful comments, which greatly improved this manuscript. The second author thanks the Smithsonian Institu­ tion Office of Fellowships and Grants for two Short-Term Visi­ tor Program Awards (1987, 1990). STUDY SITE New Trout Cave is located in a low escarpment 20 m above U.S. Highway 220, 5.5 km southwest of Franklin, Pendleton County, West Virginia. The cave is located on the Sugar Grove quadrangle, 38°39'09"N, 79°22'98"W, and the entrance is at an elevation of 570 m. The cave developed in the Coeymans-New Scotland limestones of Devonian age. In February 1979, a party of explorers from the Mononga- hela Grotto of the National Speleological Society discovered a deposit of bones in the cave. Samples discovered at that time and during subsequent trips indicated that the deposit was Pleistocene in age. A total of 12 collecting trips were made. The deposit was collected in 30 cm intervals to a depth of 220 cm and wet-screened through 5 mm and 1.5 mm mesh (Grady, 73 74 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY TABLE 1 .—Measurements (in mm) of humeri of Desmodus stocki from different localities (after Arroyo-Cabrales and Ray, 1997 (AR); or Morgan, 1991 (M)), of two specimens of Desmodus draculae (paratype and Loltun Cave, (Arroyo-Cabrales and Ray, 1997)) and of the specimens from West Virginia. Mean measurements are presented first, followed by ranges given in parentheses. Locality Desmodus stocki San Josecito Cave (M) San Josecito Cave (AR) Reddick 1, Florida (M) New Trout Cave, West Virginia Desmodus draculae Paratype Loltun Cave, Mexico Proximal width 6.3 (5.8-6.8) 6.4 (5.9-6.8) 6.3 (6.0-6.7) 6.6 6.5 7.6 Distal width 6.8 (6.4-7.3) 7.0 (6.6-7.6) 6.8 (6.4-7.2) 7.7 7.2 7.7 Shaft width 2.5 (2.0-2.9) 2.6 (2.0-2.9) 2.6 (2.2-2.7) 3.8 4.0 3.1 1984) at the field laboratory facilities of the Carnegie Museum of Natural History in New Paris, Pennsylvania. Cleaned mate­ rials were sorted to separate bones, teeth, mollusk shells, and other biological elements. All samples were deposited in the National Museum of Natural History, Smithsonian Institution, which subsumed the collections of the former United States National Museum (USNM). One of the important features of the excavations at New Trout Cave is the stratigraphic control provided by C-14 dating of the units. Arbitrary 30 cm levels were designated A-H from top to bottom of the excavation. Dates of three units were ob­ tained from 100 gm samples of bone collagen: level A, 17,060 ±220 years before present (yrs BP); level B, 28,250±850 yrs BP; and level C, 29,400±1700 yrs BP. All dating was per­ formed by the Carbon Dating Section, Smithsonian Radiation Biology Laboratory (no longer in operation). Additional dating by T. Stafford, University of Colorado (pers. comm., 1995) in­ dicated that level B ranges from about 13,000 yrs BP to more than 40,000 yrs BP and that level D is older than 50,000 yrs BP. The fauna includes fish, amphibians, reptiles, birds, and at least 60 species of mammals. Several studies have already been published about some of the taxa, including amphibians and reptiles (Holman and Grady, 1987), mollusks (Bogan and Grady, 1991), gophers (Grady, 1984), and collared lemming (Grady and Garton, 1981), and Grady and Gorton (1982) pro­ duced a general synthesis of the faunal elements. Vampire bat remains from New Trout Cave were briefly noted by Ray et al. (1988), who assigned them to Desmodus sp., although they are morphologically close to D. draculae Morgan, Linares, and Ray. Our study indicates that the specimens pertain to Desmo­ dus stocki. Taxonomic Account Order CHIROPTERA Family PHYLLOSTOMFDAE Subfamily DESMODONTINAE Desmodus stocki Jones, 1958 M A T E R I A L E X A M I N E D . — O n e right cochlea (USNM 374284), level D; one left humerus, distal end (USNM 374285), level D; one right humerus, proximal epiphysis (USNM 374286), level D; one right metacarpal III, proximal end (USNM 3742870), level B. The specimens clearly pertain to the genus Desmodus. They compare satisfactorily with the extant species, D. rotundus, but are much larger. They were compared with the bones available for specimens of D. stocki collected from the type locality of the species, San Josecito Cave, Nuevo Leon, Mexico, and with the paratype of D. draculae from Cueva del Gaucharo, Venezu­ ela. The specimens from West Virginia were found to be within the range of variation of D. stocki or close to it, as shown by the series of specimens from San Josecito Cave. The proximal hu­ meral epiphysis from New Trout Cave measured 3.30 mm in length and 3.27 mm in width, whereas three specimens of D. stocki measured 3.27-3.40 in length and 3.23-3.40 in width. The distal epiphysis from New Trout Cave, including the lower part of the shaft, is slightly larger than available specimens of D. stocki (Table 1). Arroyo-Cabrales and Ray (1997) pointed out that the San Josecito Cave vampire bat series was presum­ ably formed by specimens from both sexes, males being of larger size; therefore, it is possible that the West Virginian specimen was a male of extreme size for the species. Although it might be argued that the distal humeral end of our specimen, especially when the remainder of the shaft is included, overlaps with measurements for D. draculae (Arroyo-Cabrales and Ray, 1997), the two species can be separated by a diagnostic charac­ ter of the humerus: in D. stocki, including the New Trout speci­ men, the crest on the lateral epicondyle of the capitulum is low and rounded; in D. draculae and D. rotundus, the crest is well developed and sharp-pointed. The D. draculae paratype is also abraded, and its measurements likely underestimate its actual size. One of the New Trout specimens was collected from level B, but it may have been intrusive, even though the available date for that level is still within the previously reported time range for Desmodus stocki: late Pleistocene (ca. 120,000 yrs BP) to Holocene (2500-5000 yrs BP) (Arroyo-Cabrales and Ray, 1997). Most of the bone from the lower levels (D-H) of New Trout Cave is reddish in color, whereas most of the bone from levels A-C is ivory to tan. All four New Trout vampire bones are reddish in color. Three specimens were collected from level D, 30 cm below level C, which dated at 29,400 yrs BP. New NUMBER 93 75 dates (R.W. Graham pers. comm., 1995) suggest considerably older dates, especially for the D and lower levels of New Trout Cave. The herpetofauna found below level C in New Trout Cave was interpreted by Holman and Grady (1987) to be consistent with mild winters and warm summers. This interpretation is supported by the presence of the water rat, Neofiber alleni, which is found in level D (Grady and Garton, 1982). The pres­ ence of the extinct vampire bat Demodus stocki in association with these taxa suggests that consistently mild winters, which have been hypothesized as necessary to accommodate the physiological constraints of this bat, prevailed at that time. These associations also suggest that Desmodus stocki was not more cold tolerant than the extant vampire bat, Desmodus ro­ tundus. Rather, the extinct vampire bat and its associated fauna suggest that the late Pleistocene climatic conditions in West Virginia were substantially more equable than the tundra and boreal environments reported as typical for the late Pleistocene (e.g., Graham and Mead, 1987). Literature Cited Alvarez, Ticul 1972. Nuevo Registro para el Vampiro del Pleistoceno, Desmodus stocki del Tlapacoya, Mexico. Anales de Ciencias Biologicas, Mexico, 19:163-165. Arroyo-Cabrales, Joaquin, and Clayton E. Ray 1997. Revision de los Vampiros Fosiles (Chiroptera: Phyllostomidae, Des­ modontinae) de Mexico. In Joaquin Arroyo-Cabrales and Oscar J. Polaco, editors, Homenaje al Profesor Ticul Alvarez, pages 69-86. Mexico, D.F. [Mexico City]: Coleccion Cientifica, Instituto Nacio­ nal de Antropologia e Historia. Bogan, Arthur E., and Frederick V Grady 1991. Two Pleistocene Molluscan Faunas from Eastern West Virginia. In James R. Purdue, Walter E. Klippel, and Bonnie W. Styles, editors, Beamers, Bobwhites, and Blue-points: Tributes to the Career of Paul W. Parmalee. Illinois State Museum. Scientific Papers, 23:189-213. Grady, Frederick 1984. A Pleistocene Occurrence of Geomys (Rodentia: Geomyidae) in West Virginia. In Hugh H. Genoways and Mary R. Dawson, editors, Contributions in Quaternary Vertebrate Paleontology: A Volume in Memorial to John E. Guiday. Special Publication, Carnegie Mu­ seum of Natural History, 8:161-163. Grady, Fred, and E. Ray Garton 1981. The Collared Lemming Dicrostonyx hudsonius (Pallas) from a Late Pleistocene Deposit in West Virginia. Proceedings of the Eighth In­ ternational Congress of Speleology, pages 279-281. Americus, Georgia: National Speleological Society. 1982. Pleistocene Fauna from New Trout Cave. Capital Area Cavers Bul­ letin, 1:62-69. Graham, Russell W., and Jim I. Mead 1987. Environmental Fluctuations and Evolution of Mammalian Faunas during the Last Deglaciation in North America. In W.F. Ruddiman and H.E. Wright, Jr., editors, The Geology of North America, K-3: North America and Adjacent Oceans during the Last Deglaciation, pages 371—402. Boulder, Colorado: Geological Society of America. Holman, J. Alan, and Frederick Grady 1987. Herpetofauna of New Trout Cave. National Geographic Research, 3(3):305-317. McNab, Brian K. 1993. Energetics and the Distribution of Vampires. Journal of Mammal­ ogy. 54(1):131-144. Morgan, Gary S. 1991. Neotropical Chiroptera from the Pliocene and Pleistocene of Flor­ ida. Bulletin of The American Museum of Natural History, 206: 176-213. Ray, Clayton E., Omar J. Linares, and Gary S. Morgan 1988. Paleontology. In Arthur M. Greenhall and Uwe Schmidt, editors, Natural History of Vampire Bats, pages 19-30. Boca Raton, Florida: CRC Press. Second Record of the Badger Taxidea taxus (Schreber) from the Pleistocene of Kentucky and Its Paleoecological Implications H. Gregory McDonald A B S T R A C T A small fauna of late Pleistocene age from Great Saltpetre Cave, Rockcastle County, Kentucky, includes the lower jaw of a juvenile badger, Taxidea taxus (Schreber). The badger is not currently present in Kentucky, and the Great Saltpetre specimen is only the second known occurrence for the state. The badger is one of a number of taxa that was present during the Pleistocene in the east- em United States but whose modem distributions are restricted to the central and western United States. Introduction The badger Taxidea taxus (Schreber) presently inhabits the central and western United States. The eastern part of its distri­ bution includes southwestern Ohio, south-central Indiana and Illinois, southeastern Missouri, northwestern Arkansas, and eastern Texas. Late Pleistocene records of the badger outside its present range are few, and the recent discovery of a badger specimen from Great Saltpetre Cave, Rockcastle County, Ken­ tucky, provides additional information regarding its former dis­ tribution. ACKNOWLEDGMENTS.—It is with great pleasure that I in­ clude this paper in a volume dedicated to my friend and col­ league Clayton Ray. I would like to thank George Turner, Bill Simpson, John Agnew, and other members of the Greater Cin­ cinnati Grotto who collected the skeletal material from Great Saltpetre Cave, graciously donated the specimen to the mu­ seum, and generously shared their knowledge of the cave. Charles A. Long of the Museum of Natural History, University of Wisconsin at Stevens Point (UWSP), kindly loaned the juve­ nile badgers used for comparison. X-rays of fossil and modern specimens were provided through the courtesy of Malcolm H. Gregory McDonald, Geologic Resources Division, National Park Service, P.O. Box 25287, Denver, Colorado 80225. Meyn and Mary Thompson. Cliff Dickey photographed the specimens. Maryann Graham of the Illinois State Museum drafted Figure 2 using the FAUNMAP database. Anita Buck read the manuscript, Philip R. Bjork, South Dakota School of Mines and Technology, Rapid City, SD; and Elaine Anderson (deceased) reviewed it, and all provided many useful com­ ments. Any errors are the responsibility of the author. CAVE LOCATION The badger and associated fauna were recovered by mem­ bers of the Greater Cincinnati Grotto during excavation of a side passage near the Pig Pen portion of Great Saltpetre Cave, Rockcastle County, Livingston Quadrangle, Kentucky. This portion of the cave is close to the south entrance, 37°22'04"N, 84°12'12"W. Elevation near the south entrance is 305 m above sea level. The cave is located next to Crooked Creek, which drains into the Rockcastle River. Crooked Creek previously flowed through the main passage until it was diverted by a me­ ander cutoff. The cave is a solution feature in the St. Genevieve Member of the Mississippian Newman Limestone. Great Salt­ petre Cave is located on the eastern edge of the Mississippian Plateau physiographic province. Present habitat in the vicinity of the cave is mixed mesic deciduous forest. The earliest recorded mention of Great Saltpetre Cave was in 1798, when it was first explored by John Baker and his family. Saltpeter mining was initiated in late 1800, and in April 1805 bones reported to be those of the sloth Megalonyx jeffersonii (Demarest) were found in the cave. In November of the same year the skull and jaw of a peccary, believed to be Platygonus compressus LeConte, were found by Samuel Brown and saltpe­ ter miners in the cave (Barton, 1805; George, 1994). The present location of the sloth bones is unknown. According to Harlan (1825), the skull and jaw of the peccary were deposited in the cabinet of the American Philosophical Society. The col­ lections of the American Philosophical Society were later transferred to the Academy of Natural Sciences in Philadel- 77 78 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY phia. The specimens are not presently in the Academy's collec­ tions (T Daeschler, pers. comm., 1994), and their present loca­ tion is unknown. The first published description of the cave was by Brown in 1809. Following these initial discoveries no other fossil vertebrate material was reported from the cave until the discoveries made by the Greater Cincinnati Grotto. All of the recently discovered fossil material from Great Saltpetre Cave is housed in the vertebrate paleontology collections of the Cincinnati Museum of Natural History (CMNH). A late Pleistocene age for the badger is indicated by its asso­ ciation with skeletal material of the extinct peccary, Platygonus compressus. Platygonus compressus is known only in faunas of Rancholabrean age (Wright, 1993). Repenning (1987) consid­ ered the Irvingtonian/Rancholabrean boundary to be at about 400,000 years before present (yrs BP). Most records of Platy­ gonus compressus are late Pleistocene (Wisconsinan), with the youngest carbon-14 date for the species at 11,900±750 yrs BP (Ray et al., 1970). Until more work is done on the Saltpetre Cave material, nothing more precise than a Rancholabrean age for the fauna can be determined. In addition to the badger man­ dible and peccary bones, the group recovered the skulls of long-tailed weasel (Mustela frenata) and woodchuck (Marmota monax) and bones of unidentified bird species. The specimens were buried in a reddish silty-clay that had filled the cave pas­ sageway. Further excavation of the silty-clay may greatly ex­ pand the fauna. Description of Material The specimen is the right dentary (CMNH 3681) of a juve­ nile, with permanent canine in place but not erupted, and third and fourth deciduous premolars fully erupted. An X-ray of the specimen indicates that the crowns of the permanent third and fourth premolars and first and second lower molars are fully formed although not yet erupted (Figure lb). The mandibular symphysis is unfused and the two halves of the mandible had separated at this point. Mandible length is 76.1 mm; height at coronoid process, 31.0 mm; width of condyle, 14.9 mm; alveo­ lar length of premolar series, 21.8 mm; length of deciduous lower third premolar (dp3), 6.0 mm; width of dp3, 3.5 mm; length of deciduous lower fourth premolar (dp4), 9.0 mm; and width of talonid of dp4, 4.6 mm. The eruption of the permanent canine is anterior to the alveo­ lus of the deciduous canine, a feature used by Long (1965) to distinguish Taxidea from Meles. The presence of two distinct alveoli also indicates that the deciduous lower second premolar (dp2) was double-rooted, another feature of Taxidea noted by Long (1965). The crown of the dp3 has two distinct cusps—a prominent protoconid and a small accessory cusp on the posterolabial side of the protoconid—and the dp3 has a well-developed talonid. In these features the dp3 resembles the permanent lower fourth premolar, although it is not as massive, rather than the perma­ nent lower third premolar, which has a single cusp and no tal­ onid. In the dp4 the protoconid, paraconid, and metaconid are prominent and distinct, with the protoconid the largest of the three. The paraconid, the second largest cusp, constitutes the mesial portion of the tooth. The metaconid is the smallest of the three cusps and is positioned posterolingually to the proto­ conid. Although the dp4 is similar to the permanent lower first molar in relative positioning of the cusps, the talonid of the dp4 is less well developed and lacks the entoconid. Comparison of the Saltpetre Cave specimen with another ju­ venile badger (UWSP M-5318) shows identical features in the morphology of the deciduous premolars. Discussion The only previous record of the badger in Kentucky is a por­ tion of the left temporal from Welsh Cave, Woodford County (Guilday et al., 1971). Great Saltpetre Cave is about 100 km southeast of Welsh Cave and approximately 200 km southeast of the present range of badger in southwestern Ohio. The fossil record of Taxidea in the eastern United States ex­ tends from the middle Pleistocene (Irvingtonian) deposits of Cumberland and Port Kennedy Caves to Holocene archaeolog­ ical records. Gidley and Gazin (1933) originally described the material from Cumberland Cave as a distinct species, Taxidea marylandica, but Long (1964) considered it a subspecies of T. taxus. Material from Port Kennedy Cave was referred by Cope (1899) to the modern species. Reference of all Pleistocene bad­ ger material to the modern species has been followed by subse­ quent workers (Hall, 1936; Anderson, 1977; Kurten and Anderson, 1980). Late Pleistocene and early Holocene localities of Taxidea outside its present range in the eastern United States include Baker Bluff Cave, Sullivan County, Tennessee (Guilday et al., 1978); Peccary Cave, Newton County, Arkansas (Semken, 1984); Bootlegger Sink, York County, Pennsylvania (Guilday et al., 1966); Meyer Cave, Monroe County, Illinois (Parmalee, 1967); and New Trout Cave, Pendleton County, West Virginia (Grady, 1984). During the Pleistocene, the ranges of the many species ex­ panded and contracted in response to environmental changes (FAUNMAP Working Group, 1994). Taxidea forms part of a suite of species associated with grasslands and prairie whose distributions are currently restricted to the central and western United States but at one time included portions of the eastern United States (Figure 2). Other now-western species that share similar habitat preferences with the badger and have extralimi- nal eastern records include plains pocket gopher, Geomys bur- sarius (Parmalee and Klippel, 1981); thirteen-lined ground su- qirrel, Spermophilus tridecemlineatus (Guilday et al., 1978); northern grasshopper mouse, Onychomys leucogaster (Semken, 1984); and birds such as the black-billed magpie, Pica pica (Parmalee, 1992), and prairie chicken, Tympanuchus NUMBER 93 79 FIGURE 1.—Taxidea taxus juvenile jaws: A-D, Great Saltpetre Cave (CMNH 3681); E-H, modern (UWSP M- 5318). A,E, lateral view; B,F, X-ray, lateral view; C,G, occlusal view; D,H, X-ray, occlusal view. cupido (Guilday and Parmalee, 1971). Many of these taxa have been found together in Pleistocene sites in the eastern United States (Table 1). Mammalian species with western af­ finities still present in Kentucky include prairie vole, Microtis ochrogaster (which is still present in Rockcastle County); prairie deer mouse, Peromyscus maniculatus bairdii; spotted skunk, Spilogale putorius; and coyote, Canis latrans (Barbour and Davis, 1974). Compared with these other species, the badger is generally not a common species in fossil assemblages in the eastern United States. This spotty record permits two alternative inter­ pretations with regard to historical changes in the distribution of badger and its presence in the eastern United States. Under the first scenario, an initial widespread distribution in the early and middle Pleistocene (Irvingtonian) may have been followed by a gradual but continuous reduction in range, from its maxi­ mum extent as far east as Maryland and Pennsylvania to the more limited western distribution of today. This pattern is seen in Ochotona, which is known from middle Pleistocene sites in the east but whose present distribution is limited to western North America (Guilday, 1979; Mead, 1987). Alternatively, the range of the badger could have become re­ stricted farther west earlier in the Pleistocene, with the late Pleistocene and early Holocene records representing a second­ ary, short-term, eastward expansion of range during more opti­ mal environmental conditions, possibly related to the develop­ ment of prairie conditions. Grady (1984) suggested that the presence of Geomys in New Trout Cave, West Virginia, which also includes Taxidea in the fauna, represents a range expan­ sion during a dry, probably glacial interval before the latest Wisconsinan glaciation. Spermophilus tridecemlineatus is considered to be typically 80 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY TABLE 1.—Late Pleistocene and Holocene records of western species found outside of their present range in eastern sites. Age assignments are taken from FAUNMAP (1994): Wisconsinan (10-110 Ka), middle Wiscon­ sinan (35-65 Ka), late Wisconsinan (10-35 Ka), Glacial (9.5-20.5 Ka), late Glacial (9.5-15.5 Ka), Holocene (0-10 Ka), early Holocene (8-10 Ka), middle Holocene (4-8 Ka), late Holocene ( 0 ^ Ka). (7t=Taxidea taxus; Gb-Geomys bursarius; St=Spermophilus tridecemlineatus; 0\-Onychomys leucogasler; Pp= Pica pica; Tc= Tympanuchus cupido.) Location 1. New Paris No. 4, Pennsylvania (late Glacial) 2. Bootlegger Sink, Pennsylvania (late Wisconsinan/Holocene) 3. Eagle Cave, West Virginia (Holocene) 4. Trout Cave, West Virginia (Wisconsinan) 5. New Trout Cave, West Virginia (late Wisconsinan) 6. Natural Chimneys, Virginia (late Wisconsinan) 7. Clark's Cave, Virginia (Glacial) 8. Strait Canyon Fissure, Virginia (late Pleistocene) 9. Welsh Cave, Kentucky (late Glacial) 10. Savage Cave, Kentucky (Wisconsinan) 11. Great Saltpetre Cave, Kentucky (Wisconsinan) 12. Guy Wilson Cave, Tennessee (late Wisconsinan) 13. Baker Bluff, Tennessee (Glacial) 14. Robinson Cave, Tennessee (late Wisconsinan) 15. Cheek Bend Cave, Tennessee (Glacial) 16. First American Bank Site, Tennessee (Wisconsinan/Holocene) 17. Harrodsburg Crevice, Indiana (Wisconsinan) 18. Riverton Site, Illinois (late Holocene) 19. Meyer Cave, Illinois (Wisconsinan/Holocene) 20. Modoc Rock Shelter, Illinois (middle Holocene) 21. Black Earth Site, Illinois (late Holocene) 22. Kingston Lake, Illinois (late Holocene) 23. Crankshaft Cave, Missouri (Wisconsinan) 24. Brynjulfson Caves, Missouri (Wisconsinan/Holocene) 25. Zoo Cave, Missouri (Wisconsinan/Holocene) 26. Herculaneum Crevice, Missouri (Wisconsinan) 27. Bat Cave, Missouri (Holocene) 28. Kingston Saltpeter Cave, Georgia (Wisconsinan) 29. Bell Cave, Colbert County, Alabama (late Wisconsinan) 30. Peccary Cave, Arkansas (Glacial) 31. Zebree Site, Arkansas (late Holocene) Tt Gb St 01 Pp Tc * _ _ _ _ _ _ * _ _ * * _ _ * * _ _ _ * _ _ _ _ * _ _ _ _ * * _ _ _ _ _ _ _ _ * _ * * * _ _ _ * _ _ _ _ * _ _ _ * _ a grassland inhabitant, and throughout its present range it is found predominantly on moderately high prairies and knolls (Streubel and Fitzgerald, 1978). The plains pocket gopher, Geomys bursarius, lives mostly in open lands such as prairie grasslands and meadowlands, where it will range from river bottoms to hill tops as long as the soil has a low water-holding capacity (Schwartz and Schwartz, 1981). Semken (1984) con­ sidered the presence of Geomys bursarius and Spermophilus tridecemlineatus at Peccary Cave (which also includes Taxidea as a member of the fauna) as indicative of prairie conditions. He surmised that their subsequent disappearance in the region resulted from loss of habitat with closure of the deciduous for­ est. Guilday et al. (1971) suggested that during the Pleistocene the range of Geomys was continuous from the Midwest to Flor­ ida and that the present, more limited distribution of Geomys was a post-Pleistocene event. Given the similarity of habitat preference of badger and these species, this may have been the pattern followed by the badger. The degree of association between Taxidea taxus, Geomys bursarius, and Spermophilus tridecemlineatus from the sites listed in Table 1 can be calculated using the formula for the In­ dex of Association: 100C/N1+N2-C. Nl is the total number of sites for species 1, N2 is the total number of sites for species 2, and C is the number of sites in common for the two species. The data in Table 1 indicate that in eastern faunas Taxidea taxus and Geomys bursarius are found together 21% of the time, Taxidea taxus and Spermophilus tridecemlineatus are found together 26% of the time, and Geomys bursarius and Spermophilus tridecemlineatus are associated 22% of the time. From this analysis it appears that the three species have an equal probability of association with each other. Platygonus compressus, the extinct peccary, has been inter­ preted as an inhabitant of open country (Guilday et al., 1971). The association of P. compressus with badger at Great Saltpetre Cave supports the interpretation of former prairie, or at least open woodland, habitat in Rockcastle County during the Pleistocene. NUMBER 93 FIGURE 2.—Modem distribution of Taxidea taxus (shaded) and extralimital records during the Pleistocene. Num­ bers refer to late Pleistocene localities listed in Table 1. Irvingtonian localities are designated by letters: A, Port Kennedy Cave; B, Cumberland Cave. This is in contrast to the mesic forest present in the area today. Whether the present distribution of badger represents a contin­ uous unidirectional reduction in range or one of many fluctua­ tions of its range during the Pleistocene and early Holocene, its presence or absence in a fauna may not be a direct indication of long-term environmental change, either unidirectional or fluctu­ ating. As noted by Long (1973), badgers are labile in habitat preference; they range from below sea level (Death Valley) to above 3660 m in elevation and are found from the Arctic-alpine to lower Austral life zones. Given its ecological position as a predator, expansion or contraction of its range may not be the di­ rect result of environmental changes but rather may reflect changes in the distribution of prey, such as Geomys bursarius and Spermophilus tridecemlineatus. It may be the prey species that are sensitive to changes in available habitat, and that cli­ matic change results in changes in their distribution. Expansion or reduction in the range of Taxidea may be merely a secondary consequence. 82 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY Literature Cited Anderson, E. 1977. Pleistocene Mustelidae (Mammalia, Carnivora) from Fairbanks, Alaska. Bulletin of the Museum of Comparative Zoology, 148(1): 1-21. Barbour, R.W., and W.H. Davis 1974. Mammals of Kentucky. 322 pages. Lexington, Kentucky: University of Kentucky Press. Barton, B.S. 1805. Miscellaneous Facts and Observations. The Philadelphia Medical and Physical Journal, 1(2): 158. Brown, S. 1809. A Description of a Cave on Crooked Creek with Remaks and Obser­ vations on Nitre and Gunpowder. Transactions of the American Philosophical Society, 6:235-247. Cope, E.D. 1899. Vertebrate Remains from the Port Kennedy Bone Deposit. Journal of the Academy of Natural Sciences of Philadelphia, 11:193-267. FAUNMAP Working Group 1994. FAUNMAP: A Database Documenting Late Quaternary Distribu­ tions of Mammal Species in the United States. Illinois State Mu­ seum Scientific Papers, 1-2: 690 pages. George, A.I. 1994. Interim Chronology of Historic Events at Great Saltpetre Cave, Rockcastle County, Kentucky. The Electric Caver, 31(2): 10-14. Gidley, J.W, and C L . Gazin 1933. New Mammalia in the Pleistocene Fauna from Cumberland Cave. Journal of Mammalogy. 14:343-357. Grady, F.V 1984. A Pleistocene Occurrence of Geomys (Rodentia: Geomyidae) in West Virginia. Special Publication, Carnegie Museum of Natural History, 8:161-163. Guilday, J.E. 1979. Eastern North American Pleistocene Ochotona (Lagomorpha: Mam­ malia). Annals of the Carnegie Museum, 48(24):435—444. Guilday, J.E., H.W. Hamilton, E. Anderson, and P.W. Parmalee 1978. The Baker Bluff Cave Deposit, Tennessee, and the Late Pleistocene Faunal Gradient. Bulletin of the Carnegie Museum of Natural His­ tory, 11:1-67. Guilday, J.E., H.W. Hamilton, and A.D. McCrady 1966. The Bone Breccia of Bootlegger Sink, York County, Pennsylvania. Annals of the Carnegie Museum, 38(8): 145—163. 1971. The Welsh Cave Peccaries (Platygonus) and Associated Fauna, Kentucky Pleistocene. Annals of the Carnegie Museum, 43(9): 249-320. Guilday, J.E., and P.W. Parmalee 1971. Thirteen-lined Ground Squirrel, Prairie Chicken and Other Verte­ brates from an Archaeological Site in Northeastern Arkansas. Amer­ ican Midland Naturalist, 86(1 ):227-229. Hall, E.R. 1936. Mustelid Mammals from the Pleistocene of North America with Systematic Notes on Some Recent Members of the Genera Mustela, Taxidea and Mephitis. Carnegie Institution of Washington Publica­ tion, 473:41-119. Harlan, R. 1825. Fauna Americana, Being a Description of the Mammiferous Ani­ mals Inhabiting North America. 318 pages. Philadelphia: Anthony Finley. Kurten, B., and E. Anderson 1980. Pleistocene Mammals of North America. 442 pages. New York: Co­ lumbia University Press. Long, C A . 1964. Taxonomic Status of the Pleistocene Badger, Taxidea marylandica. American Midland Naturalist, 72:176-180. 1965. Comparison of Juvenile Skulls of the Mustelid Genera Taxidea and Meles, with Comments on the Subfamily Taxidiinae Pocock. Ameri­ can Midland Naturalist, 74(l):225-232. 1973. Taxidea taxus. Mammalian Species, 26: 4 pages. Mead, J.I. 1987. Quaternary Records of Pika, Ochotona, in North America. Boreas, 16:165-171. Parmalee, P.W. 1967. A Recent Cave Deposit in Southwestern Illinois. Bulletin of the Na­ tional Speleological Society, 29(4): 119-147. 1992. A Late Pleistocene Avifauna from Northwestern Alabama. Natural History Museum of Los Angeles County Science Series, 36: 307-318. Parmalee, P.W., and W.E. Klippel 1981. A Late Pleistocene Population of the Pocket Gopher, Geomys cf. bursarius, in the Nashville Basin, Tennessee. Journal of Mammal­ ogy, 62(4):831-835. Ray, C.E., C S . Denny, and M. Rubin 1970. A Peccary, Platygonus compressus LeConte, from Drift of Wiscon­ sinan Age in Northern Pennsylvania. American Journal of Science, 268:78-94. Repenning, CA. 1987. Biochronology of the Microtine Rodents of the United States. In M.O. Woodburne, editor, Cenozoic Mammals of North America: Geochronology and Biostratigraphy, pages 236-268. Berkeley: University of California Press. Schwartz, C.W., and E.R. Schwartz 1981. The Wild Mammals of Missouri. 356 pages. Columbia: University of Missouri Press. Semken, H.A. 1984. Paleoecology of a Late Wisconsinan/Holocene Micromammal Se­ quence in Peccary Cave, Northwestern Arkansas. Special Publica­ tion, Carnegie Museum of Natural History. 8:405-431. Streubel, D.R, and J.F. Fitzgerald 1978. Spermophilus tridecemlineatus. Mammalian Species, 103: 5 pages. Wright, D.B. 1993. Evolution of Sexually Dimorphic Characters in Peccaries (Mamma­ lia, Tayassuidae). Paleobiology, 19(l):52-70. Bison antiquus from Kenora, Ontario, and Notes on the Evolution of North American Holocene Bison Jerry N. McDonald and George E. hammers A B S T R A C T An associated skeleton of an adult male Bison antiquus occiden­ tal from Kenora, Ontario, is radiocarbon dated at 4270±65 yrs BP, making it the youngest unequivocal record for the species. It also extends the range 280 km north and 65 km east of the previ­ ously documented limits for the genus in the western Great Lakes area. The Kenora bison died in a shallow pond in an oak-pine woodland; it was experiencing nutritional stress at the time of its death. Both halves of the mandible had been fractured by trauma earlier in life. A review of morphological change in bison during the Holocene shows that later bison, Bison bison, were absolutely smaller, had absolutely shorter limbs, and had more robust upper limbs and more gracile lower limbs, relative to length, than did the earlier bison, Bison antiquus. Compared with that of Bison antiquus, the thoracic limb of Bison bison became elongated relative to the pel­ vic limb, and limb length increased relative to skull size. Relative to the norm in Bison antiquus, these traits diverged further in the wood bison, Bison bison athabascae, than in the plains bison, Bison bison bison. The patterns of pelage development and social behavior in Bison bison bison, however, are more distant from par­ allel patterns postulated for Bison antiquus antiquus than are pat­ terns of pelage development and social behavior in Bison bison athabascae. Introduction The skull, mandible, 14 teeth, and 58 postcranial bones of an adult male bison were collected late in the 1970s during dredg­ ing operations in a peat bog near Kenora, Ontario. This bison, now in the collections of the Manitoba Museum of Man and Nature, and cataloged as MMMN V-1914, has been referred to Jerry N. McDonald, Research Associate, Department of Paleobiology, National Museum of Natural History, Smithsonian Institution, Wash­ ington, D.C. 20560-0121; and McDonald & Woodward Publishing Company, 431-B East College Street, Granville, Ohio 43023. George E. hammers (deceased). Bison antiquus occidentalis, an extinct taxon that was the evo­ lutionary link between the late Pleistocene North American steppe bison, Bison antiquus, and the extant North American bison, Bison bison. Sediment from within the cranium of the Kenora bison was radiocarbon dated at 4850 ±60 years before present (yrs BP) (Beta-3779) and ribs of the bison were radio­ carbon dated at 4270±65 yrs BP (DIC-3381). Sediment from the site yielded pollen, plant macrofossils, and mollusks, which provided information about the regional environment at the time the bison lived. The Kenora bison is significant for several reasons. Associ­ ated skeletons of Bison antiquus occidentalis are uncommon from other than archaeological contexts, and descriptions of the postcranial bones of such associated specimens are rare. The Kenora bison was found farther north and east than any bison previously known in the western Great Lakes region, and it is a rare record of bison in Ontario. The radiocarbon age of 4270± 65 yrs BP is the youngest known for Bison antiquus. Patholo­ gies of the face and teeth provide information about the life his­ tory of this animal and its biophysical condition at the time of death, and contribute information about patterns of pathology in the taxon. Although most characters of the Kenora bison skull are of average to greater-than-average size for Bison anti­ quus occidentalis, the postcranial bones are smaller than might be expected, a circumstance that raises questions about the pat­ tern of coevolution of the postcranial skeleton and skull in bi­ son during the Holocene. In this paper we (1) describe the partial bison skeleton, pol­ len spectrum, and invertebrate remains from the Kenora site; (2) review the zoogeographical, paleoecological, and evolu­ tionary context of the Kenora bison; (3) document the pattern of skull and limb coevolution seen in North American Ho­ locene Bison, and, on the basis of the observed pattern of co- evolution, (4) comment upon the position of the wood bison, Bison bison athabascae, in the evolutionary sequence of North 83 84 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY Betula Lake j I _)W Kenora Lake of the Woods \ Manitoba r\ - . I 1 ^v: • Hayes Lake Lake of the Clouds < Nicollet Creek 300 km —I • _J FlGURE ]. Top: The Kenora, Ontario, area showing the location of the Kenora bison site. Bottom: Southwestern Ontario and adjacent regions showing the Kenora bison site and other places mentioned in the text in regional context. American Bison. All dates are given in uncalibrated radiocar­ bon years before present. ACKNOWLEDGMENTS.—We extend our appreciation to Nick Serduletz of Kenora, Ontario, for salvaging the Kenora bison skeleton and donating it to the Manitoba Museum of Man and Nature. Gary Haynes and Clayton E. Ray reviewed this paper and provided many helpful comments, but the authors retain full responsibility for the facts and opinions as finally pre­ sented. ABBREVIATIONS.—The following museum abbreviations are used: MMMN Manitoba Museum of Man and Nature, Winnipeg, Manitoba PIN Paleontological Institute, Russian Academy of Sciences, Mos­ cow. The Kenora Bison LOCATION AND DESCRIPTION OF THE KENORA SITE The bison skeleton was recovered during a period of several years by Nick Serduletz while dredging a peat bog located in the watershed of Laurensons Creek 3.0 km east of Kenora. The dredging extended to a depth of approximately 4.5 m, at which NUMBER 93 85 Trees Shrubs Herbs Aquatics Spores / / W V * J V y , / / / /S •>••>* /X/VVV / / / 7 / / / / / / / / / / / T ll Zones Years BP i _ l - 4685 6970 10780 Surface Skull Bottom I I I I I I I I I I I - 1 — i — 1 1 — I I — i— i—i—i—i—i -< 4850 i—i—i—i—r i— i— i—i i—i—i—i—i—i—i—i i—r- i—i r- t— •— i— •— i—n— 0 20 40 0 0 0 0 20 40 60 0 0 0 0 0 0 0 0 0 0 0 20 0 0 0 0 20 400 0 0 0 0 0 0 Percent Sum FIGURE 2.—Pollen and spore diagrams for Hayes Lake and the Kenora bison site, Ontario, simplified from McAndrews(1982). level the skeleton was found. The technical description of the bog location is Lot K101, Concession 4J, Township Jaffray; this site is mapped on the Kenora 52E/16 quadrangle, 1:50,000 series (Figure 1). PALEOENVIRONMENT OF THE KENORA SITE McAndrews (1982) analyzed pollen, spores, and plant mac- rofossils from three samples of sediment from the Kenora bi­ son site. One sample was from the bottom of the bog, another was from within the skull, and the third was from near the up­ per surface of the bog. McAndrews *s analyses indicated that the pattern of vegetation composition and change at Kenora (Figure 2) was similar to patterns at nearby Hayes Lake, On­ tario, and the more distant Lake of the Clouds, Minnesota (Fig­ ure 1). The bog began to fill about 10,000 yrs BP, when the regional vegetation was essentially pine forest. Between about 9200 yrs BP and 5000 yrs BP, in response to warming and drying associ­ ated with the Boreal and Atlantic climatic episodes, deciduous broadleaf species became more prominent than conifers in the regional forest, and open areas supporting grasses and other herbs presumably reached their maximum. The return of some­ what cooler and more mesic conditions with the onset of the Sub-Boreal climatic episode (ca. 5000-2750 yrs BP), initiated a return to a more closed pine forest. By about 3600 yrs BP this forest was well established (Wendland, 1978; McAndrews, 1982; Webb et al., 1983; Dykev and Prest, 1987; Wendland et al., 1987). The bison skeleton was entombed in organic-rich mud de­ posited in a lentic aquatic environment. Sediment from the skull yielded "a variety of marsh shrubs and herbs and pond plants.... This evidence suggests the bison died in a water-lily- 86 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY V FIGURE 3.—Kenora bison skull in dorsal view. (Scale bar=5 cm.) covered pond one or two meters deep that was bordered with a marsh dominated by shrubs; wetland conifers (spruce, larch) grew nearby" (McAndrews, 1982:47^48). Five species of mollusks were collected from within the cra­ nium of the bison. These included the small spire (Amnicola walkeri), three-keeled valve (Valvata tricarinata), ribbed valve (Valvata sincera sincera), modest gyraulus (Gyraulusparvus), and shiny pea clam (Pisidium nitidum). Although the pond of 4270 yrs BP probably was intimately associated with upper Laurensons Creek, similar to the situation of the depression to­ day, the gastropods from the bison skull suggest that the body of water containing the skeleton was relatively slow moving and pond-like. The Kenora site is within the present range of the five mollusks (Clarke, 1981), and their presence 4270 yrs BP does not require ecological conditions substantially differ­ ent than exist today. At the same time, these mollusks are wide­ spread today, and their sympatry 4270 yrs BP would have been possible even under environmental conditions somewhat dif­ ferent than those of today. SKELETAL AND DENTAL MATERIAL COLLECTED The skull, mandible, 14 teeth, and 58 complete or partial postcranial skeletal elements of the Kenora bison have been re­ covered. The skull is nearly complete, lacking only the nasals, most of the right premaxilla, and most of the palatine process of the left maxilla (Figures 3-5). Minor fragments are missing from other bones of the skull. Both halves of the mandible are complete (Figures 6-9). Teeth present include RP4-M3, LM2-M3, Rp4-m3, and Lp4-m3. Postcranial elements present include the atlas; axis; cervical vertebrae 3, 5, and 7; thoracic vertebrae (1), 2, 4, 5,11, 9, and ?10; lumbar vertebrae (2), (3), (4), and (5); sacrum; ribs RI, (R?2), (R3), (R4), (?R5), R6, R7, (R8), R l l , (R12), LI, (L2), (L3), (L4), (L5), L6, L8, L9, L10, LI 1, L13 or L14; manubrium; (sternebrae); R scapula; R radi- oulna; R radial carpal; R metacarpal; L scapula; L radioulna; L fused 2+3 carpal; proximal thoracic sesamoid; proximal tho­ racic first phalanx; medial thoracic phalanx; bony pelvis, R fe­ mur, R tibia, R calcaneum; L femur, L tibia, two proximal pel­ vic phalanges. Specimens in parentheses are incomplete; those not in parentheses are either complete or nearly complete. Cra­ nial, vertebral, and rib measurements are given in Tables 1-3; those for other elements are given in the Appendix. The bones and teeth of the Kenora bison are well preserved and exhibit no evidence of pre-entombment or diagenetic weathering, abrasion, trampling, crushing, or chew-gnaw dam­ age. The bones do vary in color, ranging from light to dark brown, which suggests exposure during diagenesis to different NUMBER 93 87 FIGURE 4.—Kenora bison skull in lateral view, left side. (Scale bar=5 cm.) concentrations of decaying organic matter or to groundwater with differing concentrations of oxygen. All breakage of the bones can be attributed to damage during excavation. AGE AND SEX.—The Kenora bison was approaching senes­ cence at the time of its death. All teeth were in full to advanced wear. The RM1 and both mis either had lost or were losing, by wear, the enamel fossettes within the dentine field. Several morphological features identify this specimen as a male. The skull is robust, the frontal and fronto-parietal sutures are obscured by fusion over most of the dorsal surface of the cranium, the supraorbital foramina are bridged by bone at the level of the orbits, the horn cores possess distinct burrs at the base, the horn core growth is spiraled posteriorly around the longitudinal axis, and the anteroposterior plane of the base of the horn cores is rotated rearward relative to the frontal plane. Females of all species of North American bison have relatively gracile skulls. Typically, among the short-homed North Ameri­ can bison, the frontal and fronto-parietal sutures are unfused over most or all of their length, the supraorbital foramina are not bridged by bone at the level of the orbits, the horn cores do not possess burrs at the base, the horn cores grow straight or spiral anteriorly around the longitudinal axis, and the antero­ posterior plane of the bases of the horn cores is either rotated forward relative to the plane of the frontals or parallel with the plane of the frontals (McDonald, 1981). TAXONOMIC IDENTITY.—The Kenora bison is referred to Bi­ son antiquus occidentalis on the basis of diagnostic criteria given in McDonald (1981). The horn cores of the Kenora bison are triangular in cross-section and symmetrical about the dor­ soventral axis at the base. The tips of the horn cores are ellipti- FlGURE 5.—Kenora bison skull in caudal view. (Scale bar=5 cm.) SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY TABLE 1.—Skull (linear measurements in mm, angular measurements in degrees) of Kenora bison (Bison antiquus occidentalis) compared with the means of similar measurements for other shorter-homed North American bison. (Data are from McDonald, 1981, except measurements for Banff bison, which are from Harington, 1984; values in single parentheses are estimated dimensions for the complete character; values in double parentheses are for in­ complete characters and are provided solely to describe the amount of bone present.) Measurement Spread of horn cores, tip to tip Horn core length, upper curve, tip to burr Straight-line distance, tip to burr, dorsal surface of horn core Dorsoventral diameter, horn core base Rostro-caudal diameter, horn core base Minimum circumference, horn core base Width of occiput at auditory opening Width of occipital condyles Depth of occiput, nuchal line to dorsal edge of fora­ men magnum Least width of frontals, between horn cores and orbits Greatest width of frontals at orbits Distance, nuchal line to rostral end of premaxillae Angle of divergence of hom cores forward from sagit­ tal Angle between foramen magnum and occipital plane Angle between foramen magnum and basioccipital plane Kenora bison (850.0) (315.0) 275.0 88.6 93.7 286.0 269.9 138.5 99.0 274.1 317.0 545.0 70.0 135.0 119.0 Banffbison - 265.0 - (82.0) 85.0 (260.0) - - - 301.0 - - - _ - B. a. antiauus] X 870.0 279.2 249.7 101.9 105.6 324.4 287.9 143.7 111.6 314.7 371.3 629.0* 72.9 125.4 115.6 s 71.0 35.1 29.2 9.7 12.2 32.6 18.6 8.4 9.2 19.4 18.6 - 4.8 1.6 1.7 B. a. occidentalis2 X 779.3 277.8 248.1 94.6 98.8 300.3 262.0 135.0 104.0 296.6 348.0 564.3 72.1 129.6 113.4 s 76.9 39.1 31.8 8.4 10.0 27.7 13.2 7.7 7.0 16.8 16.7 24.8 5.2 7.3 5.6 B. b. bison1 X 603.9 190.7 172.4 81.9 83.4 255.4 243.9 126.6 98.7 271.1 324.6 535.3 67.7 133.8 110.5 s AA.l 24.7 21.4 6.4 6.3 19.5 9.7 5.7 6.2 12.6 12.9 17.0 4.4 7.6 5.0 B.b. athabascae* X 681.2 235.1 207.0 91.5 97.2 289.1 273.6 130.1 99.6 293.4 354.0 578.6 71.0 129.4 113.8 s 92.0 43.9 34.4 8.7 9.5 22.9 15.3 6.4 6.6 10.5 14.8 15.2 5.0 8.8 6.6 1 Number of specimens measured varies from 1 to 41. 2 Number of specimens measured varies from 25 to 91. 3 Number of specimens measured varies from 56 to 142. 4 Number of specimens measured varies from 7 to 11. *Denotes sample containing only 1 specimen. cal in cross-section. Growth along the longitudinal axis is mod­ estly spiraled posteriorly, the posterior margins of the horn cores are slightly concave in dorsal view, and the anteroposte­ rior plane of the horn core is rotated rearward relative to the plane of the frontals. The length along the upper surface of the horn core, tip to burr, is approximately 315 mm, well within the range of 186-392 mm known for the taxon. In Bison antiquus antiquus, the tips of the horn cores are typically cordiform to triangular in cross-section and possess a distinct groove on the dorsal surface, growth is straight (not spiraled) around the arched longitudinal axis of the core, the posterior margin is straight, and the anteroposterior plane of the horn cores is par­ allel with the frontal plane. The horn cores of Bison bison grow straight along the longitudinal axis, they are straight along their posterior margins, and they are less slender (i.e., length: basal circumference ratio is less) than in Bison antiquus occidentalis. PATHOLOGIES.—Pathological conditions are apparent in the bones and teeth of the skull and in both halves of the mandible. The RP2 and P3 were missing during at least the later part of the life of the Kenora bison, as evidenced by the ossified alveo- TABLE 2.—Measurements (in mm) of vertebrae (exclusive of atlas and axis) of Kenora bison. (Measurements conform to those described for bovids in von den Driesch, 1976. Measurements in parentheses are estimated di­ mensions for the complete character.) Measurement H GL GLPa Bpacr Bpacd Bfcr Bfcd Hfcr Hfcd MinBr Bptr Height of spine C3 (155.0) 66.0 100.2 97.0 101.7 45.5 51.1 53.1 53.6 59.1 - - C5 168.0 62.0 89.5 113.2 102.6 41.0 48.1 55.1 57.7 80.1 - 69.9 C7 385.0 57.0 94.3 109.5 86.3 43.0 79.4 59.0 54.8 93.1 139.8 326.0 TI - 59.0 70.2 - - 65.4 77.5 51.7 50.3 67.6 129.4 - T2 - 61.0 66.3 - - 58.7 77.9 53.5 51.1 66.9 119.0 472.0 T4 62.0 63.7 - 57.1 74.9 52.7 50.1 44.6 110.5 — T5 - 59.0 63.2 - - 59.8 70.2 49.6 44.4 43.3 110.4 285.0 ?T7 - 60.0 63.9 - - 48.7 62.6 47.2 45.1 37.2 107.8 214.0 T9 - 61.0 60.7 - - 59.2 77.1 53.1 52.2 48.4 112.5 391.0 ?T10 - 62.0 63.5 - - 55.2 67.3 47.3 43.3 43.8 107.5 141.0 L2 158.9 64.0 99.6 65.6 47.6 48.0 52.7 45.9 43.4 42.8 (242.0) (101.0) L3 160.5 65.0 100.6 75.1 46.5 49.1 53.6 47.0 46.3 44.2 317.0 93.6 L4 155.9 64.0 94.8 78.1 49.5 49.9 55.8 47.8 45.7 52.5 (358.0) 87.4 L5 148.1 65.0 96.9 78.2 60.1 50.8 66.2 47.1 42.0 60.9 (369.2) 80.8 NUMBER 93 89 FIGURE 6.—Kenora bison mandible in dorsal view. (Scale bar=5 cm.) lar trough that would have accommodated the roots of these teeth (Figure 6). A hole is present in the dentine fields of LM3. The insertion tubercle for the masseteric muscle on the right mandible is conspicuously larger than the same tubercle on the left mandible (Figure 7). Both the right and left halves of the mandible had been fractured and healed. The fractures were centered approximately below the right p2-p3 and left p4-ml (Figures 6-9). The right mandibular body contains roughened FIGURE 7.—Left (top) and right (bottom) halves of Kenora bison mandible in lateral view. (Scale bar=5 cm.) 90 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY FIGURE 8.—Left (top) and right (bottom) halves of Kenora bison mandible in medial view. (Scale bar=5 cm.) FIGURE 9.—Kenora bison mandible in ventral view. (Scale bar=5 cm.) NUMBER 93 91 TABLE 3.—Measurements (in mm) of ribs of Kenora bison. (Measurements in double parentheses are for incomplete elements and are provided solely to de­ scribe the amount of the element that is present.) P R E - H O L O C E N E H O L O C E N E Rib number RI ?R2 R3 R4 ?R5 R6 R7 R8 Rll R12 LI L2 L3 L4 L5 L6 L8 L9 L10 Lll L13 Length 316 ((210)) 360 466 ((84.9)) 604 630 ((400)) 563 ((460)) 309 ((229)) ((260)) ((135)) ((224)) 609 644 645 618 568 396 Minimum anteropos­ terior diameter 27.6 - 26.1 26.7 - 27.3 25.1 - 19.6 19.9 27.4 25.4 - - 29.7 27.6 25.8 21.6 18.2 18.8 17.9 Minimum circumference 73.0 - 64.0 70.0 - 74.0 69.0 - 62.0 57.0 74.0 63.0 - - 76.0 76.0 69.0 61.0 61.0 61.0 56.0 bone on the lateral, ventral, and medial surfaces of the healed fracture, and a subcircular hole leading to the interior of the bone opens through the ventral surface of the fractured area. Rostral to the level of Rml, the body of the right ramus of the mandible is conspicuously wider than that of the left, and the diastema is displaced ventral to its normal position. Root ca­ nals for the missing Rp2 and Rp3 are present in the alveolar trough, indicating that these teeth were present at, or at least shortly before, the time of death. The medial side of Rm3 was shorn away in life, and the dentine surface thus exposed was worn smooth by use. The surface of the healed left body of the mandible is much smoother than that of the right, and no dis­ placement of the diastema occurred. A circular opening is lo­ cated between and below Lp4-Lml; both of these teeth are loosely rooted in the alveolus, and an abnormally long gap sep­ arates the two teeth. The bone surrounding the opening is dense and does not appear to have been affected by infection at the time of death. Root canals for the missing Lp2 and Lp3 are present, but they are shallow—a condition that suggests either displacement of the teeth or modification of the roots by the trauma to the mandible or by its subsequent healing. Enamel on the medial surfaces of Lp4, Lml, and Lm3 had been chipped, part of the medial wall of the rostral cusp of Lm3 is missing, and holes are present in the dentine fields of Lm3 and Lp4. The tooth rows of both halves of the mandible are unusually sinu­ ous, probably reflecting modification of the tooth alignment by the fractures and subsequent healing. No pathologic conditions were noted in any of the postcra­ nial bones. O yeore BP FIGURE 10.—The evolutionary fold of 5000^1000 radiocarbon yrs BP, when the evolution of Bison bison from Bison antiquus was rapidly completed (from McDonald, 1981). DISCUSSION The Kenora bison lived at the time when the nucleus of the North American bison range was anchored on the northern Great Plains and when the final, rapid steps in the evolution of Bison bison from Bison antiquus were taking place (Figure 10) (McDonald, 1981). The radiocarbon age of 4270±65 yrs BP is the youngest such date of which we are aware for a specimen that can be assigned with confidence to Bison antiquus occi­ dentalis. Harington (1984) reported a partial cranium from Banff, Alberta, dated at 3240±90 yrs BP (1-11638), which he referred to Bison bison occidentalis, equivalent to Bison anti­ quus occidentalis as used in this paper. Having examined the Banff cranium, we believe that this specimen is referable to Bi­ son bison. All measurements that can be obtained from the skull of the specimen are within the range of Bison bison (Ta­ ble 1), and the qualitative aspects of characters still visible on the specimen also are consistent with those typical of Bison bi­ son. Measurements of the Kenora bison skull are all within the range documented for Bison antiquus occidentalis, but some are outside the observed ranges for Bison antiquus antiquus and Bison bison (Table 1; cf. tables 21, 25, 29, and 34 in Mc­ Donald, 1981). Most features of the Kenora bison skull con­ form harmoniously with both qualitative and quantitative at­ tributes of the skull typical of Bison antiquus occidentalis. The horn cores are longer and more slender (i.e., long relative to di­ ameter at base), and the dorsal part of the cranium is narrower than typical specimens, however. The breadth of the base of the cranium is near the mean for the taxon. All measured dimensions of the long bones of the Kenora bi­ son are smaller than the observed means for the limbs of Bison antiquus occidentalis except the anteroposterior diameter of the femur, which is slightly larger (within one standard devia­ tion (s)) than the observed mean for the taxon (Table 4). The length of the radius is >2s, and that of the femur and tibia are between \s and 2s, of the mean for Bison antiquus occidentalis long bone lengths, whereas the length of the metacarpal in the Kenora bison is smaller than the heretofore-documented mini­ mum for the taxon. The transverse diameters of all four limb 92 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY TABLE 4.—Limb measurements (in mm) for shorter-horned North American bison, males only. (For each taxon, top row is mean measurement, bottom row is standard deviation. Data from McDonald, 1981. AP=anteroposterior diameter; TR=transverse diameter; L=rotational length; *=sample with only 1 specimen.) Species Bison antiquus antiquus Bison antiquus occidentalis Kenora bison (MMMN V-1914) Bison bison bison Bison bison athabascae n 1-28 38-161 1 37-118 2-3 Humerus L 339.0* S=- 356.9 5=21.6 - 325.9 5=11.1 354.5 5=6.3 AP 69.7 2.8 66.4 4.5 - 61.5 3.1 67.5 0.7 TR 58.7 0.5 55.0 4.2 - 50.6 3.1 58.5 0.7 Radius L 343.0 7.7 343.8 14.7 309.0 320.5 9.2 341.3 14.0 AP 33.7 0.9 35.5 2.1 32.0 32.2 1.7 35.0 1.0 TR 54.5 1.2 58.5 3.3 57.8 53.9 2.8 59.7 3.5 Metacarpal L 222.0 6.6 219.8 8.6 190.0 206.3 7.1 228.7 4.1 AP 30.7 1.2 30.1 1.8 28.2 27.2 1.6 30.3 1.1 TR 52.3 3.5 51.4 3.7 48.3 45.4 3.4 52.0 2.0 L 429.0 14.0 403.7 18.0 383.0 380.7 10.4 421.3 11.5 ls from the observed mean for Bison antiquus occi­ dentalis. At the same time, however, the dimensions of the Kenora bison long bones are also within the ranges observed for Bison bison bison. The two thoracic limb bones of the Kenora bison are >\s shorter than the mean, whereas the two pelvic limb bones are very near the mean for Bison bison bison. The diameters of all four bones are near the mean for Bison bi­ son bison. All of the long bones of the Kenora bison are smaller than the documented range for equivalent characters in male Bison bison athabascae. The fact that the Kenora bison lived near the end of the Bison antiquus lineage, as Bison antiquus was changing rapidly into Bison bison (Figure 10), suggests that the Kenora bison should be relatively small bodied compared with the mean for Bison antiquus occidentalis. Except for the horn cores, which are rel­ atively long for the subspecies, the cranium of the Kenora bi­ son generally conforms to this expectation of small size. The long bones are shorter than might be expected when compared with the size of the skull. If the evolutionary reduction in all skeletal elements were taken to be isometric, these long bones would probably be identified casually as those of Bison bison bison had they been found without the associated skull. In gen­ eral, however, the long bones are broad and robust relative to their length, a condition suggesting that the Kenora bison was more heavily built than Bison bison bison. In addition, the bones of the thoracic limb are shorter relative to those of the pelvic limb, a condition resembling more nearly the limb pro­ portions of Bison antiquus antiquus than Bison bison bison. (The thoracic limb:pelvic limb ratio was lowest (0.81) in Bison antiquus antiquus, greater (0.83) in Bison bison athabascae, still greater (0.846) in Bison bison bison, and greatest (0.854) in Bison antiquus occidentalis (i.e., the thoracic limb, relative to the pelvic limb, was shortest in Bison antiquus antiquus and longest in Bison bison occidentalis). Bison bison occidentalis also demonstrated the greatest variation in limb dimensions, which is not unexpected given its dynamic role as a decana- lized evolutionary bridge between two relatively canalized spe­ cies (Tables 5, 6). ZOOGEOGRAPHIC SIGNIFICANCE.—The Kenora bison is a range record for the subspecies, species, and genus. The nearest known remains of bison from natural contexts are of Bison an­ tiquus antiquus from Betula Lake, Whiteshell Provincial Park, Manitoba, and Bison antiquus occidentalis from Nicollet Creek, Itasca State Park, Minnesota, respectively 65 km west and 280 km south of Kenora (Figure 1) (Cleland, 1966; Shay, 1971; McDonald, 1981). Prehistoric Bison bison remains have been recovered from the Hungry Hall archaeological site along the east side of Rainy River near where it discharges into Lake of the Woods, Ontario (Moore, 1975). Bison from Hungry Hall are north and east of the known natural range of Bison bison; they probably were imported by Indians from the nearby bison range. TABLE 5.—Ratios of anteroposterior and transverse diameters to lengths of long limb bones. (Data from Mc­ Donald, 1981, as presented in Table 4. AP=anteroposterior diameter; TR=transverse diameter; L=rotational length; *=sample with only 1 specimen.) Species Bison antiquus antiquus Bison antiquus occidentalis Kenora bison (MMMN V-1914) Bison bison bison Bison bison athabascae n 1-28 38-161 1 37-118 2-3 Humerus AP:L TR:L 0.206* 0.173 0.186 0.154 - 0.180 0.155 0.190 0.165 Radius AP:L TR:L 0.098 0.159 0.103 0.170 0.104 0.187 0.100 0.168 0.103 0.175 Metacarpal AP:L TR:L 0.138 0.236 0.137 0.234 0.148 0.254 0.132 0.220 0.132 0.227 Femur AP:L TR:L 0.131 0.137 0.124 0.122 0.135 0.127 0.126 0.124 0.128 0.123 Tibia AP:L TR:L 0.103 0.143 0.096 0.133 0.095 0.140 0.093 0.132 0.092 0.129 Metatarsal AP:L TR:L 0.129 0.149 0.122 0.145 - 0.119 0.138 0.120 0.141 NUMBER 93 93 TABLE 6.—Ratios between selected cranial and limb biometrics for shorter-horned North American Bison, males only. (Data from McDonald, 1981. FW = least frontal breadth; N-Pm = nasion to premaxilla; BW=breadth across base of skull at mastoid processes; TL=combined rotational length of humerus, radius, and metacarpal; PL= combined rotational length of femur, tibia, and metatarsal; HCL=horn core length.) Species Bison antiquus antiquus Bison antiquus occidentalis Kenora bison (MMMN V-1914) Bison bison bison Bison bison athabascae FW/N-Pm 0.500 0.526 0.503 0.506 0.507 BW/N-Pm 0.457 0.464 0.495 0.455 0.472 TL/PL 0.811 0.854 - 0.846 0.832 HCL/TL 0.308 0.302 - 0.224 0.254 HCL/PL 0.250 0.258 - 0.189 0.212 FW/TL 0.348 0.322 - 0.318 0.317 FW/PL 0.283 0.275 - 0.269 0.264 PALEOECOLOGICAL SIGNIFICANCE.—During the Atlantic cli­ matic episode (ca. 8500-5000 yrs BP) the regional vegetation of extreme southwestern Ontario was dominated by a mixed forest. Early in the episode, when conditions were cooler and more mesic, conifers dominated. By the end of the episode, when the warming and drying trend reached its maximum, de­ ciduous broadleaf species were more prominent than conifers in the regional forest, and open areas supporting grasses and other herbs presumably reached their maximum. The onset of the Sub-Boreal climatic episode (ca. 5000-2750 yrs BP) brought a cooler, more mesic climate, which led to a closed for­ est with conifers increasing in importance relative to deciduous broadleaf species. The parkland vegetation that characterized the Kenora region around 5000 yrs BP, with the deciduous browse and sparse grasses, would have afforded the best local environmental con­ ditions available during the Holocene for supporting popula­ tions of bison. This does not mean that the population of bison in the Kenora region was either large or permanent. The Kenora bison does not possess morphological characteristics suggesting inbreeding depression in the population (Mc­ Donald, 1981), but it does show evidence of extended nutri­ tional stress. McAndrews's (1982) interpretation of the death of the Kenora bison—that it died in a shallow pond—is consistent with taphonomic evidence as well as paleobotanic and sedi­ mentary evidence. Similarly, the bones recovered consist of a random array of elements, not the patterned collection that would be expected had the bison been killed by predators or its carcass scavenged, or had flowing water sorted the bones (Voorhies, 1969; Haynes, 1980, 1982; Shipman, 1981). All breakage in the bones appears to be the result of excavation by heavy machinery rather than the result of battering in a high- energy fluvial environment. Some of the elements appear to have weathered after excavation. INTERPRETATION OF PATHOLOGIES.—Two pathologic condi­ tions appear to be represented in the Kenora bison. A trauma to the face appears to have broken the bodies of both halves of the mandible and dislodged the R/L? P2 and P3. A sharp blow to the right half of the mandible, along or immediately caudal to the diastema, could have produced the fractures that broke the bone completely through, allowing it to become displaced, and bent the body of the left side of the mandible to the left, pro­ ducing a green-stick fracture that separated the medial and ven­ tral parts but not the dorsolateral part. The incompletely sev­ ered left body could have functioned as the splint that held the rostral parts of both halves of the mandible in place while they healed. Healing occurred with slight ventral displacement of the diastema of the right body. In addition to the pathologic conditions discussed above, the broken teeth conform to a pattern of tooth pathology seen in the dentition of numerous musk oxen (Ovibos moschatus) from Greenland, the Arctic islands of Canada, and Alaska (Henrich- sen, 1981, 1982; Anne Gunn, pers. comm., 1987; McDonald, unpub. data). The usual pattern of this pathology is that holes exist in the dentine field in correlation with broken teeth. In musk oxen, the enamel walls of teeth are often conspicuously thin in individuals exhibiting the condition. The etiology of this condition is not known, but it is believed to result from inade­ quate or imbalanced intake of (= access to) essential nutrients, such as calcium or phosphorus. The hypoplastic condition of the dentine and enamel appears to weaken the teeth to the ex­ tent that normal use results in tooth breakage. In musk oxen ex­ hibiting hypoplasia and subsequent breakage, irritation of the alveolar bone is common, and teeth frequently are loosely rooted and have nodulated roots. In the Kenora bison, the medial wall of the Rm3 was broken to below the gum line (Figure 8), but no irritation of the alveo­ lar bone is apparent. Indeed, the alveolar bone is dense and most of the teeth, except those near the fractures, are well rooted. The fact that some teeth exhibit hypoplasia and are chipped or more extensively broken suggests that the Kenora bison was experiencing modest long-term nutritional stress. This stress could have been caused by the inadequate supplies of the requisite nutrients in the formerly glaciated, shallow- soiled Kenora area. Gross Allometry in Holocene Bison Evolution The evolution of limb length and proportions in North Amer­ ican Holocene bison is summarized in Tables 4 and 5. The most general tendency was toward absolutely shorter limbs, with more robust (i.e., greater shaft:length ratio) upper limb bones and more gracile (i.e., lesser shaft:length ratio) metapodials, relative to length. Superimposed on this pattern was a trend to­ ward the evolution of longer thoracic limbs relative to pelvic limbs. Both of these observations apply to the derivation of the late Holocene Bison bison from the late Wisconsinan Bison an- 94 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY tiquus. In other ways, the pattern of limb evolution in Bison bi­ son athabascae diverged from that in Bison bison bison. The latter evolved to absolutely smaller size, with the greatest rela­ tive reduction of length taking place in the upper limb bones and the least occurring in the metapodials. In Bison bison atha­ bascae, the femur and humerus evolved to be only slightly ( l%-3%) shorter than in Bison antiquus antiquus, the radius and tibia remained about the same size, and the metapodials be­ came longer. The skull of bison became smaller during the Holocene, and the proportions of some characters altered, which produced changes in shape. The most obvious changes were those in­ volving the horn cores, which became smaller, directed more caudally, and rotated caudally. The cranium became more domed on the dorsal surface, and the occipital condyles pro­ jected less far from the occiput. The skull became shorter and, on the dorsal surface, narrower in nearly equal proportions; but the rate of shortening exceeded the rate at which the base of the cranium narrowed, leaving the ventral side of the cranium pro­ portionately wider in Bison bison than it had been in Bison an­ tiquus. When the evolutionary changes in the limbs and skull of bi­ son are correlated, the pattern that emerges is a trend toward a relatively smaller head (with conspicuously allometrically shorter horns) set relatively lower on a body whose limbs, es­ pecially the thoracic limbs, were becoming relatively longer (Table 6). Bamforth (1988), drawing upon data in McDonald (1981), observed that limb length in North American bison in­ creased relative to overall body size during the early Holocene. The relative lengthening of the thoracic limbs and the changes in the robustness of the limb bones are probably asso­ ciated with the evolution of relatively more cursorial habits in Bison bison than characterized Bison antiquus antiquus. The musculature of the upper limbs was exaggerated, relative to that of Bison antiquus antiquus, paralleling the increased rela­ tive robustness of the upper limb bones, and the metapodials developed into support structures carrying absolutely less weight and better equipped for more frequent and versatile mo­ bility than had typified Bison antiquus antiquus. The length of the thoracic limb, relative to the pelvic limb, in Bison bison bi­ son is greater than in Bison bison athabascae, probably be­ cause of selection for greater cursoriality, more frequent or spe­ cialized agonistic encounters during the rut (which selected for greater forebody strength, focused on the thoracic limb, for de­ livering and enduring head-to-head clashes and various head- thrusting movements), and the essentially obligatory grazing feeding behavior (requiring regular raising and lowering of the relatively low-slung but massive head) of the former, all as dis­ cussed at greater length in McDonald (1981). Bison bison atha­ bascae, on the other hand, is less cursorial, probably has less frequent or less specialized agonistic encounters during the rut, and has greater opportunity to browse at higher than ground level with its relatively high-placed head; it has therefore de­ veloped a relatively lower hump and shorter thoracic limb. The Evolutionary Place of Bison bison athabascae The evolutionary position of Bison bison athabascae has been a subject of scholarly and practical concern for a century. The existence of a form of extant bison in north-central Canada different from the plains bison, Bison bison, has been known to science since at least 1829 (Richardson, 1829; Hind, 1859). Most early writers regarded the northern bison as little more than a phenotypic variant, or merely a marginal population, of the plains bison. Rhoads (1897), however, regarded the north­ ern, or wood, bison to be taxonomically distinct and gave it the name Bison bison athabascae. Scientific and political opinion on the taxonomic status of the wood bison has vacillated fre­ quently during the last century. There is no question, however, that the bison native to the parklands of north-central Canada during the nineteenth and twentieth centuries were demonstra­ bly larger than the contemporary plains bison. Here we follow McDonald's (1981) premise that it is scientifically useful and valid to designate the wood bison as taxonomically distinct, a position recently reinforced by van Zyll de Jong and others (1995). The wood bison became protected by law in 1891, and two years later a preserve was created as a refuge for the remaining populations. The total population of wood bison was about 1500-2000 when, between 1925 and 1928—at a time when wood bison were considered by responsible authorities as not distinct taxonomically from the plains bison—6673 plains bi­ son were introduced into Wood Buffalo National Park. Today, most zoologists familiar with North American bison recognize only a single extant North American species—Bison bison— containing the two subspecies Bison bison bison (the plains bi­ son) and Bison bison athabascae (the wood bison) (e.g., Skin­ ner and Kaisen, 1947; McDonald, 1981; Nowak and Paradiso, 1983; Meagher, 1986; van Zyll de Jong, 1986; van Zyll de Jong et al., 1995). Flerow (1971) considered the wood bison to be­ long to a separate species, Bison priscus, and he identified it as Bison priscus athabascae. Geist (1990), however, recently ar­ gued that the wood and plains bison are taxonomically identi­ cal. Interpretations of the evolutionary origins of Bison bison athabascae have fallen generally into two categories. The first interpretation is that Bison bison athabascae represents a de­ scendant (either changed or unchanged) form of bison derived from ancestral populations that originated in eastern Siberia or Beringia and dispersed southward. Bison bison athabascae then represents either (1) the evolutionary and geographic end- point of that parent dispersal or (2) an arrested phase of the evolutionary trajectory that continued as the parent dispersal continued southward to produce, either alone or with introgres- sion with more southerly bison, the plains bison (Skinner and Kaisen, 1947; Guthrie, 1970; Flerow, 1971; Harington, 1984; van Zyll de Jong, 1986). A second interpretation is that Bison bison athabascae represents a (changed) descendant form of bison derived from ancestral populations that originated in NUMBER 93 95 midlatitude North America and dispersed northward during the early and middle parts of the Holocene (McDonald, 1981). Elsewhere, McDonald (1981) has argued against the Berin- gian origin of Bison bison athabascae and Bison bison bison and for the common origin of both of these subspecies from Bi­ son antiquus. McDonald relied upon the dense and continuous factual record documenting both the spatial, temporal, and morphological continuity among midlatitude North American bison throughout the Holocene and the absence of any such spatial and morphological continuity among bison in, or those coming from, Beringia. Quite simply, the vast body of factual evidence still demonstrates that Bison bison athabascae is mor­ phologically, temporally, and spatially nearer to Bison bison bi­ son than to the late Pleistocene bison of eastern Siberia. Opin­ ions to the contrary are based upon antiquated concepts of either (1) progressive diminution of body size or (2) periodic radiation of new taxa into North America from Siberia, or both (e.g., Allen, 1876; Gromova, 1935; Schultz and Frankforter, 1946; Skinner and Kaisen, 1947; Guthrie, 1970; Wilson, 1975, 1996; Harington, 1986). Van Zyll de Jong (1986) argued the case for a Beringian ori­ gin for Bison bison athabascae and supported his thesis with a well-preserved skull (PIN 835-624-39) from the Bolshaya Chukochya River in Siberia. He considered this skull to be rep­ resentative of "athabascae-\ike bison in eastern Siberia at the close of the last glaciation." "The similarity of this bison to North American occidentalis and to athabascae is strong evi­ dence of the former existence of an ancestral Beringian popula­ tion" (van Zyll de Jong, 1986:53). Van Zyll de Jong did not ex­ amine the specimen personally, and he provided no absolute age for the specimen or the population it represented. Nor did he figure the specimen, describe any diagnostic characteristics that would allow it to be accepted unequivocally as resembling Bison bison athabascae or Bison antiquus occidentalis more nearly than a Eurasian form of Bison, or provide measure­ ments. To diagnose this specimen, he relied upon linear mea­ surements alone (which were sunk in his quantitative summa­ ries and never presented, in the 1986 paper, as raw data); but measurements alone frequently are insufficient to diagnose most bison specimens, especially those representing taxa that obviously overlap others in size. It is quite possible to obtain identical measurements from two or more bison specimens that possess substantially different qualitative (and genetically con­ trolled) characteristics. It also is theoretically possible for con­ vergent evolution to produce similar phenotypes from gener­ ally similar ancestral stock even in the absence of recent genetic unity. The morphological, behavioral, and ecological similarities between Bison bison and the Eurasian bison or wisent, Bison bonasus, is a case in point. Although these spe­ cies differ in morphological detail, they evolved generally via pathways that resulted in convergence of size, form, and func­ tion; it is obvious from the fossil record that Bison bonasus had a larger ancestor that could have passed through a stage during which linear features of its skull morphology resembled those of Bison bison athabascae. Lastly, bison collected from along the Bolshaya Chukochya River span nearly 1.7 million years, and most specimens have been collected without stratigraphic context and have not been radiocarbon dated (Sher, 1974, 1986, pers. comm., 1992; McDonald et a l , 1991). Thirty years ago, Sher cautioned against the assignment of bison to a given geological age based upon the size and shape of horn cores, noting that this practice "is risky to say the least" (Sher, 1974:207). Our position is that the weight of factual evidence supports the view that Bison bison athabascae originated in North America from the same lineage that gave rise to Bison bison bi­ son. McDonald (1981:260-261) was of the opinion that Bison bison athabascae could have attained its morphological char­ acteristics by either "(1) suspended equilibrium, in which case B. b. athabascae retained large body size as it evolved charac­ ters from B. antiquus adapted to the new r-type forest/wood­ land selection regime; or (2) more recent adaptive differentia­ tion, in which case B. b. athabascae has increased in body size from B. b. bison Elements of both alternatives could have operated in the origin and continuation of B. b. athabascae, but the second has probably been more important over time " Considering the evolutionary trends in the size, shape, and proportions of skeletal elements, as documented for bison dur­ ing the Holocene, one is left with the realization that Bison bi­ son athabascae may be considered the most "advanced" or "furthest evolved" form of Holocene bison in several important respects. When the skull and limb morphology typical of the ancestral Bison antiquus antiquus is compared with that of the derivative taxa, Bison bison bison and Bison bison athabascae, the integrated skeleton of Bison bison athabascae is less like that of Bison antiquus antiquus than is Bison bison bison. Spe­ cifically, in Bison bison athabascae, the skull is smaller, the upper limb bones are relatively more robust, and the metapodi­ als are relatively more gracile than in Bison bison bison. Rela­ tive to head size, Bison bison athabascae is the most long- limbed of short horned North American bison, whereas Bison antiquus antiquus is the shortest limbed. In other respects, Bison bison athabascae is probably not as distant from Bison antiquus antiquus as is Bison bison bison. The evolutionary lengthening of the front limb relative to the hind limb has progressed further in Bison bison bison than in Bison bison athabascae. Also, the forebody pelage used to ef­ fect social rank and breeding success during the rut is less ex­ aggerated and seasonally pronounced in Bison bison athabas­ cae than in Bison bison bison (Geist and Karsten, 1977; Lott, 1979). The horns therefore perhaps have relatively greater im­ portance for establishing social rank or dominance in Bison bi­ son athabascae than in Bison bison bison. The interpretation of skeletal measurements presented herein suggests that the evolution of the skeleton of Bison bison atha­ bascae has, collectively, proceeded further than has that of Bi­ son bison bison, whereas the opposite is true for evolution of selected morphological features and behavior of social signifi- 96 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY cance. Claims that Bison bison athabascae is more or less far evolved from its ancestral morph should, therefore, indicate ex­ actly which parameter of evolution is being considered. It is not necessary that all parameters evolve the same distance, at the same time, or in the same direction. As environments change, the multitude of selective pressures that act on individ­ ual organisms also change. The skeleton of Bison bison atha­ bascae appears to have differed allometrically from Bison anti­ quus to a greater extent than has the skeleton of Bison bison bison, but behaviorally and ecologically Bison bison athabas­ cae might differ less from Bison antiquus than from Bison bi­ son bison. Appendix Measurements (in mm) of limb bones of Kenora Bison (Measurements and abbreviations follow those given for bovids in von den Driesch, 1976) 1. Atlas: GB=199.9; GL=129.5; BFcd=111.3; BFcr= 123.8; H=102.1. 2. Axis: LCDe=122;H=162.1;LAPa=122.6;BFcd=67.7; BPtr=132.8; BPacd=88.5; BFcr=113.6; SBV=75.3. 3. Manubrium: Dorsoventral depth=92.3; lengfh= 104.5; breadth at articular surfaces=47.2; minimum breadth= 17.5. 4. Sacrum: BFcr=72.0; HFcr=35.3. 5. Scapula (right): Ld=276; HS=504; DHA=487; SLC= 79.2; GLP=95.4; LG=75.0; BG=69.8; transverse breadth of neck= 37.9; breadth, lateral border of spine to medial border of neck=76.7. 6. Radius (left): Approximate rotational length=309; GL (radius)=333; GL (radioulna)=452; minimum anteropos­ terior diameter of diaphysis=32.0; minimum transverse di­ ameter of diaphysis=57.8. 7. Ulna (left): GL=425; DPA = 98.9; SDO = 68.5; LD = 145.7; minimum transverse breadth of diaphysis=21.3. 8. Radial carpal (right): GB=48.2. 9. Fused 2+3 carpal (left): GB=47.4. 10. Metacarpal (right): GL= 190.0; SD=48.3; DD=28.2; BD =76.8; anteroposterior diameter, proximal epiphysis=46.0; anteroposterior diameter, distal epiphysis=40.3; Bp=77.5. 11. Proximal sesamoid, thoracic limb: length=26.1; trans­ verse breadth=17.8; anteroposterior depth =15.8. 12. Proximal phalanx, thoracic limb: GLpe=65.3; Bp=39.7; SD=35.8; Bd=39.6. 13. Medial phalanx, thoracic limb: GLpe=43.3; Bp=38.1; SD=31.3;Bd=35.9. 14. Bony pelvis: GL=526; GBTi=279; SBI=180; GBA= 268; SB=37.0; GBTc=503; LS=208.6; LA=89.6. 15. Femur (right): GL=436; GLC=405; rotational length= 383; Bp=154.1; SD=48.8; Bd=119.3; DC=56.4; antero­ posterior diameter of diaphysis=51.8. 16. Tibia (right): Rotational length = 370; GL=403; Bp= 127.4; Bd= 103.4; SD=51.7; minimum anteroposterior di­ ameter of shaft=35.0; Bd=75.7; Dd=53.0. 17. Calcaneum (right): GL=158.9; GB = 56.7; transverse breadth of neck=23.0. 18. Proximal phalanx 1, pelvic limb: GLpe=69.7; Bp=34.2; SD=31.6;Bd=35.2. 19. Proximal phalanx 2, pelvic limb: GLpe=70.3; Bp=34.7; SD=31.4;Bd=35.6. Literature Cited Allen, J.A. 1876. The American Bisons, Living and Extinct. Memoirs of the Museum of Comparative Zoology at Harvard College, 4( 10): 246 pages. Bamforth, D.B. 1988. Ecology and Human Organization on the Great Plains. 216 pages. New York: Plenum Press. Clarke, A.H., Jr. 1981. The Freshwater Molluscs of Canada. 446 pages. Ottawa: National Museum of Canada. Cleland, C E . 1966. The Prehistoric Animal Ecology and Ethnozoology of the Upper Great Lakes Region. Anthropological Papers, University of Michi­ gan, Museum of Anthropology, 29: 294 pages. Driesch, A. von den 1976. A Guide to the Measurement of Animal Bones from Archaeological Sites. Bulletin of the Peabody Museum of Archaeology and Elhnol- Dykev, A 1987. Flerov, C 1971. Geist, V 1990. Geist, V, 1977. ogy, Harvard University, 1: x + 138 pages. .S., and V.K. Prest Paleogeography of Northern North America; 18,000-5000 Years Ago [Map 1703A, scale 1:12,500,000]. Ottawa: Geological Survey of Canada. C. The Evolution of Certain Mammals during the Late Cenozoic. In K.K. Turekian, editor, The Late Cenozoic Glacial Ages, pages 479-491. New Haven: Yale University Press. [Letter to the Editor.] Conservation Biology, 4(4):345-346. and P. Karsten The Wood Bison (Bison bison athabascae Rhoads) in Relation to Hypotheses on the Origin of the American Bison (Bison bison Lin­ naeus). Zeitschrift fiir Sdugelierkunde, 42:119-127. NUMBER 93 97 Gromova, V.I. 1935. Pervobtnyi zubr (Bison priscus Bojanus) v SSSR. Trudy Zoological Institute, Academy Nauk SSSR, 2:77-204. Guthrie, R.D. 1970. Bison Evolution and Zoogeography in North America during the Pleistocene. Quarterly Review of Biology, 45(1): 1-15. Harington, CR. 1984. Mammoths, Bison and Time in North America. In W.C. Mahaney, editor, Quaternary Dating Methods, pages 299-309. Amsterdam: Elsevier. Haynes, G. 1980. Prey Bones and Predators: Potential Ecologic Information from Analysis of Bone Sites. Ossa, 7:75-97. 1982. Utilization and Skeletal Disturbances of North American Prey Car­ casses. Arctic, 35:266-281. Henrichsen, P. 1981. Dental Anomalies in Muskoxen from Greenland. Videnskabelige Meddelelserfra Dansk Naturhistori.sk Forening, 143:113-123. 1982. Population Analysis of Muskoxen, Ovibos moschatus (Zimmer­ mann 1780), Based on Occurrence of Dental Anomalies. Sauge- tierkundliche Mitteilungen, 30(4):260-280. Hind, H.Y. 1859. Report on the Assiniboine and Saskatchewan Exploring Expedition of 1858. Toronto: The Legislative Assembly. Lott, D.F. 1979. Hair Display Loss in Mature Male American Bison: A Temperate Zone Adaptation? Zeitschrift fur Tierpsychologie, 49:71-76. McAndrews, J.H. 1982. Holocene Environment of a Fossil Bison from Kenora, Ontario. On­ tario Archaeology, 37:41-51. McDonald, J.N. 1981. North American Bison: Their Classification and Evolution. 315 pages. Berkeley: University of California Press. McDonald, J.N., CE. Ray, and CR. Harington 1991. Taxonomy and Zoogeography of the Musk Ox Genus Praeovibos Staudinger, 1908. In J.R. Purdue, W.E. Klippel, and B.W. Styles, ed­ itors, Beamers, Bobwhites, and Blue-Points: Tributes to the Career of Paul W. Parmalee. Illinois State Museum, Scientific Papers, 23: 285-314. Meagher, M. 1986. Bison bison. Mammalian Species, 266:1-8. Moore, L. 1975. The Presence of Bison Bones at Hungry Hall Site, Rainy River Northwest Ontario. [Abstract.] In Eighth Annual Meeting, 1975, Ca­ nadian Archaeological Association, Abstracts, page 24. Nowak, R.M., and J.L. Paradiso 1983. Walker's Mammals of the World. Fourth edition, 1362 pages. Balti­ more: Johns Hopkins University Press. Rhoads, S.N. 1897. Notes on Living and Extinct Species of North American Bovidae. Proceedings of the Academy of Natural Sciences of Philadelphia, 49:483-502. Richardson, J. 1829. Fauna Boreali-Americana; Or the Zoology of the Northern Parts of British America. 140 pages. London: John Murray. Schultz, C.B., and W.D. Frankforter 1946. The Geologic History of the Bison in the Great Plains: A Prelimi­ nary Report. Bulletin of the Nebraska State Museum, 3:1-10. Shay, CT. 1971. The Itasca Bison Kill Site: An Ecological Analysis. 133 pages. Saint Paul: Minnesota Historical Society. Sher, A.V. 1974. Pleistocene Mammals and Stratigraphy of the Far Northeast USSR and North America. International Geological Review, 16(7-10): 284 pages. [Original in Russian, published by Nauka Press, Mos­ cow, 1971, 310 pages.] 1986. Olyorian Land Mammal Age of Northeastern Siberia. Palaeonto­ graphica Italica, 74:97-112. Shipman, P. 1981. Life History of a Fossil: An Introduction to Taphonomy and Paleo­ ecology. 222 pages. Cambridge, Massachusetts: Harvard University Press. Skinner, M.F, and O.C Kaisen 1947. The Fossil Bison of Alaska and a Preliminary Revision of the Ge­ nus. Bulletin of the American Museum of Natural History, 89: 123-256. van Zyll de Jong, C.G. 1986. A Systematic Study of Recent Bison, with Particular Consideration of the Wood Bison (Bison bison athabascae Rhoads 1898). Publica­ tions in Natural Sciences, National Museum of Canada, National Museum of Natural Sciences, 6: 69 pages. van Zyll de Jong, C.G., C. Gates, H. Reynolds, and W. Olson 1995. Phenotypic Variation in Remnant Populations of North American Bison. Journal of Mammalogy, 76:391-^405. Voorhies, M.R. 1969. Taphonomy and Population Dynamics of the Early Pliocene Verte­ brate Fauna, Knox County, Nebraska. Contributions to Geology [University of Wyoming], Special Paper, 1: 69 pages. Webb, T., Ill, E.J. Cushing, and H.E. Wright, Jr. 1983. Holocene Changes in the Vegetation of the Midwest. In H.E. Wright, Jr., editor, Late-Quaternary Environments of the United States, 2: The Holocene, pages 142-165. Minneapolis: University of Minnesota Press. Wendland, W.M. 1978. Holocene Man in North America: The Ecological Setting and Cli­ matic Background. Plains Anthropologist, 23:273-287'. Wendland, W.M., A. Benn, and H.A. Semken, Jr. 1987. Evaluation of Climatic Changes on the North American Great Plains Determined from Faunal Evidence. In R.W. Graham, H.A. Semken, Jr., and M.A. Graham, editors, Late Quaternary Mammalian Bioge­ ography and Environments of the Great Plains and Prairies. Illinois State Museum, Scientific Papers, 22:460^172. Wilson, M.C. 1975. Holocene Fossil Bison from Wyoming and Adjacent Areas, xiv + 276 pages. Master's thesis, University of Wyoming, Laramie, Wyo­ ming. 1996. Late Quaternary Vertebrates and the Opening of the Ice-Free Corri­ dor, with special Reference to the Genus Bison. Quaternary Interna­ tional. 32:97-105. The Terrestrial Posture of Desmostylians Daryl P. Domning ABSTRACT An attempt to reconstruct a skeleton of Paleoparadoxia Rein­ hart, 1959 (Mammalia, Desmostylia), suggests that desmostylian terrestrial posture deviated from that of typical ungulates much less than has been supposed by other authors. Desmostylians prob­ ably had a quadrupedal stance, with the body well off the ground and the limbs more or less under the body; a strongly arched spine and steeply inclined pelvis; slightly abducted elbows and more strongly abducted knees; and a digitigrade foot posture with an extended but not hyperextended wrist and hyperextended toes, the front toes pointing anterolaterad and the hind toes pointing for­ ward. Most peculiarities of the skeleton have parallels in certain large, slow-moving terrestrial mammals, such as ground sloths and chalicotheres. The desmostylian skeleton was apparently well suited to supporting the body's weight on the hindquarters, per­ haps while the animal clambered slowly over very uneven ground. This most likely occurred while it foraged for marine algae or sea grasses in rocky intertidal areas of the North Pacific shoreline, and while it crossed these areas en route to and from the water. Loco­ motion in the water probably resembled that of polar bears, with alternate pectoral paddling as the principal means of propulsion and the hind limbs used for steering. Surprisingly, desmostylian- like features of the tibia and ankle also are found in many other primitive ungulates and deserve closer study. Introduction Before and even since the discovery that desmostylians were quadrupedal, apparently amphibious marine mammals rather than wholly aquatic sirenians, their outward appearance and, in particular, their terrestrial posture have been controversial (see the remarkable collection of artists' reconstructions compiled by Inuzuka (1982)). Not even the more recent studies, based upon complete postcranial skeletons, have led to a consensus on this latter issue (cf. Repenning, 1965; Shikama, 1966, 1968; Inuzuka, 1984, 1985; Halstead, 1985; Repenning and Packard, 1990). Daryl P. Domning, Research Associate, Department of Paleobiology, National Museum of Natural History, Smithsonian Institution, Wash­ ington, D.C. 20560-0121, and Laboratory of Evolutionary Biology, Department of Anatomy, College of Medicine, Howard University, Washington, D.C. 20059. In March 1986, I was asked to design a new mount of Pa­ leoparadoxia Reinhart, 1959, in a terrestrial pose for a planned exhibit gallery at the National Museum of Natural History, Smithsonian Institution. (The mount was never actually con­ structed because of a change in the exhibit plan.) The specimen to be mounted was a cast of the neotype skeleton of Paleopara­ doxia tabatai (Tokunaga, 1939), known in the literature as the Izumi specimen. Available for comparison were several other specimens and casts of various desmostylian genera and spe­ cies (see below). Because some of these (notably the Stanford specimen) have not been formally described, most original specimens were not then accessible, and time and resources were limited, a thorough morphological or biomechanical study of the desmostylian skeleton was infeasible. Features critical for mounting the skeleton, however, were sufficiently clear from the available material, and the observations seemed worth recording as a contribution to the ongoing debate. These conclusions were first presented at the Society of Ver­ tebrate Paleontology annual meeting in Philadelphia in 1986. Subsequent discussions and collaboration with N. Inuzuka and others (see Inuzuka et al., 1995) have encouraged further re­ finement of these observations, which are offered herein for fu­ ture testing as yet another alternative hypothesis on the fasci­ nating subject of desmostylian locomotion. ACKNOWLEDGMENTS.—I thank R.J. Emry and CE. Ray for stimulating this study by inviting me to design the proposed mount of Paleoparadoxia, D. Chaney for logistical help in the course of the work, and M. Cole, M.C. Coombs, and R.J. Emry for useful discussions. J.M. Clark, M. Cole, M.C. Coombs, N. Court, N. Inuzuka, CE. Ray, and CA. Repenning commented on various drafts of the manuscript or parts thereof. I take great pleasure in dedicating this paper to Clayton E. Ray, friend, model, and mentor throughout my professional career. ABBREVIATIONS.—The following museum abbreviations are used: PV National Science Museum, Tokyo UCMP University of California Museum of Paleontology, Berkeley USNM Collections of the National Museum of Natural History, Smith­ sonian Institution, Washington, D.C, which include the collec­ tions of the former United States National Museum 99 100 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY MATERIAL AND METHODS 1. Cast (USNM 26375) of nearly complete skeleton of sub- adult Paleoparadoxia tabatai (PV-05601; Izumi specimen) from Japan, partly described by Shikama (1966, 1968). 2. Cast (USNM 25899) of nearly complete skeleton of ma­ ture adult Paleoparadoxia sp. (UCMP 81302; Stanford speci­ men) from California. 3. Cast (USNM 24727) of nearly complete skeleton of adult Desmostylus Marsh, 1888 ("ZX mirabilis'" Nagao, 1935; Hok­ kaido University, Sapporo; Keton specimen), from Sakhalin, described by Shikama (1966) and Inuzuka (1980-1982). 4. Partial skeleton of immature Desmostylus (USNM 186891) from Oregon. Also in existence are two partial skeletons possibly referable to Behemotops Domning, Ray, and McKenna, 1986 (the Ashoro specimens; Inuzuka, 1989); a well-preserved skeleton of Desmostylus (the Utanobori specimen, described in part by Inuzuka (1988)); two partial skeletons of Paleoparadoxia tabatai from the Chichibu Basin in the Saitama Prefectural Museum, Japan (Sakamoto, 1983); and the anterior one-half of an immature skeleton of Paleoparadoxia weltoni Clark, 1991 (UCMP 114285). The last three specimens do not appear to show any biomechanically significant postcranial differences from the Izumi specimen. Behemotops, Paleoparadoxia, and Desmostylus seem to have differed little, if at all, in their loco­ motor specializations (Shikama, 1966; Inuzuka et al., 1995), so observations made herein apply to all three genera unless oth­ erwise stated. I reached my conclusions mainly by manipulating the casts of the Stanford specimen, the well-preserved skeleton of a ma­ ture adult of the largest known species of Paleoparadoxia. Its joint surfaces are accordingly both fully developed and adapted to bear the maximum stresses that would have occurred in any known desmostylian. This makes the Stanford specimen the most sensitive indicator of the biomechanical constraints on desmostylian locomotion. The conclusions derived from this individual were then applied to the Izumi specimen, which is smaller, immature (judging from degree of epiphyseal fusion), less well preserved, and somewhat distorted, and which appar­ ently represents a different species. No evidence was encoun­ tered, however, that would invalidate the assumption of biome­ chanical similarity. The skeleton was built literally from the ground up, by as­ sembling the feet in their most probable poses and then placing the limbs, girdles, and axial skeleton on top of them, with as lit­ tle deviation as possible from a "normal" terrestrial ungulate or subungulate stance. Recent reconstructions that differed greatly from such a stance were not satisfactory. For example, Repen- ning's (1965) reconstruction, which was reproduced by Romer (1966, fig. 367), Barnes et al. (1985, fig. 3a), Inuzuka (1984, pl. 9: fig. 2), and others, placed the animal in a frog-like squat­ ting pose with the dorsal surface of the manus against the ground. Inuzuka (1984, 1985) advocated a "herpetiform" pos­ ture with the limbs sprawled to the sides. In contrast, my work­ ing assumption, or null hypothesis, was that Paleoparadoxia did not differ essentially in posture from typical large land mammals such as hippopotamuses or rhinoceroses. (No single species of mammal, however, was used as a model.) When (as often happened) peculiarities of the joints demanded a depar­ ture from this assumption, other mammals provided analogies to demonstrate the mechanical feasibility of the posture. The aim of this study was to clarify the terrestrial posture of desmo­ stylians and not their aquatic adaptations, so these examples were sought among land mammals; however, this is not meant to imply that desmostylians were other than largely aquatic in habits and habitat. It is somewhat reassuring that this eclectic method, although applied separately to different parts of the skeleton, repeatedly led to comparisons with animals that share a similar habitus: heavy-bodied, presumably slow-moving quadrupedal forms re­ sembling ground sloths. Perhaps Paleoparadoxia was not a paradox after all. Results PES.—The key observation of this study was made on the hind foot, which has always posed the most vexing problems for desmostylian functional anatomy. The axis of the ankle joint inclines ventromedially about 30°^40° from normal to the long axis of the tibia, and it slopes ventrolaterally about 10° from the plantar plane of the foot. If the foot is placed planti­ grade on a horizontal surface, as it has been in some recon­ structions (see Inuzuka, 1982), the tibia is thereby inclined about 40°-50° from the sagittal plane. Together with the tor­ sion in the tibia itself (see below), this necessitates strong ab­ duction of the knee, and awkward positions of other joints, in order to assume the plantigrade pose during the weight-bearing portion of the stride. This pose, however, is certainly incorrect, as is shown by the form of the metatarsals and, especially, the metatarso-phalangeal joints. The pes as a whole is paddle-like; the tarsus-metatarsus is rigid and planar, rather than forming a half-cone as in an ele­ phant. From digit II to digit V the metatarsals increase signifi­ cantly in length, and the form of curvature of the phalangeal ar­ ticular surface varies (Figure la-d). This surface lies on the dorsal side of the bone at its distal end and is nearly flat on metatarsal II, corresponding to the equally flat proximal sur­ face of the proximal phalanx. The transition from this flat dor­ sal surface to the distal part of the articular surface is abrupt and much more strongly curved than the very slight proximal concavity of the phalanx. The latter can stably articulate with the metatarsal in either the maximally (about 75°) hyperex­ tended position (the more stable position; Figure la) or the typ­ ical extended position, i.e., in line with the metatarsal. The metatarsal's palmar joint surface is slightly keeled and articu­ lates with a pair of large sesamoids, so flexion of the phalanx onto this surface is impossible. The most likely terrestrial pose NUMBER 93 101 FIGURE 1.—Paleoparadoxia sp. (UCMP 81302), left pes: a-d, metatarsals and proximal phalanges of digits II-V, respectively, in medial views, each showing maximum degree of hyperextension possible; e, left pes in anterodorsal view. Note that the tibial-astragalar joint axis is horizontal, the metatarsals lie in a nearly vertical plane, the weight-bearing axis passes through digit III, and the toes are maximally hyperextended. of the second digit is therefore the hyperextended one, with the phalanx nearly perpendicular to the metatarsal (Figure \a). The same observations hold for the more lateral toes to a pro­ gressively lesser extent. The dorsal-to-distal curvature of the metatarsal surface gradually becomes less abrupt, and the prox­ imal articular concavity of the phalanx deepens. The degree of hyperextension required (or even possible) at these joints, after increasing slightly from digit II to III, correspondingly dimin­ ishes from about 80° on III to 60° on IV and 50° on V (Figure \b-d). This pattern becomes intelligible when the foot is placed in a digitigrade position, with the plane of the metatarsals at an an­ gle to the ground and the phalanges flat on the substrate (Figure \e). Then the short, almost cuboidal medial metatarsals (espe­ cially metatarsal III) support the body's weight on their flat dis­ tal ends and their sesamoids. The greater lengths of the more lateral metatarsals require them to extend distally and laterally at progressively greater angles to the sagittal plane. Conse­ quently they bear less weight; the end of metatarsal V may al­ most have been free of the ground, and not only is its digit the least hyperextended (Figure \d), but its proximal phalanx has a canted distal articulation, causing the toe to curve medially and making the distal phalanges lie parallel to those of the other digits (Figure \e). When the metatarsals and phalanges are in this pose, one finds that, despite the peculiar shapes of the tarsals and meta­ tarsals (1) the foot posture can be digitigrade during weight- bearing (i.e., when the hip is directly over the ankle) and would be within usual limits for a large land mammal; (2) both the axis of the ankle joint and the plane of the sole are nearly per­ pendicular to the sagittal plane; (3) the toes point forward; and (4) the axis of the ankle joint is horizontal (Figure le), so the ankle joint is not prone to dislocation by the body's weight. The calcaneal tuberosity does have a strong medial inclination (Figure le), but this can be seen as an adjustment to the torsion of the tibia (see below) and of the muscles of plantar flexion. A line from the apex of the calcaneal tuberosity perpendicular to the axis of the tibial-astragalar joint passes approximately through digit III, and therefore probably along the central weight-bearing axis of the foot. If the animal were to take a long stride, the hind foot might initially be placed in a planti­ grade pose, becoming digitigrade only as the hip moved above it. Such a bulky animal as a desmostylian may have habitually taken only short strides, however, throughout which the foot could have been digitigrade or nearly so. On the whole, in view of its strange morphology, the posture of the hind foot is surprisingly normal. An elastic pad, such as exists in elephants, may have supported the foot in this digiti­ grade posture (although in elephants the foot bones surround the pad with a half-conical structure that is not seen in desmo­ stylians, and the posture of the digits themselves is actually un- guligrade; Osborn, 1942:1336). Alternatively, desmostylians may have managed without such a supporting pad because they were smaller than elephants, probably spent more time sup­ ported by water, and when hauled out probably took weight off their feet by squatting or lying on the ground as do sea lions. The modern Hippopotamus, for example, maintains a digiti­ grade posture on land without an elephant-like supporting pad in the foot. TIBIA.—The tibia has two major peculiarities: lateral torsion of the proximal end about 45°-50° relative to the distal end, and mediodistal inclination of the distal joint surface about 30°-40° from normal to the long axis of the shaft. If the shaft is held vertically with its anteroposteriorly flattened distal portion perpendicular to a sagittal plane, the knee and femur must be strongly abducted and the sole of the foot must face both dis­ tally and laterally. Conversely, with the foot in the digitigrade position described above, the tibia is inclined outward from the ankle about 30°^40° from a sagittal plane. The abduction of the knee largely compensates for this, however, so that the ankle lies not far outboard of the parasagittal plane of the hip joint (Figure 2). 102 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY FIGURE 2.—Paleoparadoxia tabatai (Izumi specimen, cast, USNM 26375), skeleton partly assembled in sand­ box in posterolateral view to show limb positions and spinal curvature postulated in this study. The three repre­ sentative ribs shown are not in correct anatomical position; their ventral ends should be abducted much farther from the midline. NUMBER 93 103 FIGURE 3.—Paleoparadoxia tabatai (Izumi specimen, PV-05601), reconstructed skeleton in lateral view, show­ ing probable terrestrial posture. Note hyperextension and anterolateral direction of front toes, anterior direction of hind toes, and strong abduction of knees. FEMUR.—The shaft of the femur is broad and greatly flat­ tened anteroposteriorly. The neck protrudes anteromedially from the plane of flattening of the shaft. The head rises slightly above the level of the greater trochanter and lacks a fovea capi­ tis. A line connecting the proximal and distal margins of the ar­ ticular surface of the head forms an angle of about 50°-65° with the femoral shaft. With the foot bones in the position de­ scribed above, with the tibia inclined outward 30°^0° from a sagittal plane and lying in a transverse plane inclined perpen­ dicular to the sole of the foot, and with the knee joint bent at a right angle, the posterior (now ventral) edge of the femoral head lies barely inside the acetabulum; a 2-3 cm wide strip of the head's articular surface is exposed dorsally; and although the femur is laterally rotated, there remains clearance between the greater trochanter and the ischium (Figures 2, 3). This leg posture would seem to provide adequately for support and movement of the hindquarters. The medial condyle is some­ what larger than the lateral in the Izumi specimen and extends farther distally and posteriorly; in the Stanford specimen, prob­ ably representing a different species, they are more nearly equal in development. Court (1994:324) stated that "[i]n me­ dial view, a line drawn from the posterior-most extent of the in­ ner condyle to its anterior margin, when orientated in the hori­ zontal plane, reflects the position of maximum articular congruence at the knee and the likely habitual posture of the fe­ mur." By this criterion, the femur of Paleoparadoxia would have habitually inclined about 67°-69° from the vertical. The crural index (lengths of tibia/femur x 100) is 88 in the Izumi and 77 in the Stanford specimen. The ratios of lengths of meta­ tarsal Ill/femur* 100 are 19 and 15, respectively. These differ­ ences might be ontogenetic, interspecific, or both. INNOMINATE.—As others have noted, the innominate is primitive looking, even rodent-like, being very long and nar­ row for such a large mammal. The ilium in particular lacks the degree of lateral expansion typical of heavy ungulates. The ver­ tical angle between innominate and sacrum is slight (about 10°-15°), as in primitive mammals, and, given the length of the ilium, would seem to be disadvantageous for weight-bearing. The strong spinal curvature (see below), however, makes possi­ ble (indeed, necessitates) a very steep inclination of both sacrum and pelvis (Figures 2, 3). This accomplishes the same purpose as the short, almost vertical ilium of most other gravi- portal mammals, namely, to bring the hip joint nearly under­ neath the sacroiliac joint. In caudal view, the posterior part of the pelvis has a pronounced U-shape, with the pubes forming an almost flat floor of the pelvic canal and the ischia lying in almost parallel, parasagittal planes (cf. Inuzuka, 1985, fig. 1). The acetabulum is relatively wide and shallow. This is reflected in the extremely acute angle seen in ventral view between its superolateral and ventrolateral margins (cf. Jenkins and Ca- mazine, 1977, fig. 7), which indicates that the superolateral margin does not extend very far laterally. Also, a line connect­ ing the medial and lateral margins of the upper, weight-bearing articular surface of the acetabulum forms an angle of about 70°-75° with the vertical, as nearly could be estimated for the disarticulated Izumi and Stanford skeletons. VERTEBRAL COLUMN.—The desmostylian neck is short for an ungulate, and, although the head is fairly large, the anterior thoracic neural spines are normally inclined and not unusually 104 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY enlarged, so the nuchal ligament was evidently not relied upon much for cranial support. The cervical and anterior thoracic centra, largely missing from the Izumi specimen but restored by its preparators, give a distinct secondary (concave-dorsal) curvature to this region. Fifteen thoracic and six lumbar vertebrae are present in the restored Izumi skeleton (although Shikama (1966) mentioned only 14 thoracics); the last 13 thoracics are preserved in the Stanford specimen. The posterior thoracics and lumbars have short, vertical spines, and there is no anticlinal vertebra. The fourth thoracic from the rear (?T12) in the Stanford specimen comes closest to being a diaphragmatic vertebra; the zygapo- physeal articular planes shift from near-horizontal to near-ver­ tical in the region of ?T11-13 but then return to an intermediate inclination in the lumbar region. This shift suggests more verti­ cal flexibility in the posterior thorax than in the adjacent re­ gions. The most-posterior thoracic and most-anterior lumbar centra are distinctly wedge shaped, thicker dorsally, and thinner ven­ trally in both the Stanford and the Izumi Paleoparadoxia spec­ imens (this, however, is not very evident in Desmostylus). This shape indicates a strong dorsal convexity of the spine in this re­ gion, with the summit of the backbone lying near the rear of the ribcage. From this point the lumbar and sacral vertebrae de­ scend at a rather steep angle, with the pelvis in an almost verti­ cal position (Figures 2, 3). The sacrum is fairly long and robust but is not fused to the pelvis, and sacral vertebrae 1 and 2 are not fused to one another, even in the Stanford adult (although sacrals 2-5 are fused). MANUS.—The metacarpals are about twice as long as the metatarsals. The metacarpo-phalangeal articulations are fairly uniform in shape and do not show a mediolateral gradient of change analogous to that seen in the hind foot. The long axis of the lunar articular surface lies at approximately a 65° angle to the plane of the palm. With the metacarpals vertical, the radius can easily be balanced upright on the lunar without any other support and with its shaft tilted backward some 15° from the vertical (Figure 3). The large radial styloid process thus would not have prevented wrist extension sufficient for digitigrade posture. In this position, the axis of the elbow joint is at about a 75° angle to the plane of the palm. Even with the elbow ab­ ducted 30° from the body, the front toes could not have pointed straight forward; at best the palm might have been "pronated" some 45° from a parasagittal plane when the animal stood on land. (True pronation and supination, however, were prevented by the firm articulation of the radius and ulna; this angle of "pronation" was controlled instead by movement at the shoul­ der joint.) The weight of the forequarters therefore appears to have been carried on metacarpals held essentially vertically, with the foot in a digitigrade stance; the toes presumably were hyperextended and pointed as much laterad as anterad, and the wrist joint was extended straight. From this position the wrist could have been flexed, but the very large styloid process of the radius would have prevented any significant hyperextension. RADIUS AND ULNA.—As mentioned above, the radius and ulna are tightly articulated, although not fused, and the radius bears a massive styloid process that overlaps the anterior side of the wrist joint. The radius forms almost the entire distal part of the semilunar notch and is, therefore, the main weight-bear­ ing bone. The ulna also bears considerable weight, however, and has an even thicker shaft than the radius. The olecranon is very long, equal to one-half the length of the radius (excluding the styloid process) in the Izumi specimen and to about two- thirds of the length of the radius in the Stanford specimen. HUMERUS.—The humerus is long and slender in the Izumi specimen. The trochlea is deeply constricted at its middle, and a corresponding ridge on both radial and ulnar parts of the semilunar notch renders the elbow joint very stable. The notch, however, opens more anteriorly than proximally, so the elbow could not have been extended as straight as in graviportal forms such as elephants. The entepicondyle is short and stout, protruding more posterad than mediad. The trochlea is canted only about 80° to the long axis of the shaft, so the shafts of the humerus and radius are only 10° out of alignment when the el­ bow is extended. This alignment indicates a very slight degree of abduction of the elbow and nearly parasagittal forelimb movement in terrestrial locomotion (Jenkins, 1973). The supi­ nator crest shows little or no expansion, but there is a long and fairly prominent deltopectoral crest reaching almost to the dis­ tal end of the shaft. The lesser tuberosity is weakly developed in the immature Izumi specimen, but the greater tuberosity is robust. The head is expanded anteromedially. The brachial in­ dex (lengths of radius/humerus x 100) is 62 in the Izumi skele­ ton; the intermembral index (lengths of (humerus + radius)/(fe- mur + tibia) x 100) is 88; the ratio of lengths of metacarpal III/ humerus x 100 is 41; and the ratio of lengths of metacarpal III/ radius x 100 is 65 (64 in the Stanford skeleton). SCAPULA.—The distal end of the scapula is not preserved in the Stanford specimen, and that of the Izumi specimen has the coracoid process detached; on the cast it has been restored as extending rather far anteriorly and distally. Although this pre­ vents the humerus from being extended straight downward from the glenoid, such extension almost certainly would not have taken place. At maximum extension of the shoulder, the humerus would have made perhaps a 140° angle with the spine of the scapula. In a normal standing pose the angle would have been smaller, perhaps 120°-130° (Figure 3). Comparisons PES.—The desmostylian hind foot has been thought to be completely without parallel, but it shows surprising resem­ blances to the pes of the perissodactyl Chalicotherium (Figure 4; Zapfe, 1979): the latter is digitigrade, the metatarsals in­ crease markedly in length from medial to lateral, the calcaneal tuberosity has a strong medial inclination, and the toes point forward despite the strong abduction of the knee (see below). There are also differences from Paleoparadoxia: the axis of the NUMBER 93 105 FIGURE 4.—Chalicotherium grande, left pes in anterior view, after Zapfe (1979). astragalus is sloping rather than horizontal, and of course the feet have only three toes, which bear claws rather than the small hooves that desmostylians presumably possessed. None­ theless, we see in Chalicotherium a large quadruped of unques­ tioned terrestrial habits that bore its weight mainly on the me­ dial side of a hind foot that resembled in obvious ways the foot of Paleoparadoxia. The notoungulate Homalodotherium also has longer lateral than medial metatarsals, but the medial ones are thin rather than stout, and the animal evidently walked on the lateral rather than medial edge of its foot (Scott, 1930). TlBIA.—Again, Chalicotherium approaches Paleoparadoxia in the torsion of the tibial shaft, although not in the tilt of the ankle joint. Even closer matches to Paleoparadoxia are found in the giant armadillo (Priodontes), the collared anteater (Tamandua), pangolins (Manis), various bears (including Hel- arctos and Ursus spelaeus RosenmuTler and Heinroth), and the giant panda (Ailuropoda). Indeed, one giant panda (USNM 258835, a wild-caught specimen) is identical to the Stanford Paleoparadoxia in values of both variables (Domning, 2001). These two features (and especially the tilt of the ankle joint) also are seen to different degrees in a variety of other, extinct forms, including the most primitive sirenians (Prorastomidae; Domning, 2001) and proboscideans (Numidotherium; Court, 1994, and below), ground sloths (Megalonyx), Coryphodon, and "condylarths," including Hyopsodus (Gazin, 1968) and Meniscotherium (Gazin, 1965; Williamson and Lucas, 1992; see also Domning, 2001). Hyopsodus and Meniscotherium be­ long to the group of primitive ungulates from which tethytheres or pantomesaxonians emerged (Thewissen and Domning, 1992), so these features of the tibia may well have been inher­ ited from the ancestors of desmostylians. Their function is un­ clear, and their taxonomic distribution (or indeed, their very ex­ istence) within this group has attracted negligible attention. Meniscotherium (which exhibits much less tibial torsion but an even greater tilt of the ankle joint than Paleoparadoxia) has, however, been characterized as "adapted for ambulatory, habit­ ual travel on an irregular terrestrial substrate" and as suited to negotiating "a rugged and obstacle-filled terrain" (Williamson and Lucas, 1992:32-33), as was also probably the case for des­ mostylians (see below). The desmostylians' short tibia, relative to the femur, is characteristic of graviportal quadrupeds in gen­ eral as well as the above-mentioned "condylarths." PATELLA.—Coombs (1983:38) suggested that a large patella (indicative of powerful knee extensors) was associated with body erection in Chalicotherium; this bone is proportionately much larger in the Stanford Paleoparadoxia than in chalico- theres. FEMUR.—The flattened desmostylian femur is strikingly reminiscent of that of large ground sloths, but some flattening of the shaft is seen in most large land animals, including Chali­ cotherium and Homalodotherium as well as elephants, rhinoc­ eroses, and sauropod dinosaurs. (This flattening is more pro­ nounced in sauropods thought to have routinely used a bipedal posture; Wilson and Carrano, 1999.) The femur of Paleopara­ doxia, like that of elephants and ground sloths, lacks a fovea capitis. The posture suggested herein for Paleoparadoxia, with the knees strongly abducted from the body, is likewise not un­ precedented among large mammals, including the largest ground sloths and Chalicotherium. Bears adopt a similar pose when supporting themselves on their hind legs. Jenkins and Ca- mazine (1977) described the form and kinematics of the hip joint in three other carnivores (raccoon, cat, and fox) that show progressively lesser degrees of hip abduction. The angle be­ tween the margin of the articular surface of the head and the femoral shaft is greater in desmostylians than in raccoons, in which it is about 50° (Jenkins and Camazine, 1977:363), sug­ gesting a greater degree of femoral abduction in the former. When desmostylian bones are manipulated in the fashion advo­ cated herein, the portions of the femoral articular surface en­ closed within the acetabulum at each stage of the stride corre­ spond well with the observations of Jenkins and Camazine (1977:359-360) on the raccoon, cat, and fox, indicating that the 106 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY present reconstruction does not involve a pattern of hip articu­ lation unusual for terrestrial mammals. INNOMINATE.—In most large terrestrial mammals, including Chalicotherium, Homalodotherium, and ground sloths, the il­ ium is much broader than in desmostylians. This broadening provides attachment area for enlarged gluteal muscles, which extend the hip against gravity. The absence of this graviportal adaptation in desmostylians may be because of their more am­ phibious habits; in any case, it does not rule out body erection. The long innominate and relatively narrow ilium of Paleopara­ doxia, as noted above, resemble those of rodents, but they also strongly resemble those of the giant panda, Ailuropoda (see Davis, 1964). Desmostylians are probably the largest mammals to display this form of pelvis. Davis (1964:110) listed seven features that he believed to characterize the pelvis "in mam­ mals in which forces parallel to the long axis of the pelvis pre­ dominate, i.e., those that stand erect and those that use their hind legs for bracing while digging." Of these seven features (which do not include a broadened ilium), at least five are seen in desmostylians: the wings of the ilia tend to be shifted into the frontal plane; the corpus of the ilium is nearly circular in cross section; the pubo-ischiadic symphysis is reduced in its anterior part (i.e., the front of the symphysis lies posterior to the acetabulum); the number of sacral vertebrae is relatively high (five); and the tail is shortened. (Although Davis (1964:113) pointed out that Ailuropoda exhibits these features "to a far greater degree than any other carnivore," he nonethe­ less believed that pandas do not stand erect more than bears and that the contrast in pelvic shape between these forms is due to nonadaptive factors. In any case, both bears and pandas fre­ quently assume a bipedal posture, in contrast to most mam­ mals.) Desmostylians also lack the noticeably enlarged ischial tuberosity that is typically seen in large ungulates, including Chalicotherium and Homalodotherium. In Chalicotherium, however, these tuberosities were thought by Zapfe (1979) to re­ flect the use of a sitting posture during feeding. The U-shape of the desmostylian pelvis in caudal view is outside the range of variation of the carnivores studied by Jenkins and Camazine (1977:368, fig. 7), but they generalized that "the ischial surface is approximately perpendicular to the middle of the range of abduction." The nearly parasagittal ischial surface in desmo­ stylians implies very pronounced abduction at the hip, as is agreed by most authors. The relatively shallow, open desmo­ stylian acetabulum and the inclination of its upper articular sur­ face likewise tend toward the end of the raccoon-to-fox spec­ trum that indicates relatively great hip abduction (Jenkins and Camazine, 1977:361-363, figs. 6, 7). VERTEBRAL COLUMN AND BODY FORM.—The strongly curved backbone of Paleoparadoxia was explained above as an adjustment to supporting a heavy body by way of an elongate pelvis. A similar degree of spinal curvature and verticality of the pelvis can be observed in rodents as large as the giant bea­ ver Castoroides (see Romer, 1966, fig. 442), which of course is still much smaller than Paleoparadoxia. A somewhat similar overall body form (albeit with different osteological details) is seen in the giant panda, however (Davis, 1964, fig. 15), and even in some very heavy mammals, such as the notoungulate Toxodon and the edentate Glyptodon (Romer, 1966, figs. 374, 427). The last two forms share a bulky body, low forequarters, a high rump, and (as in nearly all graviportal mammals) a nearly vertical ilium, although they lack a pronounced lumbar curvature. Spinal curvatures more similar to that of Paleopara­ doxia, as well as more equal limb lengths, occur in the sloth Hapalops and the marsupial Diprotodon (see Coombs, 1983, figs. 5a, 7e). Chalicotherium and Homalodotherium, on the other hand, have long front legs, low hindquarters, and straighter spines, although the schizotheriine chalicothere Moropus shows a distinct dorsal convexity of the spine at the front of the lumbar region (Coombs, 1983, fig. 7a). MANUS.—Several large land mammals have longer metacar­ pals than metatarsals, including Chalicotherium and Homa­ lodotherium. The latter used a digitigrade stance of the fore­ foot, but Chalicotherium, like anteaters, walked on the dorsal surfaces of the flexed and heavily clawed toes, with the palm in a parasagittal plane (an orientation also seen, to a lesser extent, in Paleoparadoxia). RADIUS AND ULNA.—A strong radial styloid process (albeit not so strong as in Paleoparadoxia) is found in Chalicothe­ rium; Toxodon; the smaller toxodont, Nesodon (Scott, 1912); and the polar bear, Thalarctos (as well as other ursids), in com­ bination with digitigrade or even (in ursids) plantigrade pos­ ture. The olecranon is short in Chalicotherium and Homalodo­ therium; a closer match to that of Paleoparadoxia is seen in Toxodon, Nesodon, and Glyptodon, which used the forelimbs for support and not for pulling down tree limbs (see below). HUMERUS.—The humerus of Chalicotherium is long, but with a more markedly tilted trochlea and a more expanded dis­ tal end than in Paleoparadoxia. Sloths, bears, and Nesodon also have relatively long and slender humeri. STERNUM.—Chalicotherium shows a slight broadening and dorsoventral flattening of the sternebrae, but the double row of broad, flat plates forming the desmostylian sternum (Shikama, 1966) is unique. It is worth noting, however, that in large ground sloths, such as Paramylodon and Eremotherium, the costal cartilages are ossified. This arrangement and the desmo­ stylian sternum both could be ways of providing a solid area of origin for massive pectoral muscles, used in ground sloths for pulling down branches and in desmostylians for swimming. This notion unfortunately is not readily testable, for such mus­ cles generally have fleshy proximal attachments that leave no evidence in the form of distinct scars on the bones. Discussion TERRESTRIAL LOCOMOTION.—The foregoing is not intended to be an exhaustive catalog of the desmostylian-like features found among terrestrial mammals. Its purpose is merely to show that nothing in the present reconstruction is wholly un- NUMBER 93 107 - > FIGURE 5.—Reconstruction of Paleoparadoxia, showing postures that it probably used while standing or travel­ ing in the intertidal zone of North Pacific shorelines. precedented among large land mammals. On the contrary, Pa­ leoparadoxia presents a coherent suite of features that repeat­ edly appear in mammals believed to share a particular range of lifestyles: large, slow-moving herbivores that browsed (often bipedally) on leafy vegetation, in many cases by pulling it down from overhanging branches. These forms include ground sloths in addition to chalicotheres and Homalodotherium, both of which have been compared with sloths as well as with each other. I do not suggest that desmostylians browsed on trees or even used their forelimbs to gather food. Apart from their marine habitat, their lack of claws and their non-elongated forelimbs, among other features, clearly rule this out. The sloth-like forms, however, are believed to have shared a less obvious trait, namely, the habit of supporting themselves on their hind legs while feeding. Large sloths had massive tails to help sup­ port them in this pose, but even the practically tailless Chalico­ therium is thought to have behaved similarly (Zapfe, 1979). The essential requirement is a broad, stable base formed by the hind appendages. These forms all share an abducted posture of the knees (which steadies the upright body against falling side­ ways) and striking modifications of the hind feet—to support weight on the lateral edge of the foot in sloths and Homalodot­ herium, and on the medial side in Chalicotherium and desmo­ stylians. The similarities in metatarsal, ankle, and tibial struc­ ture between the latter two seem too striking to reflect wholly different selective pressures. Coombs (1983, figs. 3, 4, table 4) summarized in more detail the osteological features that appear to characterize diggers and bipedal browsers. The majority of these are features of the fore­ limbs, and judged on these alone, Paleoparadoxia resembles a digger more than a bipedal browser. This similarity, however, is surely due to its use of the forelimb for digging-like move­ ments while swimming, because the front feet are not at all suited to efficient digging or food-gathering. Most features of the desmostylian hindlimb, in contrast, are consistent with the pattern Coombs described for semi-bipedal forms. These fea­ tures—most of which also are well exemplified by humans— include a vertical ilium, a large acetabulum, a heavy femur, dif­ fering stance of manus and pes, a long calcaneal tuber, short metatarsals with rigid articulations, and splayed hind toes. Al­ though Coombs's comparisons were designed to distinguish bi­ pedal browsers from diggers and climbers rather than from swimmers or amphibious quadrupeds, and although they cov­ ered such a diversity of mammals that they are necessarily very generalized, these resemblances nonetheless show that the present interpretation of desmostylians is not without anatomi­ cal precedent. Consequently, Paleoparadoxia likely was a slow, heavy, quadrupedal herbivore that often had to support much of its weight on its hindquarters when climbing over extremely un­ even, rocky, slippery ground. This would have been especially true while it fed in the North Pacific intertidal zone, presum­ ably on marine algae and sea grasses (see Domning et al., 1986:47-48; Figure 5), and while it traveled to and from the 108 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY water. Such a scenario fits well with Repenning and Packard's (1990) explanation for the bilateral hind-limb fractures seen in the Stanford skeleton, which also envisions a rocky seashore habitat. Although it is possible that the peculiarities of the hind limbs were instead adaptations for swimming (see below), it is not necessary to look beyond the realm of purely terrestrial mam­ mals to find detailed resemblances to the limbs of desmostyl­ ians. The latter, once envisioned as an aberrant group of sire­ nians or "sea cows," might more accurately be thought of as "sea sloths." (Interestingly, on the Pacific coast of Peru, where desmostylians apparently never occurred, Thalassocnus natans Muizon and McDonald, 1995, a newly discovered genus and species of megalonychid ground sloth, apparently filled an eco­ logical niche much like that suggested herein for desmostyl­ ians.) These parallels with semi-bipedal terrestrial browsers have been drawn from a diverse group of land mammals. The test, then, is whether the salient features of desmostylian osteology really fit together coherently. The hind feet and legs would pro­ vide stable support for most or all of the body's weight during slow, deliberate movements over rough ground. The almost vertical pelvis is appropriate for four-footed weight-bearing. The strongly arched back, although well suited for supporting a heavy trunk, is flexible enough for raising the forequarters without greatly changing the slope of the lumbosacral region. In quadrupedal stance, the base of the short neck is carried rela­ tively low to the ground, which could easily be reached by the elongated rostrum. The scapula and forelimbs are in a normal ungulate pose, and the manus is peculiar only in its "semi-pro- nated" orientation. Although its gait was doubtless slow, noth­ ing would seem to have prevented the animal from traveling with ease across dry, level ground. A further and more rigorous test is to examine the moments of resistance against bending of the vertebral centra. Slijper (1946) and Coombs (1983) calculated these quantities by mul­ tiplying the width (b) of the caudal surface of the centrum by the square of the height (h2). Calculated moments for each ver­ tebra were graphed, as in Figure 6. A curve peaking in the lum­ bar region (Figure 6a) was interpreted to indicate use of an erect or semi-erect posture (cf. Coombs, 1983, fig. 2). Humans and bears (including Ailuropoda, Figure 6b) display this pat­ tern. The results for the Stanford Paleoparadoxia (Figure 6c) agree for the most part with Slijper's theoretical curve for qua­ drupedal mammals (Figure 6d), but they also show bipedal ten­ dencies in that the highest moments seen in the preserved por­ tion of the vertebral column are relatively posterior, lying mostly in the lumbar region. Among the seven forms compared by Coombs (1983, fig. 2), this combination of a peak at the front of the thorax and one in the lumbar region is most closely matched by the chalicothere Moropus, the gerenuk (Litocra- nius), which often feeds in a bipedal pose (Figure 6e), and a goat (Capra) that was born without forelimbs and learned to walk bipedally. A normal goat did not show any lumbar or pos- brV a theoretical bipedal A i l u r o p o d a —I 7 - 6 - IT) 2 5- x 4 - 3 - P a l e o p a r a d o x i a theoretical quadrupedal 1Th 1L L i tocranius FIGURE 6.—a-d, Graphs of moments of resistance (bh2; b=greatest caudal width of centrum in mm; h=greatest caudal height) against bending of verte­ bral centra, redrawn after Slijper (1946) and Coombs (1983): a, Slijper's theo­ retical curve for mammals with erect or semi-erect posture; b, curve for giant panda (Ailuropoda); c, curve for Paleoparadoxia sp. (Stanford specimen, UCMP 81302); d, Slijper's theoretical curve for quadrupedal mammals, e, Comparable curve for gerenuk (Litocranius), from Coombs (1983) after Rich- ter; y-axis=caudal surface of centra, in mm2. For all curves, the x axis repre­ sents the vertebral series, with ITh indicating first thoracic and iL indicating first lumbar vertebra Where scales are omitted, the y-axis just indicates lesser to greater. terior thoracic peak. Although this evidence seems to support the concept of desmostylians as quadrupeds with semi-bipedal habits, it cannot be considered conclusive because patterns somewhat resembling that of Paleoparadoxia also are seen in Hippopotamus and even in horses and cows (Slijper, 1946, ta­ ble 2). This body of data clearly needs to be analyzed in more detail. Yet another possible test is to compare the limb and joint pro­ portions of desmostylians with those of other large ungulates NUMBER 93 109 and subungulates, in order to find phenetic resemblances that give clues to locomotor habits. An ongoing study will use mul­ tivariate ordination of dissimilarity (M. Cole, unpub. data) to assess patterns of overall similarity among taxa based upon rel­ ative joint size and limb shape. The study incorporates numer­ ous osteological measurements from a variety of extinct and living taxa: Desmostylus, Paleoparadoxia, Elephas, Lox- odonta, Tapirus, Moropus, Dicerorhinus, Rhinoceros, Bos, Hippopotamus, Choeropsis, Eremotherium, and Paramylodon. Preliminary results of this principal coordinates analysis (M. Cole, pers. comm., 1999) show the two desmostylians linked together, then to the two ground sloths and to the pygmy hippo­ potamus and hippopotamus. The remaining taxa, in turn, are linked to the two hippos. Although the functional reasons for these linkages have yet to be interpreted, and the possible role of facultative bipedality remains unclear, it is interesting that the closest resemblances of desmostylians are with hippos on the one hand and ground sloths on the other. Although desmo­ stylians have often been compared with hippos, no one has heretofore seen in desmostylians any resemblance to sloths. This independent result tends to corroborate my conclusion based upon the comparisons offered above. AQUATIC LOCOMOTION.—Although Paleoparadoxia, like other desmostylians, seems to have been capable of fully ter­ restrial locomotion, it displays some features that seem best ex­ plained by, or at least consistent with, a partially aquatic mode of life. These include the retracted external nares, raised orbits, peculiar sternum, long, robust olecranon process, and unusual "semi-pronated" position of the manus. The toes are not obvi­ ously adapted for digging, so the powerful forelimb must have had other uses. Toxodon shows that weight-bearing alone might explain the large olecranon; this and large pectoral muscles also would have been of use in clambering over rocky ground. Nonetheless, the combination of characters cited also makes sense in the context of aquatic locomotion. Nothing in the des­ mostylian skeleton demands an aquatic interpretation, how­ ever, and aquatic habits might not have been postulated were desmostylian remains not found exclusively in marine deposits. Repenning (in Shikama, 1966:145; Repenning and Packard, 1990, fig. 183) and Shikama (1966:148, fig. 114) visualized Paleoparadoxia as swimming with the broad, somewhat pad­ dle-like forefeet while using the hind limbs as rudders. This swimming method (alternate pectoral paddling; Fish, 1996) is essentially the one used by polar bears (Thalarctos), which (al­ though they do not possess a greatly enlarged olecranon or en­ larged pectoral muscle attachments) have limb proportions similar to those of desmostylians (intermembral index of Thal- arctos=87), are fully capable of both aquatic and terrestrial lo­ comotion, and nonetheless show no obvious aquatic adapta­ tions in the skeleton. This swimming method is entirely consis­ tent with the comparatively short hind legs and metatarsals and powerfully developed forelimbs of a desmostylian, and it readily explains the features mentioned above. With the elbow strongly abducted, the manus could have been effectively re­ tracted underneath the trunk in a sweeping motion with the palm facing posteromediad, by means of movement at the shoulder permitted by the anteromedial expansion of the hu­ meral head. Large pectoral muscles arising from the broad ster­ num and inserting on the long deltopectoral crest would have played an important role in this. With the elbow adducted against the ribcage, the manus, held almost parasagittal^ as in sirenians and other marine mammals, could have been feath­ ered for the recovery stroke. The deeply grooved elbow joint would have resisted dislocation during the power stroke, with the huge olecranon enabling forceful extension of the elbow at the end of the stroke. The large radial styloid process would si­ multaneously have resisted hyperextension of the wrist. With their dorsoventrally flattened digits, whose lateral processes at the distal ends of the phalanges may have supported webbing (Figure le; Repenning and Packard, 1990), both front and hind feet would have been effective paddles. Shikama (1966) described the desmostylians' probable mode of locomotion when feeding on the seafloor. The procumbent tusks would have been used for uprooting and raking in aquatic vegetation; the hind limbs, with the knees adducted and the soles directed posterolaterad against the substrate, would have driven the animal forward; and the forelimbs would have steadied and helped steer the body. The posterolaterad-directed hind feet might be likened to the anti-recoil spades of a split-trail field ar­ tillery carriage. The overall posture would have somewhat re­ sembled that of a bottom-feeding walrus, although the hind limbs would have propelled the animal in a different manner. Critique of Alternative Reconstructions INUZUKA'S RECONSTRUCTION.—Inuzuka's "herpetiform mammal" reconstruction has been described several times in both Japanese and English, most recently and succinctly in Inu­ zuka et al. (1995). He defended his interpretation mainly on the basis of anatomical arguments, although the splayed-limb post­ mortem position of the Utanobori skeleton was also a factor (pos­ sibly the dominant one) in his original formulation of the idea. Inuzuka postulated a highly unusual, horizontal orientation of the scapula, based only upon its position in the partly disar­ ticulated, fossilized skeleton—a questionable source of infor­ mation at best, and certainly not decisive. The broad sternum may well have served for the origin of large pectoral muscles, but a herpetiform stance is not needed to explain such muscles, as they also would be important in swimming. As evidence for a laterally extending humerus, he cites the semi-pronated posi­ tion of the mutually immobile radius and ulna, on the assump­ tion that the front toes would have had to point forward. As noted above, however, this latter constraint is sometimes re­ laxed—as in Chalicotherium—and it then becomes possible to bring the elbow and front feet much closer to the midline than Inuzuka shows. As for the hind limb, the evidence Inuzuka cited for an ab­ ducted femur and strong adductor and quadriceps musculature 110 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY is equally consistent with a sloth-like pose, as is the inclination of the calcaneal tuber when the tibial torsion is taken into ac­ count. His reconstruction of the entire hind limb, including a digitigrade pes with the toes pointing forward, is basically in agreement with the interpretation presented herein for a semi- bipedal stance, although a normal walking stance would proba­ bly involve more extended hip and knee joints and less separa­ tion of the hind feet. The pelvis, however, should be more steeply inclined to accommodate a stronger dorsal convexity of the spine than he shows (e.g., Inuzuka, 1984, pl. 11: fig. 2). This also would bring the base of the neck closer to the ground. In short, no feature of the skeleton demands a herpetiform posture; all features are at least equally explicable in the con­ text of a sloth-like restoration, which has the major advantage for weight-bearing of keeping the feet (especially the front feet) much closer to the midline. Also relevant to this discussion is Numidotherium, possibly the most prmitive proboscidean (Court, 1994). Given that pro­ boscideans have been proposed as the sister group of desmo­ stylians (Domning et al., 1986), Court's (1994) conclusion that Numidotherium had a semisprawling posture might support In- uzuka's reconstruction. The resemblances between Numido­ therium and desmostylians are relatively few, however: the fe­ mur was habitually abducted, laterally rotated, and posteriorly inclined; the cnemial crest is deflected laterally; the tibial shaft is twisted; and the talocrural joint surface of the tibia faces lat- erodistad. There are, on the other hand, many points of con­ trast: in desmostylians the lateral part of the humeral trochlea protrudes more distally; the spiral elbow joint is designed for forceful medial rotation of the humerus, as in most sirenians (cf. Domning, 1978:125-127, fig. 32c), and is dislocated by lateral rotation; the forefeet were evidently digitigrade rather than plantigrade; the head of the femur is nearly spherical and lacks a fovea, and its shaft is not bowed; the femoral condyles are much more nearly equal in size; the femur was habitually held in a more nearly horizontal position; the proximal tibia faced anterolaterad rather than anterad and lacks an interarticu- lar eminence; movements of the talocrural joint do not seem to involve significant inversion/eversion or abduction/adduction; the calcaneal tuber is inclined; and a rugose plantar tubercle is present on the calcaneum, indicating strong plantar ligaments and possibly implying a digitigrade rather than a plantigrade pedal stance. Numidotherium apparently was specialized in a direction rather different from that of desmostylians, a condition that provides no support for the idea of semisprawling posture in the latter. Furthermore, at least some of the few shared charac­ ters may merely reflect a common tethythere heritage, for ex­ ample, the twisted tibial shaft and inclined talocrural joint sur­ face, as discussed above. REPENNING'S RECONSTRUCTION.—Repenning's (1965) "frog-like" or "sealion-like" reconstruction also can be recon­ ciled with my own by means of a simple adjustment. Although Paleoparadoxia could doubtless have squatted in that position when at rest, it would not necessarily have been confined to such a posture (with the belly "on or very close to the ground," Repenning and Packard, 1990:202) while moving about on land; again the key feature is the hind foot. Repenning and Packard (1990) envisioned desmostylian ter­ restrial locomotion as resembling the "humping" or "hopping" gait of an otarioid pinniped, in which the front feet move in unison followed by the back feet moving in unison. This move­ ment also would involve strong extension and flexion of the spine. They viewed the sole of the hind foot as being nearly vertical and the toes hyperextended against the substrate during this action. They further stated that lifting the body by greater extension of the hind-limb joints (i.e., so that the femur and tibia approached vertical, as in an elephant; CA. Repenning, pers. comm., 1996) would result in "rotating the foot from a plantar position to one in which the plantar plane was ap­ proaching 45° from horizontal; ... concentrating the weight... on the medial edge of the paddle-like foot, and ... rotating the tibial-astragalar articular surface toward the maximum of 45° from horizontal and toward possible dislocation" (Repenning and Packard, 1990:202-203). The latter statement concerning ankle dislocation is true if the tibia is visualized in an elephant-like vertical position. There is, however, a less extreme possibility that also permits an alternating terrestrial gait. Repenning and Packard's inter­ pretation overlooks the features of the metatarso-phalangeal joints described above, and the fact that the pes could adopt a digitigrade stance with the axis of the ankle joint level. This stance would allow the body's weight to be stably supported on moderately flexed hip, knee, and ankle joints and with ab­ ducted knees as shown in Figure 3. With the hip joint directly above the ankle, the weight would indeed fall on the medial side of the digitigrade foot, but this is just what the foot is built for. The hindquarters could thus have been lifted well off the ground, with the hind feet beneath the hip joint, and an alternat­ ing gait could have been used. Repenning and Packard (1990) correctly noted the vertebral articulations allowing arching of the back, and they stated that the front legs "could be held vertically for support." Appar­ ently, then, they did not regard as obligatory the front-limb pos­ ture shown by Repenning (1965), with the dorsum of the manus resting on the ground. In any case, the radial styloid pro­ cess, although preventing hyperextension, does not prevent full extension of the wrist. Their anatomical interpretation and the present one therefore are essentially compatible, given the rein- terpretation of the hind foot outlined above. Their conclusion that "Paleoparadoxia was well adapted to swimming and was as poorly adapted to terrestrial locomotion as are modern otari­ oid pinnipeds" (Repenning and Packard, 1990:203) appears in­ correct, however. Paleoparadoxia was distinctly less aquati- cally specialized than otarioids: it swam more on a par with a polar bear, whereas it moved on level ground more like a typi­ cal ungulate or other terrestrial mammal than an otarioid, and most like a ground sloth. NUMBER 93 111 Literature Cited Barnes, L.G., D.P. Domning, and C E . Ray 1985. Status of Studies on Fossil Marine Mammals. Marine Mammal Sci­ ence. 1(1): 15-53. Coombs, M.C. 1983. Large Mammalian Clawed Herbivores: A Comparative Study. Trans­ actions of the American Philosophical Society. 73(7): 1-96. Court, N. 1994. Limb Posture and Gait in Numidotherium koholense, a Primitive Proboscidean from the Eocene of Algeria. Zoological Journal of the Linnean Society, lll(4):297-338. Davis, DD. 1964. The Giant Panda: A Morphological Study of Evolutionary Mecha­ nisms. Fieldiana: Zoology Memoirs, 3:1-339. Domning, D.P. 1978. Sirenian Evolution in the North Pacific Ocean. University of Cali­ fornia Publications in Geological Sciences, 118: xi + 176 pages. 2001. Supposedly Unique Features of the Desmostylian Tibia Are Also Found in Other Mammals. Bulletin of the Ashoro Museum of Pale­ ontology. 2:39^12. Domning, D.P., CE. Ray, and M.C. McKenna 1986. Two New Oligocene Desmostylians and a Discussion of Tethythe­ rian Systematics. Smithsonian Contributions to Paleobiology, 59: iii + 56 pages. Fish, F.E. 1996. Transitions from Drag-Based to Lift-Based Propulsion in Mamma­ lian Swimming. American Zoologist. 36:628-641. Gazin, CL. 1965. A Study of the Early Tertiary Condylarthran Mammal Meniscothe­ rium. Smithsonian Miscellaneous Collections, 149(2): iv + 98 pages. 1968. A Study of the Eocene Condylarthran Mammal Hyopsodus. Smith­ sonian Miscellaneous Collections. 153(4): iv + 90 pages. Halstead, L.B. 1985. On the Posture of Desmostylians: A Discussion of Inuzuka's "Her­ petiform Mammals." Memoirs of the Faculty of Science, Kyoto Uni­ versity, Biology, 10(2): 137- 144. Inuzuka, N. 1980-1982. [The Skeleton of Desmostylus mirabilis from South Sakhalin, [Parts] I-V.] Chikyu Kagaku [Earth Science], 34:205-214, 247-257; 35:1-18, 240-244; 36:117-127. [In Japanese with English sum­ mary.] 1982. [Atlas of Reconstructed Desmostylians.] Abstracts, 36th Annual Meeting, Association for the Geological Collaboration in Japan, Saitama, pages 44-61. [In Japanese.] 1984. Skeletal Restoration of the Desmostylians: Herpetiform Mammals. Memoirs of the Faculty of Science, Kyoto University, Biology, 9(2): 157-253. 1985. Are "Herpetiform Mammals" Really Impossible? A Reply to Hal- stead's Discussion. Memoirs of the Faculty of Science, Kyoto Uni­ versity, Biology. 10(2): 145-150. 1988. [The Skeleton of Desmostylus from Utanobori, Hokkaido, I: Cra­ nium.] Bulletin of the Geological Survey of Japan, 39(3): 139-190. [In Japanese with English summary.] 1989. [Desmostylus and Behemotops.] In Ashoro Educational Committee, editors, Research Report on the Ashoro Marine Mammal Fauna, pages 40-76. [In Japanese.] Inuzuka, N., D.P. Domning, and C E . Ray 1995. Summary of Taxa and Morphological Adaptations of Desmostylia. The Island Arc, 3:522-531. Jenkins, F.A., Jr. 1973. The Functional Anatomy and Evolution of the Mammalian Humero- Ulnar Articulation. American Journal of Anatomy, 137(3):281-298. Jenkins, F.A., Jr., and S.M. Camazine 1977. Hip Structure and Locomotion in Ambulatory and Cursorial Carni­ vores. Journal of Zoology. 181:351-370. Muizon, C de, and H.G. McDonald 1995. An Aquatic Sloth from the Pliocene of Peru. Nature, 375:224-227. Osborn, H.F. 1942. Proboscidea, a Monograph of the Discovery, Evolution, Migration and Extinction of the Mastodonts and Elephants of the World. Vol­ ume 2, pages xxvii + 805-1675. New York: American Museum Press. Repenning, CA. 1965. [Drawing of Paleoparadoxia Skeleton, with Caption.] Geotimes, 9(6): 1,3. Repenning, C.A., and E.L. Packard 1990. Locomotion of a Desmostylian and Evidence of Ancient Shark Pre­ dation. In A.J. Boucot, editor. Evolutionary Paleobiology of Behav­ ior and Coevolution, pages 199-203. Amsterdam: Elsevier. Romer, A.S. 1966. Vertebrate Paleontology. Third edition, 468 pages. Chicago: Univer­ sity of Chicago Press. Sakamoto, O. 1983. On the Occurrence of the Two Skeletons of Paleoparadoxia tabatai (Tokunaga) from Chichibu Basin, Central Japan. Bulletin of the Saitama Museum of Natural History, 1:17-26. Scott, W.B. 1912. Mammalia of the Santa Cruz Beds, Part II: Toxodonta. Reports of the Princeton University Expedition to Patagonia, 6:111-238, 287-300. 1930. A Partial Skeleton of Homalodontotherium from the Santa Cruz Beds of Patagonia. Geology Memoirs, Field Museum of Natural History, 1:7-34. Shikama, T. 1966. Postcranial Skeletons of Japanese Desmostylia: Limb Bones and Sternum of Desmostylus and Paleoparadoxia, with Considerations on Their Evolution. Palaeontological Society of Japan, Special Pa­ pers, 12: iii + 202 pages. 1968. Additional Notes on the Postcranial Skeletons of Japanese Desmo­ stylia. Science Reports of the Yokohama National University, section 2,14:21-26. Slijper, E.J. 1946. Comparative Biologic-Anatomical Investigations on the Vertebral Column and Spinal Musculature of Mammals. Verhandelingen der Koninklijke Nederlandsche Akademie van Wetenschappen, Afdeel- ingvoor Natuurkunde, series 2, 42:1—128. Thewissen, J.G.M., and D.P. Domning 1992. The Role of Phenacodontids in the Origin of the Modern Orders of Ungulate Mammals. Journal of Vertebrate Paleontology, 12: 494-504. Williamson, T.E., and S.G. Lucas 1992. Meniscotherium (Mammalia, "Condylarthra") from the Paleocene- Eocene of Western North America. Bulletin of the New Mexico Mu­ seum of Natural History and Science, 1: iv + 75 pages. Wilson, J.A., and M.T. Carrano 1999. Titanosaurs and the Origin of "Wide-Gauge" Trackways: A Biome- chanical and Systematic Perspective on Sauropod Locomotion. Pa­ leobiology, 25:252-267. Zapfe, H. 1979. Chalicotherium grande (Blainv.) aus der miozanen Spaltenfullung von Neudorf an der March (Devinska Nova Ves), Tschecho- slowakei. Neue Denkschriften des Naturhistorischen Museums in Wien, 2:1-282. The Miocene Pinniped Desmatophoca oregonensis Condon, 1906 (Mammalia: Carnivora), from the Astoria Formation, Oregon Thomas A. Demere and Annalisa Berta A B S T R A C T New crania, dentitions, mandibles, and postcrania of the fossil pinniped Desmatophoca oregonensis Condon, 1906, are described from the Miocene Astoria Formation of the Newport Embayment, Lincoln County, Oregon. This relatively large sample includes specimens of different developmental ages and genders and per­ mits documentation of the range of variation in this Miocene pin­ niped. This variation suggests that several characters previously considered diagnostic of different species are in fact the result of sexual dimorphism. Reevaluation of the status of Desmatophoca supports recognition of only two species, D. brachycephala Barnes, 1987, and D. oregonensis. A cladistic analysis, using 30 cranial and dental characters, supports the monophyly of the genus Desmatophoca and suggests a close sister-group relationship with a monophyletic Allodesmus clade. The parsimony analysis also confirms the monophyly of the Desmatophocidae, which is herein defined as the clade containing the most recent common ancestor of Desmatophoca and Allodesmus and all of its descendants. Phoc- ids are shown to be the sister group to the desmatophocids in a redefined monophyletic Phocoidea. Introduction The fossil pinniped genus Desmatophoca was founded on the basis of a partially prepared skull, a partial mandible, and postcranial bones collected from the middle Miocene portion of the Astoria Formation at Newport, Lincoln County, Oregon, and was described by Condon (1906) as Desmatophoca ore­ gonensis. A second species known only by a skull was de­ scribed by Barnes (1987) as Desmatophoca brachycephala. This specimen was collected from the lower Miocene portion Thomas A. Demere, Department of Paleontology, San Diego Natural History Museum, P.O. Box 121390, San Diego, California 92112. An­ nalisa Berta, Research Associate, Department of Paleobiology, Na­ tional Museum of Natural History, Smithsonian Institution, Washington, D.C. 20560-0121; and Department of Biology, San Diego State University, San Diego, California 92182. of the Astoria Formation as exposed in southwestern Washing­ ton, across the Columbia River from the stratotype at Astoria, Oregon. Desmatophoca has been variously classified in the family Desmatophocidae (Hay, 1930) or the subfamily Desmatophoci- nae (Mitchell, 1966; Barnes, 1972), the content of which has included (Repenning, 1976; Repenning and Tedford, 1977; King, 1983) or excluded (Mitchell, 1968, 1975; Barnes, 1987, 1989) the genus Allodesmus. More recently, Barnes (1987, 1989) has followed Mitchell (1968, 1975) in recognizing the Allodesminae and Desmatophocinae as separate monotypic subfamilies within the Otariidae sensu lato (=Otarioidea fide Repenning and Tedford, 1977). According to Barnes (1989), the Desmatophocinae are the sister group of walruses and the Allodesminae are more closely linked with Pinnarctidion in a group that includes all other "non-walrus otariids" (i.e., Enali­ arctos, sea lions, and fur seals) (Figure 1). In the classification arrangement adopted herein, the family Desmatophocidae is defined as the monophyletic group containing the most recent common ancestor of Desmatophoca and Allodesmus and all of i ts d e s c e n d a n t s . The O t a r i o i d e a ( E n a l i a r c t i d a e + Desmatophocidae+Odobenidae+Otariidae) of Repenning and Tedford (1977) and the Ota r i i dae ( E n a l i a r c t i n a e + Desmatophocinae+Allodesminae+Odobeninae+Otariinae) of Barnes (1989) were rejected by Wyss (1987), Berta (1991), and Berta and Wyss (1994) as nonmonophyletic groupings that are based largely upon shared primitive characters. Berta (1991) proposed a close relationship between Desmatophoca and the "enaliarctine" pinniped Pinnarctidion in a clade that included Allodesmus+Y*hociddiZ, whereas Berta and Wyss (1994) sup­ ported a slightly different arrangement in which Pinnarctidion, Desmatophoca, and Allodesmus formed a polytomy with a sis­ ter-group relationship to the Phocidae, referring to the entire clade as the Phocoidea (Figure 1). Although Desmatophoca oregonensis was named more than 90 years ago, the original description was minimal and ham- 113 114 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY Other Otariidae (incl. Otariinae, Pinnarctidion Allodesmus. Pteronarctos. Enaliarctos) Desmatophocinae (-Bssmatoohasa) Dusignathinae Odobeninae Imagotariinae Barnes, 1989 Allodesmus Odobenidae Pinnarctidion Desmatophoca Phocidae FIGURE 1.—Alternative phylogenetic hypotheses of phocoid and desmatophocid relationships. Odobenidae Pinnarctidion PHOCOIDEA r^^ OCOMORPHA Berta and Wyss, 1994 Allodesmus Desmatophoca Phocidae PHOCOIDEA PHOCOMORPHA Berta, 1994 pered by the incomplete preparation of the holotype skull. Mitchell (1966, 1975) prepared and illustrated additional por­ tions of the holotype but did not provide more complete osteo­ logical descriptions or comparisons. Such a redescription is critical to an understanding of the phylogenetic significance of this important early pinniped. Unfortunately, the holotype skull is not currently available for study. Among the large number of fossil pinnipeds collected by Douglas Emlong from the Astoria Formation along the Oregon coast near Newport (Ray, 1977), however, are several undescribed specimens referable to Des­ matophoca oregonensis. These specimens, which form the ba­ sis of this report, include new cranial, mandibular, dental, and postcranial material. New insights derived from the study of this material provide evidence for a close phylogenetic rela­ tionship between D. oregonensis, D. brachycephala, and spe­ cies ofAllodesmus. ABBREVIATIONS.—The abbreviations for institutions and collections are as follows: BVM LACM SDSNH SICC UCMP UOMNH USNM Buena Vista Museum of Natural History, Bakersfield, California Natural History Museum of Los Angeles County, Los Angeles, California San Diego Natural History Museum, San Diego, California Sado Island Community Center, Ogi-machi, Sado Island, Japan Museum of Paleontology, University of California, Berkeley, California University of Oregon, Museum of Natural History, Eugene, Ore­ gon Collections of the National Museum of Natural History (NMNH), Smithsonian Institution, Washington, D.C, which in­ clude collections of the former United States National Museum NUMBER 93 115 TABLE 1.—Suture closure scoring for specimens of Desmatophoca oregonen­ sis. Suture classification follows terminology of Doutt (1942) and Siversten (1954). Specimen USNM 335478 USNM 314645 USNM 335451 USNM 335702 USNM 243694 USNM 250283 I 4 - - 4 - 4 11 1 - 1 2 - 4 III 3 - 4 - - - IV 1 - 1 - - - V 1 - 2 3 - 4 VI 1 1 - 4 - 4 VII 1 - 1 2 4 - VIII 1 - 4 4 IX 1 1 1 - 2 - ACKNOWLEDGMENTS.—We thank Lawrence G. Barnes (LACM), Michael Metz (BVM), and Clayton E. Ray (NMNH) for allowing us to study specimens under their care. We also thank Lawrence G. Barnes for providing casts of comparative specimens. Naoki Kohno of the National Science Museum, To­ kyo, Japan, facilitated study of Japanese pinniped specimens during a research trip supported by National Science Founda­ tion (NSF) international programs grant 9417088 to AB. Rob­ ert Ernst of Bakersfield, California, donated specimens of Al­ lodesmus kernensis to SDSNH. Robert L. Clark (SDSNH) skillfully prepared several of the Emlong Desmatophoca skulls. Research was supported in part by NSF grant BSR 9006535 to AB. METHODS To determine the degree of closure for cranial sutures, we used the procedures of Doutt (1942) and Siversten (1954); the resulting scores are provided in Table 1. Measurements, to the nearest tenth of a millimeter, of crania, dentitions, and mandi­ bles were taken using digital calipers; these measurements are provided in Tables 2 and 3 and for the most part represent the standard measurements of Siversten (1954). All specimens in the photographic figures, with the exception of Figure 6B, were coated with a sublimate of aluminum chloride. MATERIAL In addition to the referred specimens listed below, the fol­ lowing specimens were examined during the course of this study: TABLE 2.—Cranial and upper dentition measurements for specimens of Desmatophoca oregonensis and D. brachycephala. (a=measurement for alveolus; e= estimated measurement; *=measurement doubled for missing half; +-broken specimen; -=no data. Measurements of Siversten (1954) indicated by numbers in parentheses.) Measurements Condylobasal length (0) Palatal length Postpalatal length Facial length Braincase length C-M2 length Orbit length Temporal fossa length Greatest width anterior nares (3) Greatest length nasals (4) Rostral width at C (12) Palate width at P4 (anterior root) Width at supraorbital processes (7) Least interorbital width Width of braincase (8) Zygomatic width (17) Auditory width (19) Mastoid width (20) Greatest width of occipital condyles Greatest width of foramen magnum Greatest height of foramen magnum I1;AP/T 12; AP/T 13; AP/T CI; AP/T P1;AP/T P2; AP/T P3; AP/T P4; AP/T Ml; AP/T M2; AP/T USNM 335478 230.0 117.1 99.7 125.0e 99.0e 63.2" 48.0e 34.0e 23.8 59.2 52.7 46.9 32.4 28.7 78.6 132.9 103.9 116.2 60.3 30.0 16.9 3.0/5.3" 3.4/6.1" 7.2/10.7" 14.4/8.0 6.8/6.1" 12.1/8.9 11.6/8.7 10.0/8.3 6.9/7.2 4.9/4.2 USNM 335702 - - 123.3 - 114.0e - 64.0e 49.0e - - - 55.4 35.0* 33.0 77.6+ 176.3 135.1 157.4 70.4 29.8 30.0 -1- -/- -/- -1- 9.9/9.0 11.8/8.0 10.7/8.1 12.8/8.2" 7.3/6.7 6.0/4.1" USNM 335451 - - - - - 96.3 59.0e 42.0e 26.7 55.8 59.8* 53.2 35.4 33.2 80.1 165.0* 130.5 139.0* - - - -/- -/- -/- 17.1/13.4 9.8/8.1 10.4/7.8 10.5/7.8 9.5/8.0 6.6/6.4 5.2/3.1" USNM 314645 255.0e 136.0e 101.9 139.0e I02.0e - 48.0e 41.0e - - 61.3 59.4 - - 84.6+ 153.8 119.1 133.3 72.0* 32.5 + 21.0 -/- -/- -/- 16.8/11.8 8.8/8.4" 13.3/8.9 12.8/8.9" 9.7/7.2" 6.6/3.7" -/- USNM 243694 - 141.8 + - 170.0e - 99.5 48.0e 38.0e - - 65.3+ 47.6 - - - 152.0* 128.0* 148.0* - - - -/- -/- 9.6/8.2 20.2/13.4 9.2/7.5 10.2/8.2+ 10.2/7.9+ 8.9/6.8 5.6/4.8+ 7.5/5.3" Cast of holotype SDSNH 61143 326.0 172.7 135.0 195.0e 120.0e - 64.0e 54.0e 28.8 - 76.7+ 68.0* 36.0+ 36.1 + 77.3 + 188.0* 144.1 161.9 70.6 31.7 18.5 -/- -/- 11.4/8.1 18.2/13.7 8.1/8.3e 12.9/9.9 15.9/8.0" 13.4/10.1 10.5/5.5" -/- Holotype of D. brachycephala LACM 120199 282.0e 144.0e 119.9 169.0s 106.0* 97. T 54.0e 46.0* - 76.0+e 63.1 - - 88.0*e 182.3 143.8 168.6 70.9 31.6 23.5 -/- -/- -/- -/- -/- -/- -711.5" 13.2/12.3" 6.2/6.0 6.8/4.2" 116 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY TABLE 3.—Mandibular and lower dentition measurements for specimens of Desmatophoca oregonensis. (a= measurement for alveolus; b=measurement for base of root; c=measurement for c-ml; +=broken specimen; -- no data.) Measurements Ramus total length c-m2 length Greatest length symphysis Greatest width symphysis Angle of symphysis to alveoli Depth ramus at p3 Depth ramus at coronoid process cAP/T pl AP/T p2 AP/T p3 AP/T p4 AP/T ml AP/T m2 AP/T USNM 335478 147.1 67.2C 52.0 18.6 48° 34.4 65.9 13.4/8.9 8.6/7.4 9.8/7.6 11.5/8.0 11.1/7.8 11.2/7.4 - USNM 335694 188.7 95.6 54.1 23.4 49.5° 37.7 83.7 18.2/11.8 8.6/6.9" 12.7/6.9" 13.5/5.9" 11.1/7.6 10.1/6.7 6.0/3.4" USNM 314536 161.8 83.9 50.4 22.1 41° + 30.2 59.3 12.2/9.0 - 10.7/7.7 11.1/7.8 11.3/7.8 10.5/7.2 6.7/4.3" USNM 335457 - 93.7 60.9 25.2 47.5° 25.0 - 17.5/10.9 9.1/4.7" 11.6/8.5 11.8/8.5 11.3/8.0 11.1/8.0 9.5/3.0" USNM 335430 - 117.5 63.6+ 32.6 59° 43.7 - 19.0/14.0 12.5/8.5" 15.2/6.3b 15.4/8.3b 16.8/7.lb 18.2/6.3b 9.0/4.4" 1. UCMP 86334, holotype of Pinnarctidion bishopi Barnes, 1979 2. LACM 1376, skeleton on slab of Allodesmus courseni (Downs, 1956) 3. LACM 4320, complete skeleton of Allodesmus kernen- sis Kellogg, 1922 (=holotype of Allodesmus kelloggi Mitchell, 1966) 4. SDSNH 54801, referred braincase of Allodesmus kern- ens is 5. SDSNH 54802, referred right squamosal of Allodesmus kernensis 6. BVM 0163, associated complete skull and mandible of Allodesmus kernensis 7. BVM 0164, complete skull of Allodesmus kernensis 8. USNM 205315, cast of holotype of Allodesmus packardi Barnes, 1972 9. SICC 0001, holotype of Allodesmus sadoensis Hirota and Barnes, 1995 10. SDSNH 61143, cast of holotype of Desmatophoca ore­ gonensis Condon, 1906 11. LACM 120199, holotype of Desmatophoca brachyceph­ ala Barnes, 1987 12. USNM 314325, holotype of Pinnarctidion rayi Berta, 1994 13. USNM 250345, holotype of Enaliarctos emlongi Berta, 1991 LOCATION, STRATIGRAPHY, AND CORRELATION All but one of the pinniped fossils described in this report were collected by the late Douglas Emlong from the marine Astoria Formation in the Newport Embayment near Newport, Oregon. The majority of these fossils were recovered from a single stratigraphic horizon, the "Iron Mountain bed" (Armen- trout, 1981), as exposed along Moolack Beach, north of Yaquina Head (Ray, 1977). In addition to Desmatophoca oregonensis, four other pin­ niped taxa have been described from the Astoria Formation of the Newport Embayment: Pteronarctos goedertae Barnes, 1989; P. piersoni Barnes, 1990; Pacificotaria hadromna Barnes, 1992; and Proneotherium repenningi Barnes in Kohno et al., 1995. Cetaceans known from these exposures include the mysticete Cophocetus oregonensis Packard and Kellogg, 1934, and an undescribed new species of squalodontid odonto- cete (Dooley, 1994). A desmostylian, Desmostylus hersperus Marsh, 1888, also is reported from the Newport Embayment exposures of the Astoria Formation (Packard and Kellogg, 1934). The molluscan and foraminiferal biostratigraphy of the Asto­ ria Formation is rather complex and suggests an age range from early to early middle Miocene. In the type area at Astoria, Ore­ gon, the formation contains molluscs typical of the provincial Pillarian Stage (Addicott, 1976), whereas outcrops in the New­ port Embayment have served as the stratotype for the younger Newportian Stage. Foraminiferal assemblages from the Astoria Formation at Astoria are correlated with the provincial Sauce- sian Stage, whereas those from the Newport Embayment sug­ gest correlation with the upper Saucesian through Relizian Stages (Armentrout, 1981). The dome-headed chalicothere re­ covered from the "Iron Mountain bed" in the Astoria Forma­ tion of the Newport Embayment suggests correlation with the Hemingfordian or early Barstovian North American Land Mammal Age (Munthe and Coombs, 1979). At Depoe Bay (Newport Embayment), the Astoria Formation is overlain and intruded by the Depoe Bay Basalt (Snavely et al., 1973), which at Cape Meares near Tillamook Head, Oregon, yielded whole- rock K/Ar dates of 14.5± 1.0 million years ago (Ma) and 15.2± 0.6 Ma (Turner, 1970). Exposures of the Depoe Bay Basalt near Ecola, Oregon, yielded a whole-rock K/Ar date of 14±2.2 Ma (Turner, 1970). These dates represent a minimum age for the Astoria Formation. Barnes (1989) suggested that the expo­ sures in the type area near Astoria may be as old as 23-20 Ma, NUMBER 93 117 and that those in the Newport Embayment probably range from 19 to 15 Ma. The type species of Desmatophoca, D. oregonensis, is known only from the Astoria Formation of the Newport Em­ bayment (Newportian, Saucesian-Relizian, Hemingfordian/ Barstovian), whereas D. brachycephala is known only from the Astoria Formation in the type area near Astoria (Pillarian, Saucesian). The geochronologic range of the genus Desmato­ phoca thus extends from the early Miocene through the early middle Miocene, approximately from 23 to 15 Ma. Family DESMATOPHOCIDAE (Hay, 1930) Genus Desmatophoca Condon, 1906 TYPE SPECIES.—Desmatophoca oregonensis Condon, 1906. DISTRIBUTION.—Early Miocene to early middle Miocene of Oregon and Washington. INCLUDED SPECIES.—Desmatophoca oregonensis Condon, 1906; Desmatophoca brachycephala Barnes, 1987. DEFINITION.—The monophyletic group containing the most recent common ancestor of D. brachycephala and D. oregon­ ensis and all of its descendants. EMENDED DIAGNOSIS.—Desmatophocine pinnipeds with posteriorly expanded zygomatic process of squamosal (autapo­ morphy of Desmatophoca), enlarged but not excavated paroc­ cipital process, raised strut separating stylomastoid foramen from tympanohyal, and 13 procumbent and laterally directed (synapomorphies shared with Allodesmus). Laterally expanded pterygoid process (convergent with Pinnarctidion), mortised jugal-squamosal contact, acutely V-shaped nasal-frontal suture, and well-developed marginal process of mandible (synapomor­ phies at the level of the common ancestry of desmatophocids and phocids). REMARKS.—Repenning and Tedford (1977:74) listed the fol­ lowing features as diagnostic of the genus Desmatophoca: large size; double-rooted cheek teeth; well-developed internal cingulum on cheek teeth; large incisive foramina (for a des- matophocid); weak mortising of jugal-squamosal articulation; relatively small orbits (for a desmatophocid; 17% of condylo- basal length); and dental formula (11-3, C, Pl-4 , Ml,2, il,2, c, pl^4, ml). It is acknowledged that this "diagnosis" was meant to be dichotomous between Desmatophoca and Allodesmus; however, all of these character states are primitive at the level of Desmatophoca, so they are uninformative for resolving in- trafamilial relationships. Barnes (1989) listed five characters as diagnostic of his mo­ notypic subfamily Desmatophocinae (includes only the genus Desmatophoca): orbit enlarged; cheek-tooth crowns bulbous, with smooth enamel; paroccipital process enlarged; tympanic crest reduced, not projecting into tympanic cavity; and ptery­ goid process (palatine process) of maxilla expanded laterally beneath orbit. Of these characters, none is unique to Desmato­ phoca. Although large orbits are diagnostic at the Pinnipedi- morpha level, very large orbits are recognized as a secondary transformation of this character shared by Desmatophoca, Al­ lodesmus, and most phocids. Bulbous cheek-tooth crowns (i.e., crowns rounded and swollen) with smooth enamel occur in both Allodesmus and Desmatophoca. A large paroccipital pro­ cess is diagnostic at the Pinnipediformes level; the secondary transformation—paroccipital process greatly enlarged postero­ laterally—is seen in both Allodesmus and Desmatophoca. An­ other purported diagnostic feature of this genus, reduced tym­ panic crest, was actually listed by Barnes (1989, fig. 9, caption) as originating independently in Desmatophoca, Allodesmus, and his Otariinae. Finally, expansion of the palatine process of the maxilla beneath the orbit is a feature shared by Pinnarctid­ ion, Allodesmus, and Desmatophoca. These previous attempts at diagnosing the genus Desmato­ phoca have essentially provided a characterization of this taxon following the usage of Demere (1994; i.e., a listing of distin­ guishing characters, both shared-derived and shared-primitive homologous characters, and their taxonomic distributions). Un­ der the phylogenetic system as proposed by Rowe (1988) and de Queiroz and Gauthier (1990), diagnosis of a clade is a list­ ing of shared-derived homologous character states and the level of generality at which they occur, whereas definition of a taxon is based upon ancestry and taxonomic membership. Our diagnosis and definition of the Desmatophoca clade is based upon the phylogenetic analysis discussed later in this report. Desmatophoca oregonensis Condon, 1906 FIGURES 2^1, 6a, 8,10-14 Desmatophoca oregonensis Condon, 1906:1-14.—Kellogg, 1922:107.—Pack­ ard and Kellogg, 1934:20.—Mitchell, 1966:36-39, pl. 29; 1968:1883-1886; 1975:12-14, fig. 2.—Barnes, 1972:60-65; 1987:4-6.—Repenning and Ted­ ford, 1977:74. EMENDED DIAGNOSIS.—A species of Desmatophoca with incisive foramina undivided by medial septum, angle formed by long axis of external auditory meatus and sagittal plane of skull less than or equal to 62° (autapomorphies of D. oregonen­ sis), well-developed cingula and cuspules/crenations on Pl—3, conical anterior and posterior cuspules on p2—4 (derived char­ acter states that also may apply to D. brachycephala), no pre- narial shelf, distinct carinae on C1, and thick lateral wall of al­ isphenoid canal (symplesiomorphies shared with basal pinni­ peds). HOLOTYPE.—UOMNH F735, nearly complete skull (miss­ ing the left zygomatic arch) with right and left 13, C, P2, and right P4, and partial left mandible with c and p2. Size of skull suggests probable mature male. TYPE LOCALITY AND HORIZON.—On the coast near New­ port, Lincoln County, Oregon; UOMNH locality 1153 (Con­ don, 1906; Packard and Kellogg, 1934). REFERRED SPECIMENS.—LACM 123285, anterior part of skull with right C, P2,3, and left 13, C, and P l - 3 . Size and de­ gree of suture closure (Table 1) suggest probable immature male. Collected by D.J. Martel, 18 April 1983, from Astoria 118 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY FIGURE 2.—Desmatophoca oregonensis: A, USNM 335478, dorsal aspect, immature female skull (stereophoto- graphs); B, USNM 335451, dorsal aspect, mature male skull (stereophotographs). (Scale bars=5 cm.) Formation, LACM locality 4851, among boulders on beach midway between mouths of Schooner and Moloch Creeks, ap­ proximately 6.5 km north of Newport, Lincoln County, Oregon (Barnes, 1987). USNM 250283, posterior portion of sectioned skull preserv­ ing basicranium, occipital region, and left and right squamosal portions of zygomatic arches. Development of rugose lambdoi- dal crest suggests probable mature male. Collected by Douglas Emlong from drift concretion in Astoria Formation approxi­ mately 1.2 km north of Yaquina Head lighthouse, Lincoln County, Oregon. USNM 243694, complete palate with right zygomatic arch and mastoid region, left 13, C, Pl^l, Ml and right 13, C, Pl,2, P4. Canine size and degree of postcanine tooth wear suggest NUMBER 93 119 FIGURE 3.—Desmatophoca oregonensis: A, USNM 335478, lateral aspect, immature female skull; B, USNM 335451, lateral aspect, mature male skull. (Scale bars=5 cm.) probable mature male. Collected by Douglas Emlong, 1961, from "Iron Mountain bed," 7.2-8 km north of mouth of Yaquina River, Lincoln County, Oregon. USNM 335478, complete skull with left C, P3,4, Ml and right P2, P4, Ml; right mandible with p3 and ml and left man­ dible with i3, c, pl^4; isolated teeth (right P3, M2, p l ; left Pl , ml); and atlas vertebra. Size, degree of suture closure (Table 1, suture age of 14), laterally placed temporal crests, and canine size indicate probable immature female. Collected by Douglas Emlong, April 1970, in place in "Iron Mountain bed," Astoria Formation, approximately 0.5 km north of Moolack Beach State Park, 75 m south of mouth of Wade Creek, Lincoln County, Oregon. USNM 314645, complete skull with canines and left P2. Size, incomplete canine eruption, and degree of suture closure (Table 1) suggest probable immature female. Collected by Douglas Emlong in place in Astoria Formation, "Fogarty Creek block," Lincoln County, Oregon. USNM 335702, skull with right and left P2,3, Ml ; anterior portion of rostrum missing. Size, degree of suture closure (Ta­ ble 1), and development of rugose lambdoidal crests indicate probable mature male. Collected by Douglas Emlong, spring 1961, in drift from "Iron Mountain bed," Astoria Formation, approximately 100 m north of mouth of Schooner Creek, Lin­ coln County, Oregon. USNM 335451, nearly complete skull missing the incisive margin of rostrum, occipital shield, and occipital condyles; in­ cludes left C, left and right Pl—4, Ml, and associated left man­ dible with p2-4, ml. Canine size and degree of suture closure (Table 1) suggest probable mature male. Collected by Douglas Emlong, spring 1964, from Astoria Formation, approximately 100 m south of mouth of Spencer Creek, Lincoln County, Ore­ gon. USNM 335457, partial left mandible with c, p2^1, ml . Col­ lected by Douglas Emlong from Astoria Formation, approxi­ mately 800 m north of Beverly Beach, Lincoln County, Ore­ gon. USNM 335723, partial left mandible with p4. Collected by Douglas Emlong from "Iron Mountain bed," Astoria Forma­ tion, approximately 1.6 km north of Agate Beach, Lincoln County, Oregon. USNM 314536, associated right and left mandibles with right p2-4, ml . Collected by Douglas Emlong in float from "Iron Mountain bed," Astoria Formation, approximately 800 m south of Moolack Beach State Park, Lincoln County, Oregon. 120 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY FIGURE 4.—Desmatophoca oregonensis: A, USNM 335478, ventral aspect, immature female skull (stereophoto- graphs); B, USNM 335702, ventral aspect; mature male skull (stereophotographs). (Scale bars=5 cm.) USNM 335721, worn partial right edentulous mandible. Col­ lected by Douglas Emlong from special concretion in "Iron Mountain bed," Astoria Formation, Lincoln County, Oregon. USNM 335694, complete right mandible with c, p4, ml. Collected by Douglas Emlong from "Iron Mountain bed," As­ toria Formation, approximately 1.6 km north of Agate Beach, Lincoln County, Oregon. USNM 335430, associated right and left mandibles with ex­ tremely worn right and left p2-4 and ml, isolated right C, Pl(?), P3(?), and Ml(?); damaged right radius; head of left fe­ mur; proximal end of left tibia; right scapholunar; left patella; cervical vertebra (C7?); thoracic vertebra (TI?); fused sacral NUMBER 93 121 FIGURE 5.—Anterior portion of palate in pinnipeds: A, Enaliarctos emlongi: B, Desmatophoca oregonensis: C, Allodesmus kernensis: D, Monachus schauin- slandi. (if=incisive foramen; m=maxilla; p=premaxilla.) vertebrae (SI—3); eight ribs; and sternebra. Collected by Dou­ glas Emlong, March 1967, from several drift concretions, Asto­ ria Formation, approximately 1.6 km south of mouth of Big Creek, Lincoln County, Oregon. USNM 335246, associated right and left humeri, left scap­ ula, right ulna, and metacarpals. Collected by Douglas Emlong from drift concretion, Astoria Formation, 0.8 km north of mouth of Spencer Creek, Lincoln County, Oregon. DESCRIPTION.—Skull: The Emlong specimens provide im­ portant new information on the anatomy of Desmatophoca ore­ gonensis, especially the dentition, palate, orbital wall, occipital shield, basicranium, and mandible. The number of available specimens together with the occurrence of individuals of dif­ ferent developmental ages and genders provides an opportunity to document the range of variation in this Miocene pinniped. The following descriptions include discussions of this varia­ tion. Analysis of sutural closure (Table 1), tooth wear, and second­ ary sexual characteristics (e.g., development of sagittal crest and lambdoidal crests) suggests that USNM 335478 and 314645 are immature females, and that USNM 335451 and 335702 are mature males. USNM 243694, 250283, and LACM 123285, all incomplete crania, also are probably males as judged by canine size, sutural wear, and/or lambdoidal crest de­ velopment. Overall, the skull of Desmatophoca oregonensis is elongate (Figure 2). The condylobasal lengths (CBL) of SDSNH 61143 (plaster cast of holotype, UOMNH F735), USNM 335478, and 314645 are 327 mm, 231 mm, and 250 mm, respectively (Table 2). Barnes (1987) estimated the CBL of Desmatophoca brachycephala (holotype, LACM 120199) to be 283 mm. The CBL dimension in Allodesmus kernensis ranges from 370 to 400 mm (Mitchell, 1966; Barnes, 1972; pers. obs.). The facial length (i.e., approximate'level of cribriform plate to prosthion) in proportion to the CBL is variable in the Em­ long sample of D. oregonensis, ranging from 54% in USNM 335478 (immature female) to 60% in the cast of the holotype (mature male). This variation is ascribed to the different devel­ opmental ages of these skulls and follows the observations of King (1972) that the braincase in phocid pinnipeds ceases growth earlier than the face. In the adult skull of D. brachy­ cephala, the facial length is approximately 60% of the CBL, whereas in A. kernensis (LACM 4320) this percentage is 65% and in Enaliarctos emlongi (USNM 250345), 54%. The orbit (i.e., antorbital rim of maxilla to postorbital pro­ cess of jugal) is large, approximately 19%-21% (x=20%, n=2) of the CBL. The percentage of 17% given by Repenning and Tedford (1977) was presumably based upon the holotype, which has an incomplete postorbital region. In D. brachyceph­ ala, this measure is 19%, and in Allodesmus kernensis it is 25% (in agreement with the conclusions of Repenning and Tedford, 1977). The temporal fossa (for operational purposes defined herein as the cranial opening between the anterior border of the preglenoid process of the squamosal and the postorbital pro­ cess of the jugal) is anteroposteriorly shorter than the orbit, ap­ proximately 71%-79% (x=15%, n=4) of the orbit length in D. oregonensis, compared with 85% inD. brachycephala, 61% in A. kernensis, and 92% in E. emlongi. In lateral aspect (Figure 3), the interorbital region is high and at the same level as the dorsal surface of the braincase. The preorbital portion of the rostrum descends slightly (20°-24° from the horizontal; x=21.7°, n=3) from the interorbital region to the level of the posterodorsal margin of the external nares, at which point there is an abrupt flexure of the facial profile, which then descends anteroventrally (41°-44°; x=42.1°, n=3>) to the incisive margin of the rostrum. The facial profile of D. brachycephala, although distorted, appears to possess a nearly horizontal (2°) preorbital region, a more marked flexure, and an incline of 42° at the narial margin. The facial profile of Al­ lodesmus kernensis lacks the abrupt flexure at the posterodor­ sal margin of the external nares and instead consists of a nearly uniformly (30°) descending dorsal margin from the interorbital region to the incisive margin of the rostrum. Palate; The palate of Desmatophoca oregonensis pre­ serves a pair of large ovoid incisive foramina (for transmission of the incisive duct and the rostral septal branch of the palatine artery) that coalesce medially to create a single palatal opening (Figure 4). The anterior margin of this common opening ex­ tends to the lingual border of the incisor alveoli. This unique structure of the incisive foramina is seen in the mature male skulls, USNM 335451 and 243694, and in the immature female skull, USNM 335478. In these latter two specimens, small wedges of bone aligned with the intermaxillary suture extend 122 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY from the posterior margin of the foramen a short distance into the incisive opening (Figure 5B). In Enaliarctos emlongi, Pin­ narctidion rayi, and Desmatophoca brachycephala, the large incisive foramina remain medially separated by a thin maxil­ lary septum (Figure 5A). In Allodesmus kernensis, the incisive foramina are extremely reduced and coalesce to form a very small circular to ovoid opening on the intermaxillary suture (Figure 5c). This extreme reduction of the incisive foramina also is seen in species of the extant phocids Monachus and Mir- ounga and is presumably correlated with possession of a pre- narial shelf. The palatal segment of the premaxillary-maxillary suture, as preserved in USNM 335478 and LACM 123285 emerges from the posterolateral corner of each incisive foramen and makes an anteriorly concave arc to intersect the posteromedial corner of the canine alveolus (Figure 5B). This suture is closed and not visible in the holotype of D. brachycephala. In Allodesmus kernensis, this suture is considerably longer and leaves the in­ termaxillary suture well posterior to the reduced incisive fora­ men and runs anterolateral^ to join with the anteromedial cor­ ner of the canine alveolus (Figure 5c). Large anterior palatine foramina (for transmission of the ma­ jor palatine artery) are positioned on either side of the inter­ maxillary suture at the level of the anterior root of P4. Each fo­ ramen lies at the posterior end of an elongate channel that extends anteriorly along either side of the flattened medial por­ tion of the palate. Several smaller foramina are scattered on the palate posterior to the anterior palatine foramina. These smaller palatine foramina appear to increase in number with sexual ma­ turity (e.g., USNM 335702). In D. brachycephala, the anterior palatine foramina are positioned more posteriorly, at the level ofMl. Overall, the palate of D. oregonensis is narrow anteriorly and broad posteriorly (Figure 4). The width between the canine al­ veoli (lingual borders) ranges from 27.1 to 39.4 mm (x=32.1 mm, n-6), whereas the width between the anterior alveoli (lin­ gual borders) of Ml ranges from 54.8 to 82.2 mm (x=66A mm, n=l). For both measurements, the greatest width was found in the cast of the holotype skull. Expressed as a ratio (i.e., interca- nine width/inter-Ml width), the range for D. oregonensis is 0.48 to 0.52 (x=0.49, n=6). This ratio will be greatest in spe­ cies with nearly parallel tooth rows and least in species with posteriorly broadened palates. Allodesmus kernensis (LACM 4320) has a ratio of 0.58, A. packardi (USNM 205315), 0.43, Pinnarctidion rayi (USNM 314325), 0.53, and Enaliarctos em­ longi (USNM 250345), 0.65. In D. brachycephala (LACM 120199), the intercanine and inter-Ml widths are 31.8 and 70.8 mm, respectively, yielding a ratio of 0.45. These values fall within the range of variation seen in D. oregonensis and sug­ gest that there is no significant difference in palate shape be­ tween these two species of Desmatophoca. The palate of D. oregonensis is slightly arched transversely and longitudinally. The degree of transverse arching was quan­ tified, and a ratio of palate depth to width (i.e., arch of palate) was calculated (see Appendix, character 4). Ratios computed at the level of each maxillary tooth position reveal that the great­ est arching occurs on the anterior portion of the palate between the canine and P2. The P1-M2 palate arch ratios range from only 0.15 to 0.26 and average about 0.19. InD. brachycephala, the greatest palatal arching also is anteriorly placed and the ra­ tios average only 0.18. In Allodesmus kernensis (LACM 4320), the greatest transverse arching occurs on the posterior portion of the palate between P4 and Ml , and the P1-M2 palate arch ratios range from 0.21 to 0.30 and average 0.26. Desmatophoca oregonensis, D. brachycephala, and Allodes­ mus kernensis lack all traces of the embrasure pits between P3 and P4 and between P4 and Ml that are preserved on the pal­ ates of species of Enaliarctos, Pinnarctidion, and Proneothe- rium. The palatine process of the maxilla (=pterygoid process of maxilla of Barnes, 1987) is well developed in LACM 123285, USNM 335451, 335702, 335478, and 314645 and extends as a thin and broad shelf behind M2 and beneath the orbit (Figure 4). In USNM 335478 (Figure 4A), the lateral margin of the pa­ latine process makes a sharp flexure as it curves posteromedi­ ally to merge with the narrow palatines in the vicinity of the an­ gle for the descending minor palatine artery. In USNM 335702 (Figure 4B), the lateral margin of the palatine process is elon­ gate and extends 22 mm behind the M2 alveoli. The broad pa­ latine process in Desmatophoca oregonensis is similar to that in D. brachycephala, Pinnarctidion rayi, and P. bishopi. In Al­ lodesmus kernensis, the lateral margin of the palatine process is broadly convex and not as abruptly flexed. A variation in this feature in the Emlong specimens of D. oregonensis is seen in USNM 243694, in which the palatine process lacks a promi­ nent lateral flexure, especially on the right side. The palatine process is broken in the cast of the holotype but presumably was also broad. In LACM 123285, the palatine process is eroded on its posterolateral corner but still forms a shelf behind M2 (Barnes, 1987, fig. IB). The maxillary-palatine suture in Desmatophoca oregonensis is transverse medially and occurs at the level of the posterior root of Ml in USNM 335478 and 335702 and the anterior root of M2 in USNM 243694 and 335451. In all specimens, the su­ ture then curves posterolaterally to intersect the lateral margin of the palatine process of the maxilla just medial to its postero­ lateral corner. The palatines are transversely broad and generally flattened. The anterior border of the internal nares is smooth, bluntly rounded, and positioned at the level of the cribriform plate. The lateral margins of the internal nares are formed by the palatines anteriorly and the pterygoids posteriorly. The palatine-ptery- goid suture is not visible on any specimen. Pterygoid and Alisphenoid: The pterygoid hamuli, best preserved on USNM 335702, are slender and laterally re­ curved, terminating in anteroposteriorly flattened (not trans­ versely flattened) processes (Figure 4B) for origination of pha­ ryngeal muscles (e.g., pterygopharyngeus). The laterally NUMBER 93 123 recurved portion of the hamulus marks the position of the ten­ sor veli palatini. The pterygoids of USNM 335702 are tightly co-ossified with the alisphenoid, thus obscuring all sutures. The form of the alisphenoid, however, is discernible on other specimens and is short on the pterygoid strut where it contacts the pterygoid and palatine. The area on the strut for origination of the internal pterygoid muscle is planar to slightly convex and is marked neither by a depression, as in Allodesmus kern­ ensis, nor by a broad concave shelf, as in Pinnarctidion rayi and fossil odobenids (e.g., Imagotaria downsi Mitchell, 1968). In lateral aspect, the pterygoid strut is relatively thicker than in Pinnarctidion bishopi ox A. kernensis, but it is not as thick as in later-diverging odobenids (e.g., I. downsi). The posterior end of the pterygoid strut is penetrated by the alisphenoid canal (for transmission of the maxillary artery), which lies in a common pit with the foramen ovalis. In USNM 335702 and 335451, this common pit is short and deep as in D. brachycephala, whereas in USNM 335478 the pit is anteroposteriorly elongate and shal­ low. On some specimens (USNM 335451 and 250283), a dis­ tinct pit-like fossa is located immediately anterolateral to the posterior alisphenoid canal and dorsal to the base of the ptery­ goid strut. This fossa is presumably for origination of the exter­ nal pterygoid muscle. Such a fossa does not occur in D. brachycephala or A. kernensis. The lateral wall of the alisphenoid canal is thick and well de­ fined in both species of Desmatophoca, unlike the condition in Allodesmus kernensis in which the lateral wall is quite thin and usually not preserved. In such specimens of A. kernensis, there is often no obvious trace of the lateral wall of the canal. This condition "mimics" the condition in phocids in which the al­ isphenoid canal typically is missing. The temporal wing of the alisphenoid extends dorsally onto the sharply angled anterolateral corner of the braincase, as in Allodesmus kernensis, early-diverging odobenids, otariids, and presumably species of Enaliarctos. Basicranium: The presphenoid-basisphenoid suture, as preserved in USNM 335478 and the cast of the holotype, is un­ fused and allows recognition of the presphenoid, which is acutely triangular anteriorly and broader posteriorly than in^4/- lodesmus kernensis. The basisphenoid-basioccipital suture is unfused in USNM 335478 and 314645 but fused in all other skulls. This suture is straight in the transverse plane and is po­ sitioned midway between the anterior and posterior openings of the carotid canal. The basisphenoid is a narrow trapezoidal bone with a broad sagittal sulcus. The basioccipital is broadly trapezoidal with a low sagittal ridge separating the raised inser­ tions for the longus capitis muscle anteriorly and the shallow fossae for insertion of the rectus capitis lateralis muscle poste­ riorly (Figure 4). In more mature specimens (e.g., USNM 335702), the basioccipital and basisphenoid are strongly co-os­ sified, the rectus capitis insertions are more rugose, the sagittal ridge is more keeled, and the posterior fossae are broader and deeper. The hypoglossal foramina, for passage of cranial nerve XII, are small (relative to those in A. kernensis) and positioned midway between the posterior lacerate foramen and the occipi­ tal condyles. The auditory bulla in D. oregonensis is moderately inflated relative to the distinctly inflated bullae in Enaliarctos and Pinn­ arctidion and the flattened bullae in Allodesmus and many fos­ sil odobenids. Unlike the condition in Allodesmus kernensis, there is no clear distinction between the ectotympanic and ento- tympanic portions of the bulla, even on immature skulls. Ante­ riorly, the bulla is closely appressed to the posterior side of the postglenoid process. Medially, the entotympanic is distinct from the basioccipital and descends to a level below the basioc­ cipital. The carotid canal runs through a short, swollen tube, the posterior opening of which is retracted from the posterior lacer­ ate foramen, and forms a distinct dorsal bony shelf between these two foramina as in Pinnarctidion and Allodesmus (Figure 6B) . The ectotympanic has a rugose transverse crest that ex­ tends from the medial portion of the bulla to a point ventral to the external auditory meatus (Figure 4). The portion of the ecto­ tympanic anterior to this crest is smoother than that posterior to it. An elongate styliform process at the anteromedial edge of the bulla projects anterior to the anterior opening of the carotid canal. In USNM 335478, the styliform processes on both the left and right sides are marked by a small, but deep anteroposte­ riorly, oval pit (Figure 4A) measuring 2.4 by 4.5 mm. This pit does not penetrate into the carotid canal. Although similar pits were not observed in other specimens of Desmatophoca ore­ gonensis or D. brachycephala, or in outgroup taxa, they were seen in specimens of Allodesmus kernensis (SDSNH 54801 and LACM 4320) (Figure 6B). A distinct styloid process underlies the opening for the eustachian canal and is best developed in the cast of the holotype and in USNM 250283 and 335702. In A. kernensis, there is no such styloid process; instead this re­ gion of the bulla is marked by a distinct medially directed em­ bayment or retraction of the anterior margin of the median lac­ erate foramen (Figure 6 B ) . In the latter taxon and D. ore­ gonensis, there is no bony meatal tube, and the external audi­ tory meatus (EAM) is long and anterolaterally directed. The long axis of the EAM forms an angle of 56°-61° (x=57.8°, n= 4) with the sagittal plane. In D. brachycephala, the EAM is more laterally directed and forms an angle of 75°. In A. kernen­ sis, this angular measurement is 66° Posterolateral to the bulla, the circular and ventrolaterally di­ rected stylomastoid foramen is separated from the tympano- hyal pit by either a thin bony septum (USNM 250283) or a broader bony strut (as in all other USNM specimens). The tym- panohyal pit in this taxon is circular and directed posteroven- trally. In D. brachycephala, the two openings are small, closely appressed, and separated by a thin bony septum, and the tym- panohyal pit is circular and faces posterolaterally. In A. ker­ nensis, the stylomastoid foramen is anteroposteriorly elongate and the tympanohyal pit is larger (relative to D. oregonensis), circular, and posterolaterally directed (Figure 6B). In D. ore­ gonensis the posterior lacerate foramen is large and trans­ versely oval, not circular as in D. brachycephala and A. kern- 124 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY FIGURE 6.—A, Desmatophoca oregonensis, USNM 335451, right middle ear region (stereophotographs); B, Allodesmus kernensis, SDSNH 54801, basicranium, left side (stereophotographs). (Scale bar in A=2 cm; scale bar in B=5 cm.) ensis. In all three taxa, the posterior margin of the petrosal is exposed in ventral aspect at the anterior end of the posterior lacerate foramen. In Desmatophoca oregonensis, the paroccipital process for insertion of the rectus capitis lateralis muscle (flexor of the at- lanto-occipital joint) and origination of the digastricus muscle (mandible depressor) is large (relative to that of Enaliarctos) and dorsoventrally flattened. In ventral aspect it projects poste­ riorly and slightly posteroventrally (Figure 4A) and is roughly parallel to the sagittal plane. The ventral surface of this process in mature skulls (USNM 250283 and 335702) is marked by several low crests and fossae (Figure 4B). In D. brachycephala, the paroccipital processes are constructed as in D. oregonensis except that they project more posterolaterally, forming an angle of 62° with the sagittal plane. Ventrally, the exoccipital-squa- mosal suture reveals that the paroccipital process in D. oregon­ ensis is formed entirely from the exoccipital (Figure 7C). This suture is open and discernible even in mature skulls and can be traced from the lateral border of the posterior lacerate foramen laterally around the paroccipital process and onto the lateral NUMBER 93 125 B FIGURE 7.—Mastoid region in pinnipeds: A, Enaliarctos emlongi: B, Pinnarc­ tidion bishopi (after Barnes, 1979); C, Desmatophoca oregonensis; D, Allodes­ mus kernensis; E, Piscophoca pacifica (after Muizon, 1981). (b = auditory bulla; e=exoccipital; m=mastoid; p=parietal; s=squamosal.) surface of the skull to join dorsally with the lambdoidal crest. In lateral aspect, the paroccipital process is anteroposteriorly thick, rather than thin and splint-like as in odobenids (Figure 3). Species of Allodesmus also have posteriorly elongate paroc­ cipital processes (best discerned in ventral aspect), but these are relatively shorter than in D. oregonensis and are more transversely broader along the exoccipital-squamosal suture (Figure 6B) . AS in D. oregonensis, this suture remains open even in mature skulls of A. kernensis. In lateral aspect, the lat­ eral fossa on the side of the braincase between the mastoid and paroccipital processes (here called the mastoid fossa) is deeper and more anteroposteriorly elongate in D. oregonensis and D. brachycephala than in A. kernensis. In all three taxa, however, the paroccipital process (as viewed in lateral aspect) is sepa­ rated from the mastoid process by a continuous horizontal shelf as in the type of Pinnarctidion bishopi (Figure 7B). In species of Enaliarctos, the two processes are separated by a dorsally arched ridge (Figure 7A). In lateral aspect, the mastoid process in D. oregonensis is lunate in mature skulls (USNM 335451 and 335702) and cubic in immature skulls (USNM 335478). In D. brachycephala, the mastoid process also is lunate, but the entire process is more laterally extended (Barnes, 1987). The rugosity on the ventrolateral process of the mastoid surface marks the origination of lateral flexors of the head (e.g., sterno- mastoid). In A. kernensis, the mastoid process appears to be less rugose laterally and is cubic in mature skulls (SDSNH 54801); immature skulls have more lunate processes (Figure 7D). Occiput: The occipital condyles are set off from the exoc- cipitals by well-developed ventral condyloid fossae and vari­ ably developed dorsal condyloid fossae (Figure 8B). Ventrally, the articular surfaces of the right and left condyles are either continuous (USNM 335702) or separated by a smooth saddle (USNM 335478 and 250283) (Figure 3). A segment of the exoccipital lies in the saddle between the ventromedial articu­ lar surfaces of the condyles. Dorsally, the condyles tilt between 35° and 40° from the vertical axis. In the cast of the holotype and in USNM 335478, a large circular opening (condyloid ca­ nal) lies on the lateral wall of the foramen magnum. The lambdoidal (nuchal) crests are variably developed in the sample of Emlong skulls, a reflection of different develop­ mental ages. In immature individuals (USNM 335478 and 314645), the lambdoid suture is marked by a low and wide rugose surface that is nearly flush with the surface of the oc­ cipital shield (Figure 3 A ) . In mature male skulls (USNM 335702 and 250283), the rugose surfaces are enlarged into the lambdoidal crests for insertion of the splenius muscle and other neck extensors (Figure 3B). From the interparietal suture the crests diverge posterolaterally to overhang the occipital shield before curving laterally and descending in a keeled an­ teroventral arc to merge with the mastoid processes. In these specimens, the occipital shield is deeply embayed dorsomedi- ally with distinct fossae on either side of the midline for inser­ tion of the rectus capitis dorsalis muscle and other neck exten­ sors. The occipital surface also is distinctly rugose in these mature male skulls. Premaxillary: The anterior termini of the premaxillae in the cast of the holotype and LACM 123285 have weak nasal processes relative to those seen in fossil and living otariids and odobenids. These weak processes are not seen in USNM 335478 (Figure 8A) and probably develop with age. The narial basin (i.e., the region of the rostrum from the incisive margin of the premaxillae to the nasal-premaxilla contact; King, 1972) is not enlarged, and the snout is not elongate. In most species of Allodesmus (A. kernensis, A. sadoensis Hirota, A. sinanoensis Nagao, A. packardi Barnes, and A. naorai Kohno), there are no nasal processes; instead the premaxillae (and maxillae) are de­ veloped into an elongate "prenarial shelf (Kohno, 1996) that serves to enlarge the narial basin. A shallow fossa is positioned on the premaxillary-maxillary suture between the incisive margin and the external narial 126 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY FIGURE 8.—Desmatophoca oregonensis, USNM 335478: A, anterior aspect; B, posterior aspect. opening. Similar fossae occur in Enaliarctos, Pinnarctidion, Prototaria, and Allodesmus. The ascending processes of the premaxillae curve around the external nares and have a short overlap with the nasals medially (Figure 2 A ) . In USNM 335478 and 335451, this overlap is 36% and 27%, respectively, of the total nasal length. The posterodorsal corner of the as­ cending process of the premaxilla, adjacent to the dorsolateral corner of the narial opening, is marked by a small keeled pro­ cess that projects anteriorly. These processes (one right and one left) are especially well developed in LACM 123285 and to a lesser extent in USNM 335451. In USNM 335478, an imma­ ture female skull, the processes are weakly developed, suggest­ ing that they probably increase in size with age and/or gender. The position of these processes suggests that they may repre­ sent attachment sites for ligaments of the nasal cartilages. Nasal: Posteriorly the nasals extend between the frontals in an acute V-shaped termination (Figure 2k), as in Allodesmus kernensis and phocids (Demere, 1994). Medially the frontal- maxillary suture is characterized by a narrow, anteriorly di­ rected, V-shaped segment of the frontal that invades the max­ illa immediately adjacent to the nasals (the anterior process of the frontal after Kohno, 1996). This configuration also occurs in Enaliarctos, Pteronarctos, Prototaria, and Neotherium (Demere, 1994). In ,4. kernensis, the V-shaped frontal segment is indistinct. Lateral to the V-shaped frontal segment in D. ore­ gonensis, the frontal-maxillary suture turns anteroventrolater- ally to descend into the orbit. Anterior to the suture on the dor- somedial side of the maxilla is a low, rugose longitudinal crest that marks the origination of the naso-labialis muscle. This crest is absent in species of Allodesmus and is not preserved in the holotype of D. brachycephala. Maxillary and Zygomatic Arch: The zygomatic arch is hor­ izontally oriented and only slightly arched dorsally at the mid­ point. The maxillary root of the zygomatic arch is at the level of M l - 2 , and it is penetrated by a transversely elliptical in­ fraorbital foramen of moderate size (relative to the larger fo­ ramina in odobenids). As described by Barnes (1987), the ante­ rior dorsal margin of the maxillary root of the zygomatic arch flares anteriorly over the anterior opening of the infraorbital ca­ nal to form an orbital cup for the eye. This structure is seen in most pinnipeds, except for species of Allodesmus, which pos­ sess dorsal margins that are retracted posteriorly from the ante­ rior opening of the infraorbital foramen (Barnes, 1972). The anteroventral margin of the zygomatic arch is elevated above the tooth row, as in species of Enaliarctos and Pinnarctidion, and is not excavated or pitted for origination of the maxillo- naso-labialis muscle. In Allodesmus, the ventral margin of the maxillary root also is smooth, but it is lower and at the level of the tooth row. There is no evidence of a lacrimal bone, even on the imma­ ture female skull (USNM 335478). The anterior border of the orbit is thus made up entirely of maxilla as in all other pinni­ peds (Berta and Wyss, 1994). There are no antorbital processes. The maxillary and jugal meet in a deep foliate suture, with the jugal sending dorsal and ventral processes anteriorly to overlie and underlie the zygomatic process of the maxilla. In USNM 335451, the lateral portions of the dorsal and ventral processes of the jugal overlap and underlap 34 mm and 17 mm, respec­ tively, the posteriorly tapering zygomatic process of the max­ illa. This same construction of the jugal is seen in Allodesmus, odobenids, and otariids. In phocids, there is no ventral process of the jugal. Instead, the jugal has a split-like contact with the maxillary (jugal above maxillary; Muizon, 1981). Posteriorly, in D. oregonensis, the jugal ascends the anterior border of the zygomatic process of the squamosal and terminates dorsally in a sharply pointed and anteroposteriorly compressed postorbital process (Figures 3, 9E,F) . The posterior portion of the jugal NUMBER 93 127 FIGURE 9 (right).—Jugal-squamosal suture pattern in pinnipeds: A, Enaliarctos emlongi; B, Pinnarctidion bishopi (after Barnes, 1979); C, Allodesmuspackardi (after Barnes, 1972); D, Piscoph- ocapaciftca (after Muizon, 1981); E, Desmatophoca oregonen­ sis (female); F, D. oregonensis (male). (j=jugal; s=squamosal.) broadly underlies the zygomatic process of the squamosal in a splint-like suture and extends posteriorly to the anterior border of the glenoid fossa, as in Enaliarctos and Pinnarctidion (Fig­ ure 9A). In Allodesmus, the postorbital process of the jugal is lower than the zygomatic process of the squamosal (Figure 9c), and the posteroventral portion of the jugal is short, not reaching the glenoid fossa. In D. oregonensis and D. brachycephala, the jugal is transversely compressed, as in Enaliarctos, Pinnarctid­ ion, and odobenids, and contrasts with the nearly circular cross section seen in Allodesmus kernensis. In D. oregonensis, the zygomatic process of the squamosal is transversely compressed and variable in dorsoventral thickness. In USNM 335451, the anterior portion of the process is deep and forms a distinctly mortised contact with the jugal (Figures 3B, 9F) . In USNM 335478, the process is narrower (Figures 3A, 9E), which conse­ quently limits the extent of mortising. This variation in mortis­ ing in D. oregonensis, also seen in phocids of different devel­ opmental ages, is suggested to be ontogenetic in the fossil species (see Ronald and Healey, 1981, fig. 9). A mortised con­ tact also is well developed in species of Allodesmus and in pho­ cids (Figure 9c,D). In these groups, however, the zygomatic process of the squamosal is deeper anteriorly adjacent to the ju­ gal. In D. oregonensis, the zygomatic process is deeper posteri­ orly (Figure 9E,F). Frontal: The frontals are narrow along the interorbital con­ striction and preserve small but distinct supraorbital processes or crests for attachment of the orbital ligament (Figure 2). These processes are similarly developed in Enaliarctos, Pin­ narctidion, Allodesmus, and basal odobenids (e.g., Prototaria and Neotherium). In Allodesmus, however, the position of the processes relative to the condylobasal dimension is more poste­ rior than in Desmatophoca (Kohno, 1996). This is most likely correlated with the greatly enlarged orbits of the former taxon. The orbital wall as preserved in USNM 335478 is solid bone and lacks any obvious vacuities. In USNM 335451 and 335702, however, a small horizontally oval opening (9.4 and 9.8 mm in length, respectively) occurs on the orbital wall pos­ terodorsal to the sphenopalatine foramen. In USNM 335702, the ventral margin of this opening is smooth, which suggests that it is a true orbital vacuity rather than a preservational arti­ fact. Its position at the frontal-palatine suture on the orbital wall is similar to that of small vacuities in certain phocids (e.g., Mirounga and Monachus). The oval sphenopalatine foramen in USNM 335451 measures 4.4 mm in length and is positioned 4.5 mm dorsal to the maxillary-palatine suture. The frontal- maxillary suture as preserved in USNM 335478 enters the an­ terodorsal border of the orbit and traverses the ventral one- quarter of the orbital wall. In USNM 335451 and 335702, the suture can be traced to the junction with the palatine bone. In the latter specimen, the frontal-palatine suture appears to be preserved and is marked posteriorly by the small orbital vacu­ ity discussed above. The sutural boundaries of the orbitosphe­ noid cannot be reliably discerned. In USNM 335702, however, a distinct horizontal break in the orbital wall aligns with the predicted position of the ventral margin of this bone. Posteri­ orly, the circular optic foramen, within the orbitosphenoid, lies almost directly anterior and slightly dorsal to the foramen ro- tundum. Both foramina are positioned beneath the anterior end of the braincase, as in Allodesmus kernensis and Pinnarctidion rayi. The medial wall separating the right and left optic foram­ ina is solid and not perforate as in A. kernensis and certain otariids. The alisphenoid canal is merged with the foramen ro- tundum externally, forming a large orbital fissure. In USNM 250283, there is a deeply recessed septum dividing the two fo- 128 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY ramina internally. The lateral wall of the alisphenoid canal is thick and well formed, unlike the thin and variably preserved lateral wall described above for A kernensis. Parietal: The braincase is transversely inflated and has a sharply angled anterolateral margin composed of the frontal dorsally and the alisphenoid ventrally. The coronal suture as preserved in USNM 335478 is transverse dorsomedially and curves slightly anteriorly as it descends to the alisphenoid. Ob­ lique depressions marking the position of the pseudosylvian sulcus occur on the parietal behind the anterolateral corners of the braincase just above the squamosal fossa. In USNM 335478, these depressions are shallow, relative to the deep de­ pressions preserved in USNM 335702. Such depressions also occur in D. brachycephala, Pinnarctidion, and Enaliarctos but are absent in Allodesmus. The parasagittal fossae located on the midline of the dorsal part of the braincase in the holotype of D. oregonensis discussed by Barnes (1987:11) are not present on any of the Emlong Desmatophoca specimens, suggesting that these fossae may be pathologic. Development of a sagittal crest is variable in specimens of Desmatophoca and has not been noted by previous workers. In USNM 335478, the immature female skull, there is no sagittal crest; instead the temporal crests marking the dorsal extent of the temporalis muscle on the parietal are clearly visible and well separated from the interparietal suture (Figures 2A, 3A). These crests converge anteriorly on the interparietal suture and merge with it on the posterior portion of the frontal. This struc­ ture is typical of immature carnivorans and reflects the early stage in the progressive medial migration of the temporal crests, which eventually join in the well-developed sagittal crests of mature individuals. USNM 335451, a mature male skull, has a distinct sagittal crest that measures 10 mm in height at the coronal suture and extends from the posterior portion of the frontal to the broken posterior extremity of the parietal (Figures 2B, 3B). As expected, there are no traces of temporal crests on this specimen (they have merged medially to form the saeittal crestL Petrosal (Inner Ear): Preparation of the cerebral surface of the petrosal in USNM 335451 reveals a distinctly bilobed, dor­ soventrally compressed internal auditory meatus (IAM) similar to that reported in Enaliarctos mitchelli Barnes by Berta (1991), Pinnarctidion bishopi by Barnes (1979), Prototaria planicephala Kohno by Kohno (1994), Imagotaria downsi Mitchell by Repenning and Tedford (1977), Odobenus by Wyss (1987), and fossil odobenines by Demere (1994). Specimens of Allodesmus kernensis (SDSNH 54801 and 54802) show the same type of IAM morphology, suggesting that a bilobed and compressed internal auditory meatus with separate canals for the facial and vestibulocochlear nerves is primitive for pinni­ peds. Internally, the IAM in D. oregonensis is incompletely di­ vided by a low septum that rises from the petrosal but does not reach the meatal roof. The posterior margin of the petrosal is thick and broadly triangular in dorsal aspect, as in A. kernensis. The apex of this triangle aligns with the dorsal lip of the inter­ nal auditory meatus. The floccular fossa is domed and only slightly smaller (USNM 335451, approximately 5.4 mm, as measured from a cast of the middle ear) than the fossa in a sim­ ilarly sized specimen of A. kernensis (SDSNH 54802, 6.8 mm). Petrosal (Middle Ear): The middle ear region as prepared in USNM 335451 and 250283 reveals a small but inflated promontorium (Figure 6A) as in Allodesmus kernensis (SD­ SNH 54802). In Pinnarctidion bishopi and Enaliarctos mealsi Mitchell and Tedford, the promontorium also is inflated (rela­ tive to otariids and odobenids) but slopes gradually anteriorly instead of having an abrupt flexure as in D. oregonensis. A shallow groove crosses the promontorium from the posterolat­ eral corner to the anteromedial corner. This groove also is seen in A. kernensis and was reported by Barnes (1979) for P. bish­ opi. Barnes (1979) suggested that this groove marked the course of the promontory branch of the internal carotid artery. The round window and oval window in USNM 250283 mea­ sure 4.3 mm and 2.2 mm in width, respectively. The former faces posteriorly and the latter faces laterally. The anterolateral wall of the tympanic cavity in D. oregonensis and A. kernensis is smooth and broadly concave. A narrow crest is evident on this wall beginning at the septum between the tympanic open­ ing and the epitympanic recess. This crest extends anterodor­ sally toward the promontorium. Its posterior origination is sim­ ilar in position to that of the tympanic crest in Enaliarctos (Mitchell and Tedford, 1973:228; Berta, 1991) and Pinnarctid­ ion (Barnes, 1979), except that in these taxa the crest maintains a roughly horizontal orientation and does not converge dorsally on the promontorium. It is unclear if these two crests are ho­ mologous. The process of the anterior tympanic septum described by Mitchell and Tedford (1973:228) in Enaliarctos mealsi and by Barnes (1979) in Pinnarctidion bishopi is not present in D. ore­ gonensis and A. kernensis. The taxonomic distribution of this character state is not known in fossil odobenids, otariids, or phocids. The epitympanic recess in USNM 335451 is damaged but appears to have been deep relative to its length; it is small relative to the large and hemispherical condition in a specimen of A. kernensis (SDSNH 54802) of similar size. Endocranium: Special preparation of USNM 250283 re­ veals the right half of an endocranial cast. This cast preserves a broad, deep, and vertically oriented pseudosylvian sulcus. This differs from the condition in Enaliarctos mealsi, which has a narrower and shallower sulcus (Mitchell and Tedford, 1973). Anterior to the pseudosylvian sulcus the cerebrum of D. ore­ gonensis is more elongate than in E. mealsi. Although more morphology is represented in USNM 250283, further descrip­ tion of the braincase must await comparative study of other fossil and Holocene pinniped endocrania. Upper Dentition: The upper incisor alveoli are arranged in an arc and represent 11-3 (Figure 5B). As judged from the alve­ oli, the roots of 11,2 were transversely compressed. The crowns of 11,2 are not preserved in any of the skulls of D. oregonensis. NUMBER 93 129 FIGURE 10.—Desmatophoca oregonensis, USNM 335451, left tooth row and palate (stereophotographs). (Scale bar=2 cm.) The root of 13 is larger than the roots of the medial incisors (Figure 4A). The tooth is procumbent and has an ovoid cross section. As preserved in the cast of the holotype and in LACM 123285, the crown of 13 is a single conical, anterolaterally di­ rected cusp with smooth enamel. In USNM 243694, the crown of 13 is heavily worn, presumably from occlusion with the lower canine. With the exception of the young female skull (USNM 335478), a short (4.4-6.6 mm, x=5.8 mm, n=3) di­ astema is present between 13 and the canine. Similar diastemata are present in species of Enaliarctos and Allodesmus, as well as inD. brachycephala. The canine in USNM 335478 is shorter and more slenderly proportioned than in the cast of the holotype or in USNM 243694 (Table 2). A longitudinal crista is prominent posteriorly in USNM 243694, 335451, and 335478. In USNM 314645, 335451, and LACM 123285, a second crista occurs on the an- teromedial side of the crown and merges tangentially with the base of the crown. The enamel on the canine is typically smooth except in LACM 123285 where the posteromedial cor­ ner of the crown between the anteromedial and posterior cristae is marked by dorsoventrally elongate enamel "wrinkles." In USNM 243694 and 314645 apical wear is present. In addition, USNM 243694 and 335478 preserve vertical wear facets de­ veloped anteromedially on the canine crown, presumably by occlusion with the lower canine. The postcanine teeth possess smooth enamel crowns that have been described by other workers as bulbous. Presumably, this description means that the crowns are swollen and rounded relative to the more secodont dentitions of Enaliarctos and Pin­ narctidion. By this definition of bulbous, the postcanine teeth of Desmatophoca are less bulbous than those of species of^4/- lodesmus. With the exception of Pl and M2, all of the upper postcanine teeth are double rooted. Pl is single rooted and its crown, as preserved in USNM 335451 and 335478, has a single principal cusp and anterior and posterior cristae (Figure 10). The cingulum is nearly cir­ cumferential except for a small area on the labial side of the crown above the principal cusp. The lingual margin of the cin­ gulum is crenulated with at least five unequally sized crena- tions. Wear facets are developed on the posterolingual portion of the tooth at the base of the crown. The root of Pl is long (19.6 mm in USNM 335478) and closely appressed to and un­ derlying the canine root. The crown of P2 has a single large principal cusp with a tiny cuspule on its posterior flank (Figure 10). A weak crista ex­ tends from this cuspule to the apex of the crown, which is slightly posteriorly directed. As preserved in USNM 335451, 335478, and 335702, the crenulated cingulum (with up to nine crenations) is developed only lingually. USNM 314645 differs from other specimens in possessing an additional small cuspule on the anterior margin of the principal cusp just ventrad of the cingulum. Large vertical wear facets are seen on the posterolin­ gual margins of P2 in USNM 243694 and 335451. In USNM 314645 and 335478 the anteroposterior axis of P2 is oblique to the tooth row. In all other specimens P2 has a more anteropos- 130 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY terior alignment. The distinct anterior and posterior roots are circular in cross section and are of nearly equal size. The crown of P3 consists of a single large principal cusp with weak anterior and posterior cristae (Figure 10). The prin­ cipal cusp has smooth enamel and is about the same size as P2 but is less apically pointed. A tiny cuspule is positioned on the posterodorsal flank of the principal cusp. The lingual margin of the crown base is marked by a continuous and crenulated cin­ gulum (with up to nine unequal-sized crenations), which in oc­ clusal aspect is expanded posterolingually. This expansion is correlated with the transversely elongate and superficially bi­ lobed posterior root. The separate anterior root is smaller and nearly circular in cross section. The crown of P4, as seen in USNM 335451 and 335478, has a lower and more bulbous principal cusp, a larger posterior cus­ pule, and a stronger posterior crista than P2,3 (Figure 10). In occlusal aspect, the principal cusp is not as expanded postero­ lingually, and the lingual cingulum is less crenulated (five crenations on the posterolingual corner). Wear facets are devel­ oped on the posterior edge of the crown. The tooth has two sep­ arate roots; the posterior one is transversely bilobed, whereas the smaller anterior root is nearly circular in cross section. A short (4.5-6.5 mm; x=5.7 mm, n=4) diastema separates P4 from M1. The crown of Ml is triangular to quadrate in occlusal aspect and smaller than in Pl—4 (Figure 10). The paracone, metacone, and protocone are low-crowned and separate in USNM 335702; there is no talon. In USNM 335478 and 335451, the paracone is larger than the metacone and is connected to it by a crest. The protocone is indistinct. A well-developed cingulum nearly encircles the crown in USNM 335478 but is confined to the anterolateral margin in USNM 335702. The number of roots is variable in Ml: Ml is single-rooted in USNM 243694, 335451, and probably 335702; weakly bilobed in USNM 314645; and distinctly double rooted as shown by alveoli in the cast of the holotype and LACM 123285 (Barnes, 1987). This degree of variability may herald the single-rooted condition seen in divergent species of Allodesmus. A short (3.7-6.0 mm; jt=4.8 mm, n=4) diastema separates Ml from M2. An isolated right M2 is associated with USNM 335478 and preserves a circular crown (in occlusal aspect) with a low, cen­ trally placed cusp. A cingulum extends from the posterior edge around the lingual margin to the anterior edge. A faint crista extends medially from the crown apex to the lingual margin of the cingulum. The tooth has a single, short, transversely com­ pressed root. The presence of M2 is variable in Desmatophoca oregonensis; it occurs in USNM 335451, 335478, 335702, and LACM 123285 and is absent in USNM 243694 and 314645. The postcanine upper dentition of Desmatophoca oregonen­ sis differs from that of species of Enaliarctos and Pinnarctidion in having (1) principal cusps that are bulbous rather than blade­ like, (2) small posterior cuspules on the principal cusps rather than large cuspules, (3) anterior and posterior cristae that are faint rather than sharp, and (4) lingual cingula that are finely crenulated rather than smooth. The derived states of these char­ acters appear to be characteristic of D. oregonensis. Relative to species of Allodesmus, however, the condition in D. oregonen­ sis is primitive for many of these characters. The bulbous crowns of the postcanine dentition of Allodesmus consist of only the large principal cusp without cuspules, cristae, cingula, or crenations. In addition, P2-M2 are all single rooted in A. kernensis, A. sadoensis, A. packardi, and A. naorai. The dou­ ble-rooted condition reported (Barnes, 1972) in A. courseni is actually only observable on p2,3 and cannot be confirmed for the upper dentition. Mandible: The Emlong collection contains seven mandi­ bles herein referred to Desmatophoca oregonensis. This mate­ rial provides information, previously unreported, on the mor­ phology of the lower jaw and lower dentition. The new specimens include right and left dentaries associated with the mature female skull (USNM 335478), a partial left dentary as­ sociated with the mature male skull (USNM 335451), and right and left dentaries of a mature male individual (USNM 335430). Table 2 summarizes measurements of dentaries and individual teeth for the five most complete specimens. The largest speci­ men (USNM 335430) is compatible in size with the holotype skull. The postcanine teeth with this specimen have large wear facets on the anterior and posterior crown margins and emer­ gent and distally flared roots. The ramus is deep relative to that in species of Enaliarctos and has a teardrop-shaped, unfused, bony symphysis (Figure 11B). The symphysis extends in most specimens to the level of the posterior root of p2 and is marked in older individuals (USNM 335430) by a thickened genial tuberosity. Mental fo­ ramina are positioned on the midline of the jaw, one below p3 and two to three smaller ones located below p2 and pl (Figure 11A). In lateral aspect, the ventral margin of the horizontal ra­ mus is straight (not curved) and extends posteriorly as a con­ spicuous bony flange (the marginal process of Davis, 1964) for insertion of the digastricus muscle. This flange provides the mandible with a distinctly angled posterior margin as is seen also in species of Allodesmus, as well as in phocids. Enaliarc­ tos, Pteronarctos, and basal odobenids lack a well-developed marginal process. A shelf-like, anteroposteriorly elongate an­ gular process is positioned above the marginal process and marks the area of insertion for the internal pterygoid muscle. This process is reduced and not shelf-like in species of Allodes­ mus. The mandibular condyle in D. oregonensis is transversely elongate and conspicuous in its elevated position above the tooth row. The condyle is closely appressed to the posterior margin of the coronoid process and does not extend as posteri­ orly as it does in Allodesmus kernensis. The coronoid process is broad and low with a rounded dorsal margin (Figure 11 A). The masseteric fossa is shallow with a smooth surface. The small mandibular foramen is positioned directly above the mar­ ginal process at the level of the angular process. In occlusal as­ pect the mandible has a weakly sinuous outline, broadly con­ cave laterally between c 1 and p4 and broadly convex laterally NUMBER 93 131 FIGURE 11.—Desmatophoca oregonensis: A,B, USNM 335478, left mandible: A, lateral aspect; B, medial aspect (stereophotographs). C, USNM 335457, left mandible, occlusal aspect (stereophotographs). D, USNM 335478, left dentary, occlusal aspect (stereophotographs). (Scale bars in A and B=5 cm; scale bars in C and D=2 cm.) between p4 and the mandibular condyle. The postcanine alve­ oli are closely appressed and lack diastemata. Lower Dentition: The lower dentition consists of i2?,3, cl, pl-4, ml,2. Variation in this formula is seen in USNM 335430, which lacks all incisive alveoli, and in USNM 335478, which lacks alveoli for right and left m2 (compare C and D of Figure 11). The incisive region of USNM 335694 has an anterolaterally 132 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY elongate and bilobed alveolus that suggests the presence of two incisors, presumably, i2,3 (i2 posterior to i3). No specimens preserve i2. The left i3 is preserved in USNM 335478 (Figure 1 ID). The root is transversely compressed in cross section. The crown consists of a well-formed central cusp that is triangular in anterior aspect, and it has a weak and incomplete lingual cin­ gulum. A small wear facet on the medial side of the cusp indi­ cates occlusion with II or 12. The lower canine is slenderly proportioned in the young fe­ male specimen (USNM 335478) and preserves a large vertical wear facet positioned posterolaterally and extending from the apex to the middle of the crown (Figure llD). A weak vertical crista is present posteriorly on the thickly enameled crown. The canine teeth in USNM 335430, 335457, and 335694 (all pre­ sumably male specimens) are longer with broader bases (Table 3). The latter two specimens also preserve weak posterior cris­ tae. Variation in crown wear noted within the sample includes a single apical wear facet (USNM 335694), apical and postero­ lateral wear facets, and no wear facet (USNM 335457). The root of c as preserved in USNM 335430 (presumably an old male) is distinctly fluted with longitudinal furrows and ridges. The postcanine teeth are closely positioned in the tooth row. Except for p 1, the premolars possess two separate roots that are circular in cross section. The crowns of p2-ml are character­ ized by strong principal cusps with convexly curved anterior margins and concavely curved posterior margins. In labial and occlusal aspects, the enamel surfaces of the principal cusps are rounded and swollen and not secodont as in Enaliarctos and Pinnarctidion. The crown of pl is nearly circular in occlusal aspect and con­ sists of a small, low, and rounded principal cusp positioned slightly anterior of center (Figure 11D). Two low, tiny cuspules are located on the labial cingulum, one anterior to the principal cusp and the other posterolateral to it. The lingual portion of the crown is broad and encircled by a cingulum. The tooth has a single conical root and is closely appressed to the canine root. The principal cusp of p2 is larger and more apically pointed than on p 1 and is flanked by a low and tiny anterior cuspule po­ sitioned on the anterior corner of the lingual cingulum (Figure 1 lC,D). A second tiny cuspule is positioned on the posterior margin of the cingulum in line with the principal cusp. The lin­ gual cingulum is continuous from the anterior cuspule to the posterior cuspule and is irregularly crenulated especially at the posterolingual corner. There is no labial cingulum. As seen in USNM 335457 and 335694, p2 has two separate roots that are circular in cross section. Large oblique wear facets on the p2 in USNM 335430 occur on both the anterior and posterior mar­ gins of the tooth and expose the dentine core. The crown of p3 also is characterized by a large, apically pointed principal cusp. Relative to p2, the anterior cuspule is taller and more conical and is positioned more lingually on the anterior margin of the lingual cingulum (Figure llC,D). A tiny cuspule is positioned on the posterior corner of the cingulum. As preserved in USNM 335457, this posterior cuspule appears to be just the largest in a series of fine crenations that mark the posterior portion of the lingual cingulum. This cingulum, as preserved in USNM 335457, is crenulated, the crenation on the posterolabial corner being the largest. In other specimens (USNM 314536 and 335478), the cingulum is smooth. Be­ tween the tiny posterior cingular cuspule and the principal cusp is a large conical cuspule whose base is not on the cingulum but rather is positioned on the posterolingual flank of the pri­ mary cusp (Figure 11B). There is no labial cingulum. The tooth has two separate roots that are circular in cross section. The crown morphology of p4 is similar to that of p3 except that the principal cusp is lower and the conical posterior cus­ pule occurs farther up its posterolingual flank (Figure llC,D). The lingual cingulum is well developed and in USNM 335457 has a finely crenulated margin. The tall and conical, anterior cuspule at the anterior corner of the lingual cingulum is well separated from the principal cusp and is positioned on its anter- omedial flank. The tooth has two separate roots that are circular in cross section. The crown of ml differs from the premolar crowns in lack­ ing a well-developed and continuous lingual cingulum, in pos­ sessing a lower principal cusp, and in lacking the prominent conical cuspule on the posteromedial flank of the principal cusp (Figure l i e ) . Instead, there is a prominent, transversely elongated, posterior cuspule located in the position of the small, posterior, cingular cuspule of p3,4. In one specimen (USNM 335457), a slight swelling on the posterior flank of the principal cusp may be homologous with the conical posterior cuspule of p3,4. Anterior to the principal cusp is a well-devel­ oped, conical cuspule of slightly larger size than the cuspule lo­ cated on the posterior extremity of the crown. The bases of these cuspules are positioned on incomplete lingual cingula. The tooth has two separate roots that are circular in cross sec­ tion. None of the mandibles studied preserves the crown of m2, and of the seven mandibles available only one (USNM 335478) lacks alveoli for this tooth. Of the six specimens with m2 alveoli, five preserve a single oval alveolus and one (USNM 335457) preserves two separate circular alveoli (Fig­ ure l ie ) . This degree of variation is reminiscent of the condi­ tion in modern otariids (King, 1983). The lower dentition of Desmatophoca oregonensis differs from that of all other pinnipeds in possessing the combination of distinctly conical anterior and posterior cuspules, finely crenulated lingual cingula, and broadly rounded principal cusps. Atlas: USNM 335478 includes a nearly complete atlas ver­ tebra (Figure 12A,B). This element measures 68 mm across the widely spaced anterior articulation and 105 mm across the nearly complete transverse process. The anterior articular fac­ ets are deeply concave and are separated medially by a 12.4 mm wide ventral notch. The posterior articular facets are dam­ aged but appear to have been relatively planar. There is no dis­ tinct fovea for the dens of the axis vertebra. There is also no tu- NUMBER 93 133 FIGURE 12.—Desmatophoca oregonensis, USNM 335478, atlas vertebra: A, dorsal aspect (stereophotographs); B, anterior aspect (stereophotographs). C,D, USNM 335246, left scapula: C, anterior aspect; D, lateral aspect (ste­ reophotographs). (Scale bars=5 cm.) bercle on the ventral border of the atlas vertebral body. The transverse processes descend from the body of the vertebra at an angle of 53° to the horizontal plane. In dorsal aspect the pro­ cesses diverge laterally from the sagittal plane at an angle of 50° to 60°. The matrix-filled, left lateral vertebral foramen for passage of the vertebral artery measures approximately 9 mm wide by 11 mm high. The transverse foramen for passage of the vertebral artery exits the posteromedial side of the transverse process above the ventral border of the posterior articular facet. A deep, oval fossa on the ventral side of the transverse process marks the exit of the artery. In comparison with the atlas verte­ bra of Allodesmus kernensis (Kellogg, 1931, figs. 9-12), USNM 335478 is actually smaller, the dorsal surfaces of the transverse processes are planar rather than broadly excavated, and the transverse foramen is positioned more ventrally on the posterior side of the transverse process. Scapula: USNM 335246 includes a nearly complete left scapula with a broken vertebral border (Figure 13c,D). Referral of this element to D. oregonensis is based upon its association with humeri showing strong affinities with the referred hu­ merus of Packard (1947; see below). Overall, the blade is otariid-like in shape and lacks the transverse elongation of pho- 134 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY FIGURE 13.—Desmatophoca oregonensis, USNM 335246, left humerus: A, anterior aspect (stereophotographs); B, posterior aspect (stereophotographs); c, medial aspect (stereophotographs). (Scale bars=5 cm.) cine scapulae and the longitudinal elongation of odobenine scapulae. The supraspinous fossa is very large, roughly twice the size of the narrow, triangular infraspinous fossa. The latter shows no obvious morphology indicating the origins of the in­ fraspinatus or teres minor muscles. The scapular spine is keel­ like, is not thickened apically, and is strongly tilted caudally to NUMBER 93 135 overhang the cranial portion of the infraspinous fossa (Figure 13B). The axis of the spine meets the glenoid fossa at an angle of approximately 40° from the perpendicular. The acromion is large, flange-like, curved caudally, and positioned on the proxi­ mal edge of the spine. In Allodesmus kernensis, the acromion is blunt and knob-like and does not overhang the infraspinous fossa (Mitchell, 1966). The caudal angle of the blade has a dis­ tinct flange-like process for the teres major. This process is not apparent on the scapula of Allodesmus illustrated by Mitchell (1966, pl. 15). The caudal border is distinctly raised and keeled on its lateral margin and not blunt as in A. kernensis. The gle­ noid fossa is deeply embayed. The coracoid process is robust and rounded, and on the anteromedial corner it shows a slight indication of a bump for the origination of the biceps muscle. The medial surface of the scapular blade is broadly concave and marked by at least five low longitudinal ridges with inter­ vening broadly concave furrows. Humerus: USNM 335246 includes both humeri. These are referred to Desmatophoca oregonensis on the basis of their overall size and morphological similarity to the left humerus referred to D. oregonensis by Packard (1947). The proximodis- tal length of the right humerus of USNM 335246 is 218 mm, about 10 mm longer than the referred humerus of Packard. The greater tuberosity is higher than the lesser tuberosity and equal to the level of the humeral head (Figure 13A-C). The deltopec­ toral crest is transversely broad and longitudinally elongate and meets the humeral shaft proximal to the level of the entepi- condyle. The longitudinal axis of the deltopectoral crest is di­ rected toward the middle of the distal trochlea. The area for in­ sertion of the deltoid muscle is large and well defined, extending along the entire lateral border of the crest (Figure 13A). The distal end of this insertion is developed as a laterally projecting knob. Caudal to the deltoid insertion is a conspicu­ ous sulcus for origination of the brachialis muscle. The bicipi­ tal groove is deep and broadly open. The humeral head is rounded and extended caudally. The entepicondyle is robust and extends medially to produce a broad distinct concave out­ line to the humeral shaft. The supinator ridge is indistinct. The radial capitulum is broader than the medial ulnar lip of the dis­ tal trochlea. Mitchell (1961:15) briefly discussed the morphol­ ogy of the referred left humerus of Packard (1947), mentioning its slender proportions, laterally bowed shaft, and high degree of shaft torsion. Comparison of a cast of this humerus with the humeri of USNM 335246 suggests that the bowing and torsion are the result of diagenetic distortions. Humeri of Allodesmus kernensis display many features distinct from the humeri of Desmatophoca oregonensis, including the following: the over­ all size is larger; the humeral head is higher than either tuberos­ ity; the bicipital groove is shallow and open; the supinator por­ tion of the distal end is stronger, resulting in a more central position of the shaft relative to the entepicondyle and the ect- epicondyle; the medial portion of the distal ulnar condyle is not flared as a lip; the deltopectoral crest is relatively longer and meets the shaft at the level of the entepicondyle; and the del­ topectoral crest is laterally inclined and directed toward the medial margin of the ulnar facet. The taxonomic distribution of these various character states was not investigated for the present report. Ulna: A nearly complete right ulna (Figure 14A-C) miss­ ing only the extreme distal end and styloid process is associ­ ated with the scapula and humeri of USNM 335246. As pre­ served, the proximodistal length of the ulna is 208 mm, which suggests that the complete element would likely have been longer than the associated humerus. The proximal end is trans­ versely compressed and is developed anteroposteriorly into the broad fan-shaped olecranon process characteristic of all pinni­ peds (Howell, 1929). The proximal crest of the olecranon pro­ cess for insertion of the triceps muscle is broadly curved in lat­ eral aspect and terminates in a distinct posterior or ulnar process (Howell, 1929); the olecranon crest in Allodesmus is relatively shorter and less falcate and has a broader triceps in­ sertion proximomedially. Anteriorly, the height of the olecra­ non crest above the trochlear or sigmoid notch is 45 mm. The lateral surface of the olecranon process has a low longitudinal ridge adjacent to the midpoint of the olecranon crest. This ridge blends distally with the main area of the extensor fossa and pre­ sumably marks the division between the extensor metacarpi pollicis and the extensor pollicis longus. Its position is more posteriorly placed than in modern otariids (Howell, 1929). The anterior portion of the extensor fossa adjacent to the sigmoid notch is longitudinally excavated into a conspicuous depres­ sion. The distal terminus of this depression lies at the proximal edge of a raised, oblong rugosity on the anterolateral margin of the ulnar shaft. This rugosity, as discussed by Howell (1929), is the site in Phoca of origination of a strong interosseous liga­ ment that binds the ulna to the radius. Well-developed rugosi­ ties like these are not recorded for Allodesmus kernensis. The shaft in the distal region of the ulna is transversely compressed into an elongate oval. The medial surface of the olecranon process is broad for origination of the flexor musculature. This flexor fossa is flat­ tened, as in Allodesmus kernensis, and not deeply concave, as in modern otariids. The greater sigmoid cavity for articulation with the humerus is taller than it is broad and has a uniformly convex trochlear facet. The lesser sigmoid cavity for articula­ tion with the radius is lunate in shape and only slightly con­ cave. Ulnae of Allodesmus kernensis are larger than in USNM 335246, have a relatively shorter and less falcate olecranon crest, and have a more quadrate greater sigmoid cavity. Desmatophoca brachycephala Barnes, 1987 EMENDED DIAGNOSIS.—A species of Desmatophoca with widely flared zygomatic arches, posterior portion of squamosal fossa smaller than anterior fossa (autapomorphies of D. brachycephala), paroccipital process oriented at an angle greater than 50° to the sagittal plane (possible autapomorphy), angle formed by long axis of external auditory meatus and sag­ ittal plane of skull greater than 62°, no prenarial shelf, distinct 136 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY FIGURE 14.—Desmatophoca oregonensis. USNM 335246, right ulna: A, anterior aspect (stereophotographs); B, medial aspect (stereophotographs); C, lateral aspect (stereophotographs). (Scale bars=5 cm.) carinae on C, and thick lateral wall of alisphenoid canal (sym- plesiomorphies shared with basal pinnipeds). HOLOTYPE.—LACM 120199, incomplete skull with parts of both canines and crowns of left 11-3 and Ml, collected by J.L. Goeddert and G.H. Goeddert in 1979 (Barnes, 1987). TYPE LOCALITY.—LACM 4584, east of Knappton, Pacific County, Washington. DISCUSSION.—Desmatophoca brachycephala is represented solely by the holotype skull, which is damaged in the rostral re­ gion and lacks all postcanine teeth except the left Ml. Pre­ served alveoli include those for the right Pl, P4, and Ml and the left Pl (lingual margin only), P3 (posterior root only), P4, and Ml,2. The size and morphology of the right and left ca­ nines is extrapolated from a small preserved portion of the dor- NUMBER 93 137 solingual surface of the crowns. The well-developed lambdoi­ dal crests suggest that the skull belongs to a mature male individual. Barnes (1987) listed 12 characters to distinguish D. brachycephala from D. oregonensis. Subsequent to the de­ scription of D. brachycephala, the remainder of the right palate with the M2 alveolus intact was discovered, negating one of the characters (M2 absent) initially used to diagnose this taxon (L.G. Barnes, pers. comm., 1998). Of Barnes's remaining 11 characters, seven are not exclusive to D. brachycephala but are also characteristic of D. oregonensis given the larger sample now available. Those characters determined not to be diagnos­ tic of D. brachycephala include the following: the pterygoid process beneath the orbit is larger (also seen in USNM 335702, 335478, and 335451); the interorbital region is wider, espe­ cially in the anterior part (also seen in USNM 335451 and 335702); the zygomatic arch is more slender (also seen in USNM 335478; as discussed above, this character is age and sex dimorphic); the optic foramina are located more poster- oventrally within the orbits (variable in D. oregonensis, simi­ larly positioned in USNM 335478 and 335451); and the poste­ rior lacerate foramen is larger and more circular in outline (variable in D. oregonensis, similar size and shape in USNM 335451). Another purported diagnostic character that we could not confirm given the larger sample is the size and orientation of the mastoid process. According to Barnes (1987:6) the mastoid process is "larger and extended more posterodorsally" in D. brachycephala. This character is actually a combination of both size and orientation of the mastoid process, so our discus­ sion is divided accordingly. Although mastoid size has not been quantified, in a qualitative sense it is clear that some specimens of D. oregonensis (e.g., USNM 335702) possess mastoid pro­ cesses that are the same size as those of D. brachycephala. Size, therefore, is not a distinguishing feature. Orientation of the mastoid process involves at least two as­ pects, lateral orientation of the long axis of the process as viewed in ventral aspect and vertical orientation of the rugose lateral face of the process as viewed in posterior aspect. The former is correlated with orientation of the EAM and, as dis­ cussed below, D. brachycephala has an EAM (and mastoid process) that are directed more laterally than anterolaterally. Orientation of the rugose lateral face of the mastoid process is generally the same in both species, although D. brachycephala tends to have a broader lateral surface area compared with the more ventrally placed surface ofD. oregonensis. Three of Barnes's characters purported to be diagnostic for D. brachycephala were not quantified by Barnes and, when quantified by us, could not be confirmed as dichotomous: (1) the skull has a shorter and wider rostrum, more expanded later­ ally around the canines, and the palate is wider, especially in posterior part; (2) the external auditory meatus is wider and di­ rected more laterally, and (3) the paroccipital process is di­ rected more laterally, instead of posteriorly. We redefined these characters so that they could be quantified. Rostrum length is defined herein as the distance from the an­ terior border of the orbital rim to the prosthion. Comparing measurements of rostrum length to CBL we found that in both Desmatophoca oregonensis and D. brachycephala this mea­ surement was 28% of CBL, thus failing to confirm the dichoto­ mous nature of this character. A further modification of ros­ trum size is the degree of lateral expansion around the canines. As discussed by Barnes (1987), D. brachycephala is distin­ guished from D. oregonensis by its relatively short rostrum that flares around the roots of the canine teeth. The holotype of the former is missing the lateral portion of the maxilla, however, so it is not possible to precisely determine the extent of expansion of the rostrum around the canines. Palate width was measured at two points—the interalveolar distance at P1 and at M1 (lin­ gual borders)—and a ratio of the former to the latter was calcu­ lated. Thus quantified, a uniformly narrow palate would have a ratio close to 1.00, whereas a posteriorly broadened palate would have a ratio closer to 0.50. In D. oregonensis, the palate width ratio is 0.44-0.54 (x=0.49, n=4) compared with 0.42 in D. brachycephala. This ratio is 0.46 in Allodesmus packardi, 0.60 in Allodesmus kernensis (LACM 4320), and 0.70 in Enali­ arctos emlongi. Quantification of this character reveals that D. brachycephala has only a slightly broader palate than D. ore­ gonensis. Unfortunately, the sample size is not large enough to determine if this difference is significant. The orientation of the EAM was quantified by measuring the angle that the axis of the EAM makes with the sagittal plane of the skull in ventral aspect. Desmatophoca brachycephala pos­ sesses the primitive condition also seen in outgroup taxa, in which the axis of the EAM is oriented at an angle greater than 62° from the sagittal plane of the skull. In D. oregonensis, this angular measurement is less than 62° (see Appendix, character 19), thus confirming the characterization of Barnes (1987). This is also true for Barnes's identification of the more lateral orientation of the paroccipital process in Desmatophoca brachycephala. This character was quantified by measuring the angle between the long axis of the paroccipital process and the sagittal plane of the skull in ventral aspect. Desmatophoca ore­ gonensis possesses the primitive condition also seen in out­ group taxa, in which the paroccipital process is oriented at an angle less than 25° from the sagittal plane of the skull. In D. or­ egonensis this angular measurement is 17°-22° (x=20°, n=4), whereas in D. brachycephala the angle is 54° and in A. kernen­ sis, 52°-62° (JC=57°, «=3). Although only two of Barnes's purported dichotomous char­ acters for Desmatophoca brachycephala were confirmed (EAM oriented at an angle greater than 62° from the sagittal plane, and paroccipital process oriented at an angle greater than 50° from the sagittal plane), the present analysis recognizes an additional apomorphy useful in diagnosing this taxon: zygo­ matic arches more widely flared (see also Appendix, character 15). Zygomatic width, defined relative to CBL, was 65% of 138 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY condylobasal length in D. brachycephala, compared with val­ ues ranging from 54% to 60% in other pinnipeds. Phylogenetic Relationships of Desmatophoca As previously mentioned, recent hypotheses for higher-level relationships of species of Desmatophoca can be summarized as two major alternative topologies (Figure 1). In one arrange­ ment, proposed by Barnes (1989), species of Desmatophoca are more closely related to odobenids than to other "otarioids," including Allodesmus. In the other arrangement proposed by Repenning and Tedford (1977), species of Desmatophoca and Allodesmus are sister taxa. Building on this lattter arrangement, Wyss and Flynn (1993), Berta and Wyss (1994), and Berta (1994) further suggested that the Desmatophoca + Allodesmus clade is more closely related to phocids than to odobenids. To analyze the phylogenetic relationships of species of Des­ matophoca, we evaluated 30 cranial and dental characters (20 binary and 10 multistate) (Table 4). We used the branch-and- bound search option of PAUP*4.0b8 (Swofford, 1998) with all multistate characters treated as unordered. Using the higher- level relationships summarized above, we considered the fol­ lowing as ingroup taxa: Desmatophoca oregonensis, Desmato­ phoca brachycephala, Pinnarctidion rayi, Pinnarctidion bish­ opi, Allodesmus courseni, Allodesmus sadoensis, Allodesmus packardi, Allodesmus kernensis, and the Phocidae (basal phoc­ ids Homiphoca capensis Hendey and Repenning, 1972, and Piscophoca pacifica Muizon, 1981, and a living basal mona- chine, Monachus schauinslandi Matschie). We used the follow­ ing taxa as successive outgroups: Odobenidae (basal odobenids Proneotherium repenningi and Prototariaplanicephala Kohno, 1994), Otariidae (basal otariid Thalassoleon mexicanus Repen­ ning and Tedford, 1977), Pteronarctos goedertae Barnes, 1989, and Enaliarctos emlongi Berta, 1991. The cladistic analysis produced five equally parsimonious trees that differed only in the relationships among outgroup taxa. A strict consensus tree yielded a tree length of 49 steps and a rescaled consistency index of 0.785 (Figure 15). A clade consisting of species of Allodesmus and Desmatophoca was consistently recognized as the sister group to phocids. An un­ expected result (not figured) was the consistent position of the Otariidae (not Odobenidae) as the sister taxon to the ingroup. This arrangement can be explained if one considers that basal otariids (i.e., Thalassoleon) possess character states that are de­ rived relative to those of the earliest identified odobenids (i.e., Proneotherium and Prototaria). Another contributing factor for the unusual sister-group placement of otariids involves the lim­ ited character data set employed, which because of the focus of our study does not include many characters relevant to inter­ preting the relationships of otariids. For a more inclusive study of pinniped relationships, see Berta and Wyss (1994). The position of Pinnarctidion as a member of the Phocoidea clade (see Figure 1, middle and bottom) was not confirmed us­ ing this character data set. In this analysis Pinnarctidion was variously placed among outgroup taxa, and we do not consider it to be a member of the ingroup. On the basis of this analysis, we recognize the monophyletic groups/taxa presented below together with our evaluation of previous arrangements of these clades/taxa. Bootstrap percent­ ages based upon 500 replicates were calculated and are sum­ marized in Figure 15. With one exception (the D. oregonensis- D. brachycephala clade), taxonomic associations are well sup­ ported. Phocoidea (Allodesmus+Desmatophoca+Phocidae): This monophyletic group is diagnosed by three unequivocal synapo­ morphies (Figure 15): premaxilla-nasal contact reduced (1), squamosal-jugal articulation mortised with zygomatic process expanded anteriorly (transformation to state 1 in Desmatophoca) (11), and flange (=marginal process) below ascending ramus well developed (22). An additional five equivocal synapomor­ phies are potentially diagnostic of this clade: paroccipital pro­ cess enlarged and not excavated (9), posterior portion of squa­ mosal fossa smaller than anterior portion (18), angular process reduced or absent (21), and p 2 ^ crowns with conical cuspules (transformation to state 2 in Allodesmus) (27). Barnes and Hirota (1995) argued against the validity of rec­ ognizing the Phocoidea clade (Allodesmus+Desmatophoca+ Pinnarctidion + Phocidae) as proposed by Berta and Wyss (1994). As evidence, they refuted the seven synapomorphies used by Berta and Wyss to support this clade. We respond to their conclusions in the following discussion. 1. Nasals extending posterior to frontal-maxillary contact. We have modified the definition of this character following Demere (1994). The significant morphology is the shape of the nasal-frontal suture and not merely its posterior position. The derived condition, a narrowly acute, V-shaped nasal-frontal su­ ture is seen in Desmatophoca, Allodesmus, and phocids (see Appendix, character 2). Dusignathine odobenids possess V- shaped nasal-frontal sutures (Demere, 1994), but the sutures are broadly V-shaped and are not posteriorly placed. Pinnarc­ tidion was incorrectly coded by Berta and Wyss (1994) and ac­ tually has the primitive condition in which the suture is blunt and nearly transverse. This character thus does not support a close relationship between Pinnarctidion and the Phocoidea. It does, however, support the grouping of Desmatophoca, Al­ lodesmus, and phocids. 2. Pterygoid process with flat, concave lateral margin. Barnes and Hirota (1995) considered this character actually to be a combination of two characters—the form of the pterygoid strut (i.e., rounded and convex; excavated; or flat and rugose) and the arrangement of the bones (i.e., palatine, pterygoid, and alisphenoid) making up the strut. Although this distinction is valid, Berta and Wyss (1994:48) were only concerned with the form of the pterygoid strut, which they noted was flat with a concave lateral margin in Pinnarctidion, Desmatophoca, Al­ lodesmus, and phocids. We have determined that this is not a useful character at this taxonomic level (also see below), but we disagree with the specific reasons given by Barnes and Hi- NUMBER 93 139 DESMATOPHOCIDAE 15,24 PHOCOIDEA 1, 11,12,23 FIGURE 15.—Phylogenetic hypothesis of phocoid and desmatophocid relationships. (Plain numbers indicate character support for selected nodes; numbers in bold are bootstrap values discussed in text; dashed lines for Allodesmus courseni branch were added a posteriori, as discussed in text.) rota. The implication that in all phocids there is no well-formed hamulus and that the palatine does not contact the alisphenoid bone is incorrect (see for example Mirounga). 3. Mastoid process distant from paroccipital process. Ac­ cording to Barnes and Hirota, this character was neither cor­ rectly described nor quantified. Although Berta and Wyss did oversimplify the complex states for this character and did not quantify it, the discussion offered by Barnes and Hirota also was oversimplified and also failed to quantify the character. We agree that the character is complex and requires a more detailed description (see Figure 7). 4. IAM absent and canals for vestibulocochlear nerve com­ pletely separated. According to Barnes and Hirota, this charac­ ter was incorrectly described and attributed by Berta and Wyss (1994). As reevaluated in this study, a bilobed IAM with sepa­ rate canals for cranial nerves VI and VII occurs in Enaliarctos, Pinnarctidion, Desmatophoca, Allodesmus, and Imagotaria and is primitive for pinnipedimorphs. 5. Auditory bulla underlaps basioccipital. According to Barnes and Hirota, this character was incorrectly described and attributed by Berta and Wyss (1994). We agree. 6. Bony flange below ascending ramus. Barnes and Hirota (1995:356) claimed that this is an incorrectly described and at­ tributed character. This character, they argued, is "related to the musculature and functional morphology of the dentary and not of phylogenetic significance." They offered no evidence to support this assertion, however. Clearly, this flange (the mar­ ginal process of Davis, 1964) is correlated with the jaw abduc­ tion musculature, but this does not negate any phylogenetic sig­ nificance. In fact, our proposed phylogeny (Figure 15) indicates that the derived condition (character 22) evolved only once in the common ancestor of phocoids. 7. Pericardial plexus well developed. As Barnes and Hirota (1995:356) wrote, "soft anatomical structures that are not re­ flected by bone morphology cannot elucidate the relationships of fossil taxa." This is, of course, true. We wish to point out, however, that this character was identified by parsimony analy­ sis (Berta and Wyss, 1994) as a potential synapomorphy of this clade. Because it was scored as "?" in fossil taxa, it should be treated as apomorphic for phocids, the minimum level at which observation confirms its distribution. In summary, we agree with Barnes and Hirota that the Phoc­ oidea as originally defined (i.e., Pinnarctidion, Allodesmus, Desmatophoca, and phocids) is not supported by our revised character evidence. Our analysis, however, does recognize a monophyletic Phocoidea, but one that we redefine as the clade containing Desmatophoca, Allodesmus, and phocids. Desmatophocidae (Allodesmus^Desmatophoca): This monophyletic group is diagnosed by four unequivocal synapo­ morphies (Figure 15): stylomastoid foramen separated by raised strut from tympanohyal (14), tympanic crest present and dorsally convergent with petrosal (16), 13 procumbent and lat- 140 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY erally directed (23), and Pl—3 with well-developed cingula with numerous cuspules/crenations (26). One equivocal syna­ pomorphy is potentially diagnostic of this clade: laterally ex­ panded pterygoid process of the maxilla (3, also independently derived in basal odobenids and species of Pinnarctidion). Recognition of this clade differs from Barnes's (1979) ar­ rangement in which Allodesmus was removed from the Des­ matophocidae because of the presence of three "derived" char­ acters shared between Pinnarctidion and Allodesmus: (1) extreme lateral excavation of the palatine-pterygoid strut be­ tween the palate and the braincase; (2) more dorsally flattened posterior narial opening; and (3) more posterior and ventral po­ sition of the orbital aperture of the optic foramen. The posses­ sion of the "primitive" states of these characters in species of Desmatophoca was sufficient to convince Barnes (1979) that Allodesmus evolved directly from the "enaliarctine" Pinnarc­ tidion without "passing through" a Desmatophoca ancestor. Our evaluation of these characters suggests that they are more variable than observed by Barnes (1979) and not seen in Al­ lodesmus and Pinnarctidion exclusively. First, several of the Emlong specimens of Desmatophoca oregonensis (USNM 335451 and 335478) have a pterygoid strut that is shelf-like and roughly as wide as in Pinnarctidion bishopi, whereas an­ other specimen (USNM 314645) has a narrow, rounded ptery­ goid strut. Second, although Barnes claims that the internal narial opening is more compressed in Pinnarctidion bishopi and Allodesmus kernensis, this character also is variable in Desmatophoca oregonensis and in some cases is probably re­ lated to diagenesis. Third, we observed variability in the posi­ tion of the optic foramen. In the cast of the holotype of D. ore­ gonensis the foramen is more anterior; however, in USNM 335702, the foramen is posteriorly placed. In a recent phylogenetic study of Allodesmus from Japan, Kohno (1996) supported recognition of the Desmatophocidae as a monophyletic clade including Desmatophoca and Allodes­ mus. Although he did not discuss specific character evidence for this arrangement, the data in his character-taxon matrix (his table 3) do provide support. Desmatophoca: Diagnosis of this monophyletic taxon is based upon two synapomorphies (Figure 15): mortised contact with the zygomatic process expanded posteriorly (11) and p2-4 crowns with anterior and posterior cuspules conical (27). A posteriori character analysis recognizes another possible syna­ pomorphy: Pl—3 crowns with well-developed cingula and nu­ merous cuspules (26). The derived state evolved first in Des­ matophoca; absence of cingula or cuspules represents a secondary transformation (derived state 2) in Allodesmus. If correct, this transformation series predicts that well-developed cingula and cuspules should be found in D. brachycephala. Desmatophoca brachycephala is diagnosed on the basis of the following autapomorphies (Figure 15): widely flared zygo­ matic arch (15) and squamosal fossa divided, with anterior and posterior portions of equal size (18). Desmatophoca oregonen­ sis is distinguished from Desmatophoca brachycephala by two unequivocal autapomorphies (Figure 15): incisive foramen without medial septum (7) and external auditory meatus axis oriented less than or equal to 62° (19). Allodesmus: Diagnosis of a monophyletic Allodesmus is based upon six synapomorphies (Figure 15): dorsal margin of zygomatic arch retracted dorsal to infraorbital foramen (6), su­ praorbital processes of frontals developed as a posteriorly posi­ tioned swelling (8), styloid process of auditory bulla absent (13), upper canine crowns lacking posterior and medial carinae (24), upper postcanine teeth single rooted (25), and upper post­ canine tooth roots inclined (28). Although recent work by Barnes and Hirota (1995) and Kohno (1996) recognized as many as nine species of al- lodesmines from the Miocene of California and Japan, a review of their evidence suggests that the taxa are probably oversplit. This is especially true for species of Allodesmus from Califor­ nia, three of which alone are reported from a very limited stratigraphic interval within the middle Miocene, Round Mountain Silt. Because of this oversplitting and the fact that resolution of phylogenetic relationships within this clade was not a primary goal of our study, we decided to include only four species of Allodesmus in the present analysis. These four spe­ cies (Allodesmus kernensis, A. packardi, A. courseni, and A. sa- doensis) were selected in part for their long taxonomic standing and/or representation by fairly complete fossil material. The re­ sults reveal that A. kernensis and A. sadoensis are sister taxa, distinguished from A. packardi by four unequivocal synapo­ morphies (Figure 15): palate transversely arched (4), lateral wall of alisphenoid canal thin (5), infraorbital canal small (17), and moderately sized orbit (20). Allodesmus courseni was omitted from the computer analysis because it could only be scored for eight of 30 characters (see Table 4). This taxon does possess certain synapomorphies, however, so it could be manu­ ally placed in an unresolved position within the Allodesmus clade. A basal position within this clade is suggested by pos­ session of the primitive condition for one character: multiple rooted upper postcanine teeth (25). For a more complete analy­ sis of phylogenetic relationships among species of Allodesmus see Kohno (1996). Conclusions Analysis of new fossil skulls, mandibles, and postcrania of Desmatophoca oregonensis from the Astoria Formation of the Newport Embayment of coastal Oregon, USA, documents the range of morphologic variation in this taxon and concludes that several characters previously considered diagnostic of different species are actually the result of sexual dimorphism and/or de­ velopmental age. As a basal desmatophocid, Desmatophoca oregonensis has an elongate skull and upper postcanine teeth characterized by bulbous crowns with prominent central cusps, small posterior cuspules, and distinct crenulated lingual cin­ gula. The lower postcanine teeth have distinctly conical ante- NUMBER 93 141 TABLE 4.—Character-taxon matrix showing distribution of cranial, dental, and mandibular character states among selected fossil and modern pinnipeds. Character state 0 represents the ancestral state; states 1-3 represent the de­ rived states. Missing data are scored as ? Taxon Character 1 2 3 4 5 6 7 8 9 10 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 Enaliarchos spp. Pteronarchos spp. Otariidae Odobenidae Phocidae Pinnarctidion bishopi P. rayi Desmatophoca oregonensis D. brachycephala Allodessmus kernensis A. courseni A. packardi A. sadoensis 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 2 0 0 0 0 0 3 0 0 0 0 0 0 0 0 0 0 ? 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 1 0 0 1 0 0 0 0 0 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 2 0 0 1 3 2 2 1 2 0 0 0 0 0/2 0 1 1 0 0 / 1 0 0 0/2 0 I 0 0 0 0 0 0 0 0 0 ? 0 0 0 1 0 0 1 1 I 0 0 1 1 0 0 0 0 0 0 0 ? 1 0 0 1 0 0 2 0 1 0 0 0 ? 0 0 1 0 ? 0 1 1 ? ? ? 0 ? 1 0 0 ? ? 0 ? 1 0 0 0 0 ? 0 1 1 7 7 7 0 0 ? 0 ? 0 0 0 0 0 0 1 0 0 1 0 0 1 1 0 0 0 0 0 1 0 ? 2 2 1 1 1 1 I rior and posterior cuspules, finely crenulated lingual cingula, and broadly rounded principal cusps. Phylogenetic analysis confirms the monophyly of the des- matophocids, which represent a group of early and middle Mi­ ocene North Pacific endemic pinnipeds with enlarged paroccip­ ital processes and bulbous-crowned postcanine teeth. The family includes two genera, Desmatophoca and Allodesmus, which together contain at least six described species from the early and middle Miocene. The oldest record of Desmatopho­ cidae is currently established by Desmatophoca brachycephala Barnes, 1987, from the early Miocene (Aquitanian correlative) Astoria Formation of Washington, USA, suggesting that diver­ gence of desmatophocids from phocids likely occurred some­ time before 18 Ma, probably in the North Pacific Ocean basin. Appendix Characters and Character-states for Desmatophocid Pinnipeds and Related Taxa (0=primitive; I-3=derived; ?=missing) 1. Premaxilla-nasal contact. 0=extensive, 1 =reduced (Wyss, 1987:7, 15, fig. 5). In Enaliarctos, Pteronarctos, and Pinnarctidion, the premax­ illa has a broad area of contact with the nasals on the rostrum (primitive condition). A short, narrow contact between the pre­ maxilla and nasal occurs in Desmatophoca, Allodesmus, and the Phocidae (derived condition) and represents a phocoid syn­ apomorphy. 2. Nasal-frontal suture. 0=transverse, l=V-shaped, 2=W- shaped (Demere, 1994:105, fig. 3). As discussed by Demere (1994) the posterior border of the nasal is blunt and nearly transverse in Enaliarctos, Pteronarc­ tos, and Pinnarctidion. Although two derived states are recog­ nized, the homology of the first, a V-shaped naso-frontal suture (point of V directed posteriorly between frontals) is question­ able. Two variations of this condition are seen: In the dusig- nathine odobenids (e.g., Gomphotaria and Dusignathus), there is a broad V-shaped suture. In Desmatophoca, Allodesmus, and phocids there is an acute V-shaped naso-frontal suture. Otariids display a W-shaped naso-frontal suture, in which the frontals extend anteriorly between the nasals at the midline (King, 1983:151, fig. 6.4), identified as derived state 2. A posteriori character analysis suggests the evolution of character state 1 in phocoids+Otariidae with transformation to state 2 occurring in the Otariidae (ACCTRAN optimization). Another possibility is the independent evolution of state 1 in the Phocoidea clade with the independent evolution from state 0 to state 2 in the Otariidae (DELTRAN optimization). 3. Laterally expanded pterygoid process of the maxilla. 0= absent, 1 =present (Barnes, 1979:23). The pterygoid process of the maxilla is defined by Barnes (1979:23) as a "wide, thin, squared posterolaterally projecting shelf of the palate beneath each orbit." This structure in Des­ matophoca brachycephala is described as an expansive ptery­ goid process of the maxilla that forms a thin infraorbital shelf with a prominent posterolateral corner (Barnes, 1987). Lateral expansion of the pterygoid process also is seen among basal odobenids (e.g., Prototaria; Kohno, 1994). Berta (1991) de­ fined this character differently. She identified the presence of a "palatine shelf in Pinnarctidion, Desmatophoca, and Allodes­ mus but recognized it as a further modification (derived state 2) of derived condition 1, in which the palatine process of the maxilla extends posterior to the last molar. Without further in- 142 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY formation about the homology of these character states, it seems best to represent them as separate characters. A posteriori character analysis suggests that the derived con­ dition, a laterally expanded pterygoid process of the maxilla evolved at least three times: in basal odobenids, Pinnarctidion, and desmatophocids (Allodesmus and Desmatophoca). The primitive condition seen in all other pinnipeds is the lack of a laterally expanded pterygoid process of the maxilla. 4. Palate. 0=relatively flat (palatal arch means of 0.18-0.19), 1 = transversely arched (palatal arch means greater than 0.25) (Demere, 1994:106; this study). The palate of pinnipeds varies from relatively transversely flattened to transversely arched. Palatal arch can be quantified as the ratio of the depth of the palate at the midline to the chord length of the transverse arch measured between the lingual al­ veolar borders of individual right and left tooth positions. Tak­ ing ratios for each tooth position (i.e., C, Pl-4 , Ml,2) and com­ puting the mean ratio provides a means for comparison. Palatal arch ratios of 0.15-0.26 (x=0.19) occur in Desmatophoca ore­ gonensis, 0.14-0.21 in Desmatophoca brachycephala (x— 0.18), and 0.21-0.30 in Allodesmus kernensis (x=0.26). Palatal arch in A. sadoensis measures 0.30 (based upon a single mea­ surement taken at P4). Given these ratios, we define the de­ rived condition, a transversely arched palate as having a palatal arch mean greater than 0.25. The derived condition occurs in Allodesmus kernensis and A. sadoensis. All other pinnipeds, with the exception of the later-diverging odobenine walruses and the divergent otariid Otaria, are characterized by relatively flat palates (Demere, 1994) with palatal arch means that range between 0.18 and 0.19. 5. Lateral wall of alisphenoid canal. 0=thick and well de­ fined, l=thin, 2=absent (Barnes and Hirota, 1995; this study). The lateral wall of the alisphenoid canal is thick and well de­ fined in species of Enaliarctos, Pteronarctos, Pinnarctidion, Desmatophoca, odobenids, and otariids. In the derived condi­ tion seen in Allodesmus kernensis and A. sadoensis, the lateral wall is extremely thin when preserved, and it is often lost through postmortem abrasion and/or diagenesis. In those speci­ mens with non-preservation, there is no obvious trace of the wall. Allodesmus packardi retains the primitive condition (Barnes and Hirota, 1995). In phocids, the alisphenoid canal is always absent, identified as derived condition 2. 6. Dorsal margin of maxillary root of zygomatic arch. 0=not retracted dorsal to infraorbital foramen, l=retracted dorsal to infraorbital foramen (Barnes, 1989). Barnes (1989) observed that species of Allodesmus could be distinguished from those of Desmatophoca by having the dor­ sal margin of the maxillary root of the zygomatic arch posteri­ orly retracted dorsal to the infraorbital foramen. Associated with the receding dorsal margin of the zygomatic arch in Al­ lodesmus are very large orbits. In Desmatophoca and all other pinnipeds, the dorsal margin of the zygomatic arch flares ante­ riorly above the anterior opening of the infraorbital canal and forms a "cup" for the eyeball (Barnes, 1987). The derived con­ dition is interpreted as a synapomorphy for species of Allodes­ mus. 7. Incisive foramina. 0=divided by medial septum, l=medial septum absent, 2=foramina reduced (this study). The incisive foramina in Enaliarctos, Pinnarctidion, Pteron­ arctos, basal odobenids, and otariids are divided by a distinct medial bony septum. This configuration, representing the prim­ itive condition, also occurs in Desmatophoca brachycephala. In D. oregonensis, there is no medial septum developed, and the right and left foramina are joined into a single incisive opening, derived state 1. This condition also occurs in the later- diverging odobenid Dusignathus seftoni. In Allodesmus ker­ nensis and A. sadoensis, the incisive foramina are very small and positioned nearly on the intermaxillary suture; derived condition 2. DELTRAN optimization indicates that the transformation from state 0 to state 2 is a synapomorphy for Allodesmus ker­ nensis and A. sadoensis. ACCTRAN optimization holds that this transformation occurred at the level of Allodesmus and pre­ dicts its occurrence in A. packardi. 8. Supraorbital process of frontals. 0-small, distinct ridge, 1 ^extreme reduction or complete absence, 2=posteriorly posi­ tioned swelling, 3=strongly developed (King, 1983, fig. 6.7; Berta and Wyss, 1994; this study). A small but distinct supraorbital ridge is seen in species of Enaliarctos, Pteronarctos, Desmatophoca, Pinnarctidion, and basal odobenids (e.g., Prototaria). Later-diverging odobenids and phocids are characterized by extreme reduction or com­ plete absence of the supraorbital process, derived state 1. Al­ lodesmus is characterized by a supraorbital process that is de­ veloped as a posterior-placed swelling, derived state 2. A strongly developed supraorbital process, derived state 3, occurs only in otariids. A posteriori character analysis suggests that each of the three derived character states evolved independently from the primi­ tive condition. 9. Paroccipital process. 0=small, blunt, l=enlarged and ex­ cavated, 2 = enlarged and not excavated, 3 = unenlarged and crest-like (this study, Figure 7). The shape, size, orientation, and relationship of the paroccip­ ital process to surrounding bones have been variously used to distinguish among pinnipeds (e.g., Mitchell, 1968; Repenning and Tedford, 1977). In the primitive condition seen in Enaliar­ ctos, the paroccipital process is small and blunt. Three derived states are recognized. Pteronarctos, Pinnarctidion, basal odo­ benids, and otariids are characterized by having an enlarged pa­ roccipital process with an excavated ventral surface (derived state 1). Although the paroccipital process in Desmatophoca and Allodesmus is enlarged, the ventral surface is not excavated (derived state 2). The paroccipital process in phocids is unen- larged and crest-like (Muizon and Hendey, 1980:110) (derived state 3), a condition that is not homologous with other pinni­ peds. NUMBER 93 143 DELTRAN optimization indicates independent transforma­ tions from state 1 to state 0 in Enaliarctos, from state 1 to 2 in the Desmatophocidae, and from state 1 to state 3 in the Pho­ cidae. ACCTRAN optimization indicates transformations from state 1 to state 2 in the Phocoidea and from state 2 to state 3 in the Phocidae. 10. Auditory bulla. 0=ectotympanic inflated, 1= ectotympanic flattened, 2=entotympanic inflated (King, 1983, fig. 6.3). In Enaliarctos, Pteronarctos, and Pinnarctidion, the ecto­ tympanic portion of the bulla is slightly inflated (primitive con­ dition). In Allodesmus, Desmatophoca, the Otariidae, and the Odobenidae, the ectotympanic is flattened, derived condition 1. Only phocids have a greatly inflated entotympanic, derived condition 2 (Wyss, 1987). 11. Squamosal-jugal articulation. 0=splint-like, l=mortised zygomatic process expanded posteriorly, 2=mortised zygo­ matic process expanded anteriorly (this study, Figure 9). Barnes (1979) described the splint-like arrangement of the squamosal and jugal in which the zygomatic portion of the squamosal tapers to a sharp point anteriorly but does not touch the postorbital process of the jugal. This condition, seen in Enaliarctos and otariids, is identified as the primitive condi­ tion. In Allodesmus, Desmatophoca, and phocids, the squamo­ sal jugal contact is mortised, an arrangement in which both the postorbital process of the jugal and the zygomatic process of the squamosal are expanded dorsally to form a broad contact. The condition in Pinnarctidion, in which the zygomatic pro­ cess of the squamosal is only slightly expanded and fits into a shallow notch in the postorbital process of the jugal, has been interpreted as an intermediate stage of mortising (Barnes, 1987; Berta, 1994; Berta and Wyss, 1994). We do not support this earlier conclusion, however, because there is no mutual expan­ sion of both the jugal and squamosal. Among those pinnipeds that display a mortised contact of the squamosal and jugal, two derived states are recognized. In Des­ matophoca, the greatest dorsal expansion of the zygomatic pro­ cess is positioned posteriorly (derived state 1), whereas in both Allodesmus and phocids the greatest dorsal expansion of the zygomatic process of the squamosal is positioned anteriorly (derived state 2). A posteriori character analysis suggests that anterior place­ ment of the dorsal expansion evolved in the common ancestor of phocids and desmatophocids, with evolution of the posterior position representing a secondary transformation in Desmato­ phoca. 12. Jugal-maxillary suture. 0=jugal with anterodorsal and an­ teroventral splints, l=jugal with anterodorsal splint only (Kohno, 1996, listed but not described). The jugal contacts the maxillary in a deeply interdigitating foliate suture with anterodorsal and anteroventral splints. This is the primitive condition, which occurs in all pinnipeds except phocids. In phocids, the suture is a simple splint with the an­ terodorsal splint overriding the posteroventral splint of the maxillary, the derived condition. There is no anteroventral splint of the jugal. 13. Styloid process of auditory bulla. 0=present, l=absent, 2 = ectotympanic reduced (this study). The primitive condition of a distinct styloid process of the ectotympanic ventral to the Eustachian tube occurs in all out­ group taxa and Desmatophoca. In Allodesmus kernensis, the styloid process is absent (derived condition 1) and instead the anterior corner of the ectotympanic is retracted posteriorly in the region of the eustachian tube (Figure 6B). In phocids, the entotympanic is greatly enlarged, resulting in reduction of the ectotympanic and reconfiguration of the exit of the eustachian tube (derived state 2). 14. Stylomastoid foramen. 0=in a common fossa with tympa- nohyal pit, l=separated by raised strut from tympanohyal (Berta, 1994, listed but not described; this study). In Desmatophoca and Allodesmus, the stylomastoid foramen is separated from the tympanohyal by a raised bony strut, the derived condition. In all other pinnipeds, the stylomastoid fora­ men and the tympanohyal lie very close to one another in a common fossa. The derived condition of this character was in­ correctly reported in Pinnarctidion (Berta, 1994). 15. Zygomatic arch. 0=not widely flared (less than 62% of condylobasal length), 1 = widely flared (greater than or equal to 62% of condylobasal length) (this study). Although widely flared zygomatic arches have been used to distinguish Desmatophoca brachycephala from all other pinni­ peds (Barnes, 1987), this character has not previously been quantified. We redefine this character as zygomatic width rela­ tive to condylobasal length (CBL). Zygomatic width measures 65% of CBL in Desmatophoca brachycephala, 58%-60% in Desmatophoca oregonensis (x=59%, n=3), 56% in Enaliarctos emlongi, 54%-57% in Allodesmus kernensis (x=55%, «=3), 61% in Homiphoca capensis (measured from a photograph in Muizon and Hendey, 1980, fig. 3), and 57% in Piscophoca pacifica (measured from a photograph in Muizon, 1981, pl. 1: fig. 2). Given this range of measurements among fossil pinni­ peds, we defined the derived condition as zygomatic arches measuring greater than or equal to 62% of CBL. By this defini­ tion only Desmatophoca brachycephala possesses the derived condition. 16. Tympanic crest. 0=present and horizontally continuous, 1 =present and dorsally convergent with petrosal (Mitchell and Tedford, 1973:228). The tympanic crest (crista tympani) is described as occurring along the lateral wall of the auditory bulla in Enaliarctos mealsi (Mitchell and Tedford, 1973). This is the primitive con­ dition seen in outgroup taxa and Pinnarctidion bishopi (Barnes, 1979). In Desmatophoca oregonensis and Allodesmus kernen­ sis, a narrow crest is evident on the lateral wall of the bulla, be­ ginning at the same position as the tympanic crest (septum be­ tween the tympanic opening and the epitympanic recess) and extending anterodorsally toward the promontorium. Although there is some question about the homology of these crests, we 144 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY have coded the condition in Desmatophoca and Allodesmus as derived. DELTRAN optimization indicates independent character- state transformations for Desmatophoca oregonensis and for Allodesmus kernensis. ACCTRAN optimization indicates that the derived condition is a synapomorphy for the Desmatopho­ cidae. 17. Infraorbital foramen. 0=large, l=small (this study). The primitive condition, infraorbital foramen large relative to condylobasal length, occurs in Enaliarctos, Pinnarctidion, Desmatophoca, and Allodesmus packardi. The derived condi­ tion, a small infraorbital foramen, occurs in Allodesmus kern­ ensis and Allodesmus sadoensis. 18. Squamosal fossa. 0=undivided, l=divided, posterior and anterior portions of equal size, 2=divided, with posterior por­ tion smaller than anterior portion (Barnes, 1987; this study). As defined by Barnes (1987:11), the squamosal fossa is "the broad recess in the squamosal between the braincase and the zygomatic arch, dorsal to the glenoid fossa, and which floors the temporal fossa." Barnes further noted that the squamosal fossa is relatively wider in Desmatophoca brachycephala than in D. oregonensis and other pinnipeds. We redefine this character as the recess between the anterior border of the preglenoid process of the squamosal and the lambdoidal crest portion of the mastoid process. In addition, we further divide the fossa into anterior and posterior portions. The anterior portion is dorsal to the glenoid fossa and is roughly equivalent to Barnes's characterization of the squamo­ sal fossa. The posterior portion of the squamosal fossa lies above the mastoid process and the external auditory meatus. In the primitive condition seen in Enaliarctos, Pteronarctos, Pin­ narctidion, odobenids, and otariids, the squamosal fossa is transversely narrow and continuous (i.e., anterior and posterior portions undivided). In Desmatophoca and Allodesmus, the two portions are divided by a transverse raised bony strut that extends from the lateral wall of the braincase to the middle of the squamosal root of the zygomatic arch. In D. brachycephala, the posterior portion of the fossa is roughly equivalent in size to the anterior portion (derived condition 1). In D. oregonensis and Allodesmus kernensis, the posterior portion of the fossa is relatively smaller than the anterior portion (derived condition 2). Basal phocids possess both the primitive condition (e.g., Monachus schauinslandi, SDSNH 18711) and the derived state (e.g., Piscophocapacifica; Muizon, 1981, pl. 1: fig. 3) (derived condition 2) of this character. 19. External auditory meatus (EAM). 0=axis oriented greater than 62°, l=axis oriented less than or equal to 62° (Barnes, 1987; this study). We quantified this character by measuring the angle that the axis of the EAM makes with the sagittal plane of the skull. This angle measures 75° in the holotype of D, brachycephala and ranges between 56° and 61° in four skulls of D. oregonensis (x =57.8°). Polarization of this character is difficult to assess be­ cause of the shortness of the EAM in species of Enaliarctos (the angle measures approximately 72° in Enaliarctos mitchelli and 70° in Enaliarctos mealsi). This angle measures 73° in Pteronarctos, about 67° in Pinnarctidion bishopi, 64° in Al­ lodesmus kernensis, 70° in Allodesmus packardi, 78°-84° in otariids (84° in Thalassoleon mexicanus, measured from a pho­ tograph in Repenning and Tedford (1977, pl. 20)), and at least 77°-85° in phocids (77° in Mirounga angustirostris (Gill), and 85° in Homiphoca capensis, measured from a photograph in Muizon and Hendey (1980, fig. 2). Using these measurements, we define the primitive condition as an EAM axis oriented greater than 62° to the sagittal plane and the derived condition as an EAM axis oriented less than or equal to 62°. Following this definition, the derived condition occurs only in Desmatophoca oregonensis. 20. Orbit size. 0=small (less than or equal to the temporal fossa length, 1= intermediate (110% to 130% of temporal fossa length), 2=large (greater than 130% of temporal fossa length) (this study). We define the orbit as the region from the antorbital rim of the maxilla to the postorbital process of the jugal. We define the temporal fossa as the opening between the anterior border of the preglenoid process of the squamosal and the postorbital process of the jugal. In the primitive condition, the orbit is small, measuring 98% of the temporal fossa length in Enaliarc­ tos emlongi. In derived condit ion 1, the orbit size is 110%-130% of the temporal length (approximately 125% of the orbit length in D. oregonensis and 119% in D. brachyceph­ ala). In derived condition 2, seen in species of Allodesmus and phocids, the orbit size is greater than 130% of the temporal fossa length (139% in A. kernensis, 161% in A. packardi, 193% in Piscophoca pacifica (measured from a photograph in Muizon, 1981, pl. 1), and 261% in Homiphoca capensis (mea­ sured from a photograph in Hendey and Repenning, 1972, fig. 5)). Correlated with relative size of the orbit and temporal fossa is the relative position of the supraorbital process along the in- terorbital/temporal bar. We have chosen not to recognize this as a separate character because it is directly related to orbit size. Kohno (1996) originally defined this character and distin­ guished two character states. In the primitive condition, the su­ praorbital process is located relatively anterior to the orbital rim, whereas in the derived condition the supraorbital process is positioned far posterior on the postorbital bar. The position of the supraorbital process is best described relative to the mid­ point of the postorbital bar. In all pinnipeds except species of Allodesmus, the supraorbital process is positioned anterior to the midpoint of the postorbital bar. Allodesmus possesses the derived condition in having the supraorbital process positioned posterior to the midpoint of the postorbital bar. A posteriori character analysis suggests that derived state 2 evolved independently in phocids and species of Allodesmus. 21. Angular (=pterygoid) process. 0=well developed, ^ r e ­ duced or absent (Berta and Wyss, 1994). NUMBER 93 145 The angular process is the area for insertion of the internal pterygoid muscle. A well-developed angular process posi­ tioned near the base of the mandibular condyle characterizes terrestrial carnivorans, Enaliarctos, Pteronarctos, otariids, odo­ benids, and Desmatophoca oregonensis. The derived condition of a reduced or absent angular process occurs in monachine phocids and Allodesmus (Berta and Wyss, 1994). DELTRAN optimization indicates independent transforma­ tions of this character state in Allodesmus kernensis and the Phocidae. ACCTRAN optimization indicates that this character transformation evolved in the Phocoidea and that a reversal oc­ curred in Desmatophoca. 22. Flange (=marginal process) below ascending ramus. 0 = small or absent, l=well developed (Berta, 1991; Demere, 1994). The marginal process (Davis, 1964) is the area for insertion of the digastricus muscle on the medioventral margin of the mandible. The process is small in species of Enaliarctos (Berta, 1991), Pteronarctos (Berta, 1994), and basal odobenids (Demere, 1994). A well-developed flange-like marginal pro­ cess occurs in Desmatophoca and Allodesmus. 23. 13. 0=not procumbent, l=procumbent and laterally di­ rected (this study). The 13 in Enaliarctos, Pinnarctidion, and Pteronarctos is vertically directed. In Desmatophoca, 13 is procumbent and lat­ erally directed. This tooth is not present among published spec­ imens of Allodesmus, although Barnes (1972:14) mentioned the procumbent roots of this tooth. In a complete skull of Allodes­ mus kernensis (BVM 0163) with preserved right and left 13, the roots are strongly procumbent and the crowns are laterally di­ rected. The derived condition is a synapomorphy linking Des­ matophoca and Allodesmus. 24. Upper canine, crowns with posterior and medial carinae. 0 =present, l=absent (Barnes, 1987). The posterior and medial margins of the upper canine in Enaliarctos, Pteronarctos, Pinnarctidion, and Desmatophoca are characterized by distinct longitudinal carinae, the primitive condition. Carinae are absent in Allodesmus (Barnes, 1987) and later-diverging odobenids (i.e., Alachtherium, Valenictis, Odo­ benus), the derived condition. Basal phocids vary in their pos­ session (e.g., Monachus schauinslandi; pers. obs.) or lack of carinae (e.g., Piscophoca pacifica; Muizon, 1981:22). DELTRAN optimization indicates that the derived condition is a synapomorphy linking Allodesmus kernensis and Allodes­ mus sadoensis. ACCTRAN optimization indicates that the de­ rived condition is a synapomorphy uniting species of Allodes­ mus. 25. Upper postcanine teeth. 0=triple/double rooted, l=single rooted (Barnes, 1972). The derived condition, single-rooted upper postcanine teeth (except for Pl , which is always single rooted) occurs in Al­ lodesmus kernensis, A. packardi (Barnes, 1972), and A. sado­ ensis. All other pinnipeds (except later-diverging odobenids and otariids) retain the primitive condition of having multiple- rooted cheek teeth. Although the upper postcanine teeth are not present (or are hidden in matrix) in the holotype of A. courseni, the left p4 and right p3 are clearly double rooted, suggesting that the upper teeth also were double rooted (pers. obs.). 26. Pl—3 crowns. 0=narrow lingual cingula without numerous cuspules/crenations, 1 =well-developed cingula with numerous cuspules/crenations, 2=no cingula, no cuspules/crenations (this study). In Enaliarctos emlongi and E. mitchelli, Pl—3 have distinct but narrow lingual cingula without any cingular cuspules, the primitive condition. In Desmatophoca oregonensis, these teeth have well-defined cingula surrounding the lingual border of the principal cusp, and a series of small cuspules/crenations char­ acterizes the edge of these cingula (derived condition 1). In Al­ lodesmus kernensis, there is no cingulum nor cuspules/crena­ tions on P1-3 (derived condition 2). Instead, the lingual profile of the crown is smoothly concave from tip to base. Barnes (1972) and other workers have described the postcanine teeth of this taxon as bulbous, referring to the swollen base of the tooth crown. The term is misleading, however, and the more significant feature is the lack of a cingulum. DELTRAN optimization indicates independent character transformations from state 0 to state 1 in Desmatophoca ore­ gonensis and from state 0 to state 2 in Allodesmus kernensis + A. sadoensis. ACCTRAN optimization indicates transforma­ tion from state 0 to state 1 as a synapomorphy for the Desmato­ phocidae clade, with transformation from state 1 to state 2 as a synapomorphy supporting the monophyly of the genus Al­ lodesmus. 27. p2-4 crowns. 0=anterior and posterior cuspules trenchant, l=cuspules conical, 2=cuspules absent (this study). In Enaliarctos emlongi, the crowns of p2-4 are characterized by a principal cusp (protoconid?) flanked by an anterior (para­ conid?) and posterior (metaconid?) trenchant cusp/cuspule. This represents the primitive condition. In Desmatophoca ore­ gonensis, the anterior and posterior cusps/cuspules are conical (derived condition 1). In Allodesmus kernensis, the tooth crowns are smooth and lack cusps/cuspules (derived condition 2). Derived condition 2 also applies to later-diverging members of other pinniped groups except phocids. DELTRAN optimization indicates independent transforma­ tions from state 0 to state 1 in Desmatophoca oregonensis and from state 0 to state 2 in Allodesmus kernensis. ACCTRAN op­ timization indicates transformation from state 0 to state 2 as a synapomorphy uniting the Phocoidea clade with a second transformation from state 2 to state 1 occurring in the Desmato­ phocidae. 28. Upper postcanine tooth roots. 0=vertical, l=inclined (Barnes, 1989; this study). In all outgroup taxa, as well as in Desmatophoca and phoc­ ids, the roots of the upper postcanine teeth are oriented verti­ cally in the maxillary, the primitive condition. In the derived condition seen in Allodesmus kernensis, A. packardi, and A. sa­ doensis, the postcanine roots are ventrolateral^ inclined. 146 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY 29. P4 metacone. 0 = distinct, l=reduced or absent (Berta, 1994). In Enaliarctos, Pteronarctos, Pinnarctidion, and basal odo­ benids, P4 has a distinct metacone, representing the primitive condition. In Allodesmus, Desmatophoca, otariids and phocids, as well as later-diverging odobenids, the metacone is reduced or absent on the upper carnassial (Berta, 1994), the derived condition. 30. P4, protocone shelf. 0=present l=absent (Barnes, 1979). The presence of a protocone shelf on the upper carnassial has been used to distinguish Enaliarctos and Pteronarctos from all other pinnipedimorphs (Barnes, 1979, 1989). The shelf-like protocone is an ursid+pinnipedimorph synapomorphy (Flynn et al., 1988; Berta et al., 1989) seen in Enaliarctos, Pteronarc­ tos, Pinnarctidion, and basal odobenids. The derived condition, absence of the P4 protocone shelf, occurs in otariids, Desmato­ phoca, Allodesmus, later-diverging odobenids, and phocids. Literature Cited Addicott, W.A. 1976. Neogene Molluscan Stages of Oregon and Washington. In A.E. Fritsche, H. Ter Best, Jr., and W.W. Wornardt, editors, The Neogene Symposium. Society of Economic Paleontologists and Mineralo­ gists, Annual Meeting, Pacific Section, pages 95-116. Armentrout, J.M. 1981. Correlation and Ages of Cenozoic Stratigraphic Units in Oregon and Washington. Geological Society of America Special Paper, 184: 137-148. Barnes, L.G. 1972. Miocene Desmatophocinae (Mammalia: Carnivora) from Califor­ nia. University of California Publications in Geological Sciences, 89: 68 pages. 1979. Fossil Enaliarctine Pinnipeds (Mammalia: Otariidae) from Pyramid Hill, Kern County, California. Contributions in Science, Natural History Museum of Los Angeles County, 318: 41 pages. 1987. An Early Miocene Pinniped of the Genus Desmatophoca (Mamma­ lia: Otariidae) from Washington. Contributions in Science, Natural History Museum of Los Angeles County, 382: 20 pages. 1989. A New Enaliarctine Pinniped from the Astoria Formation, Oregon, and a Classification of the Otariidae (Mammalia: Carnivora). Con­ tributions in Science, Natural History Museum of Los Angeles County, 403: 26 pages. Bames, L.G., and K. Hirota 1995. Miocene Pinnipeds of the Otariid Subfamily Allodesminae in the North Pacific Ocean: Systematics and Relationships. The Island Arc, 3:329-360. Berta, A. 1991. New Enaliarctos (Pinnipedimorpha) from the Oligocene and Mi­ ocene of Oregon and the Role of "Enaliarctids" in Pinniped Phylog­ eny. Smithsonian Contributions to Paleobiology, 69:1-33. 1994. A New Species of Phocoid Pinniped Pinnarctidion from the Early Miocene of Oregon. Journal of Vertebrate Paleontology, 14:405-413. Berta, A., C E . Ray, and A.R. Wyss 1989. Skeleton of the Oldest Known Pinniped, Enaliarctos mealsi. Sci­ ence. 244:60-62. Berta, A., and A.R. Wyss 1994. Pinniped Phylogeny. In A. Berta and T.A. Demere, editors, Contri­ butions in Marine Mammal Paleontology Honoring Frank C. Whit­ more, Jr. Proceedings of the San Diego Society of Natural History, 29:33-56. Condon, T. 1906. A New Fossil Pinniped (Desmatophoca oregonensis) from the Mi­ ocene of the Oregon Coast. University of Oregon Bulletin, supple­ ment, 3(3): 14 pages. Davis, D.D. 1964. The Giant Panda: A Morphological Study of Evolutionary Mecha­ nisms. Fieldiana: Zoology Memoirs, 3: 334 pages. Demere, T.A. 1994. The Family Odobenidae: A Phylogenetic Analysis of Fossil and Living Taxa. In A. Berta and T.A. Demere, editors, Contributions in Marine Mammal Paleontology Honoring Frank C. Whitmore, Jr. Proceedings of the San Diego Society of Natural History, 2:99- 123. Dooley, A.C., Jr. 1994. The First Well Preserved Squalodont (Cetacea) from the West Coast of North America. Journal of Vertebrate Paleontology, supplement, 14(3):23A. Doutt, J.K. 1942. A Review of the Genus Phoca. Annals of the Carnegie Museum, 29:61-125. Downs, T.D. 1956. A New Pinniped from the Miocene of Southern California. Journal of Paleontology, 30:115-131. Flynn, J.J., N.N. Neff, and R.H. Tedford 1988. Phylogeny of the Carnivora. In M.J. Benton, editor, 77ie Phylogeny and Classification of the Tetrapods, 2:73-116. Oxford: Clarendon Press. Hay, O.P. 1930. Second Bibliography and Catalogue of the Fossil Vertebrata of North America. Two volumes, 1074 pages. Washington, D.C: Car­ negie Institution of Washington. Hendey, Q.B., and C A . Repenning 1972. A Pliocene Phocid from South Africa. Annals of the South African Museum. 59:71-98. Howell, A.B. 1929. Contribution to the Comparative Anatomy of the Eared and Earless Seals (Genera Zalophus and Phoca). Proceedings of the United States National Museum, 73:1-142. 1930. Aquatic Mammals. 338 pages. Springfield, Illinois: Charles C Tho­ mas. Kellogg, R. 1922. Pinnipeds from the Miocene and Pleistocene of California. Univer­ sity of California Publications in Geological Sciences, 13:23-132. 1931. Pelagic Mammals from the Temblor Formation of the Kern River Region, California. Proceedings of the California Academy of Sci­ ences, series 4, 19:217-397. King, J. 1972. Observations on Phocid Skulls. In R.J. Harrison, editor, Functional Anatomy of Marine Mammals, 1:81-115. London: Academic Press. 1983. Seals of the World. Second edition, 240 pages. Ithaca, New York: Cornell University Press. Kohno, N. 1994. A New Miocene Pinniped in the Genus Prototaria (Carnivora: Odo­ benidae) from the Moniwa Formation Miyagi, Japan. Journal of Vertebrate Paleontology, 14:414-426. NUMBER 93 147 1996. Miocene Pinniped Allodesmus (Mammalia: Carnivora); with Spe­ cial Reference to the "Mito Seal" from Ibaraki Prefecture, Central Japan. Transactions and Proceedings of the Palaeontological Soci­ ety of Japan, new series, 181:388-404. Kohno, N., L.G. Barnes, and K. Hirota 1995. Miocene Fossil Pinnipeds of the Genera Prototaria and Neotherium (Carnivora; Otariidae; Imagotariinae) in the North Pacific Ocean: Evolution, Relationships, and Distribution. The Island Arc, 3: 285-308. Mitchell, E. 1961. A New Walrus from the Imperial Pliocene of Southern California: with Notes on Odobenid and Otariid Humeri. Contributions in Sci­ ence, Los Angeles County Museum, 44: 28 pages. 1966. The Miocene Pinniped Allodesmus. University of California Publi­ cations in Geological Sciences. 61:1-46. 1968. The Mio-Pliocene Pinniped lmagotaria. Journal of the Fisheries Research Board of Canada, 25:1843-1900. 1975. Parallelism and Convergence in the Evolution of Otariidae and Pho­ cidae. Rapports et Proces-verbaux des Reunions Conseil Interna­ tional pour I Exploration de la Mer, 169:12-26. Mitchell, E.D., and R.H. Tedford 1973. The Enaliarctinae, a New Group of Extinct Aquatic Carnivora and a Consideration of the Otariidae. Bulletin of the American Museum of Natural History, 151:201-284. Muizon, C. de 1981. Les vertebres fossiles de la formation Pisco (Perou); Recherche sur les grandes civilisations. Memoires de VInstitut Francais d'Etudes Andines, 6: 150 pages. Muizon, C. de, and Q.B. Hendey 1980. Late Tertiary Seals of the South Atlantic Ocean. Annals of the South African Museum, 82:91-128. Munthe, J., and M.C. Coombs 1979. Miocene Dome-Skulled Chalicotheres (Mammalia, Perissodactyla) from the Western United States: A Preliminary Discussion of a Bi­ zarre Structure. Journal of Paleontology, 53:77-91. Packard, E.L. 1947. A Pinniped Humerus from the Astoria Miocene of Oregon. Oregon State Monographs, Studies in Geology (Oregon State College, Cor- vallis), 7:23-32. Packard, E.L., and R. Kellogg 1934. A New Cetothere from the Miocene Astoria Formation of Newport, Oregon. Carnegie Institution of Washington Publication, 447:1-62. Queiroz, K de, and J. Gauthier 1990. Phylogeny as a Central Principle in Taxonomy: Phylogenetic Defi­ nitions of Taxon Names. Systematic Zoology, 39:307-322. Ray, C E . 1977. Fossil Marine Mammals of Oregon. Systematic Zoology, 25:420^136. Repenning, C A . 1976. Adaptive Evolution of Sea Lions and Walruses. Systematic Zoology, 25:375-390. Repenning, C.A., and R.H. Tedford 1977. Otarioid Seals of the Neogene. U.S. Geological Survey Professional Paper, 992:1-93. Ronald, K., and P.J. Healey 1981. Harp Seal, Phoca groenlandica Erxleben, 1777. In S.A. Ridgway and R.J. Harrison, editors, Handbook of Marine Mammals, 2: Seals, pages 55-87. San Francisco: Academic Press. Rowe, T. 1988. Definition, Diagnosis, and Origin of Mammalia. Journal of Verte­ brate Paleontology, 8:241-264. Siversten, E. 1954. A Survey of Eared Seals (Family Otariidae) with Remarks on the Antarctic Seals Collected by M/K "Norvegia" in 1928-1929. Det Vorske Videnskaps-Akademii Oslo, 36:1-76. Snavely, P.D., Jr., N.S. MacLeod, and H.C. Wagner 1973. Miocene Tholeiitic Basalts of Coastal Oregon and Washington and Their Relations to Coeval Basalts of the Columbia Plateau. Bulletin of the Geological Society of America, 84:387-424. Swofford, D.L. 1998. PA UP*: Phylogenetic Analysis Using Parsimony (*and Other Meth­ ods), Version 4.0. Sunderland, Massachusetts: Sinauer Associates. Turner, D.L. 1970. Potassium-Argon Dating of Pacific Coast Miocene Foraminiferal Stages. In O.L. Bandy, editor, Radiometric Dating and Paleontologi­ cal Zonation. Geological Society of America Special Paper, 124: 91-129. Wyss, A.R. 1987. The Walrus Auditory Region and Monophyly of Pinnipeds. Ameri­ can Museum Novitates, 2924: 38 pages. Wyss, A.R., and J.J. Flynn 1993. A Phylogenetic Analysis and Definition of the Carnivora. In F.S. Szalay, M.J. Novacek, and M.C. McKenna, editors, Mammal Phy­ logeny. pages 32-52. Berlin: Springer-Verlag. The Fossil Monk Seal Pontophoca sarmatica (Alekseev) (Mammalia: Phocidae: Monachinae) from the Miocene of Eastern Europe Irina A. Koretsky and Dan Grigorescu A B S T R A C T We present a phylogenetic analysis of the middle Miocene Euro­ pean seal Pontophoca sarmatica (Alekseev) based upon characters of the mandible, humerus (both described herein for the first time), and femur. The diagnoses of the subfamily Monachinae and the genus Pontophoca are emended to include the postcranial charac­ ters. Pontophoca, as revised, is proposed as the sister group of the modem Monachus and is included in the monophyletic subfamily Monachinae. We also include in this subfamily two extinct taxa that occur in both the eastern United States and western Europe: Callophoca and Pliophoca. Introduction The purpose of this study is to clarify the taxonomic status of Pontophoca sarmatica (Alekseev, 1924) in light of new mate­ rial found in the last 30 years and of previously undescribed bones in various European collections. The most numerous fossil remains of true, or earless, seals (Phocidae) in the Old World have been found in the middle Sarmatian-Maeotian, and probably Pontian, deposits of the Eu­ ropean part of the former USSR, especially in the northern coastal region of the Black Sea in Ukraine, Moldavia, and Ro­ mania. In this study we seek to improve the state of knowledge of Monachinae of the middle Sarmatian and Maeotian, and possibly of the early Sarmatian and Pontian, as well as (within limits) of the subfamily as a whole. We also present a phyloge­ netic analysis and classification based upon morphological characters of the mandible, humerus, and femur of the Mo­ nachinae. The age of the material of Pontophoca sarmatica was con- Irina A. Koretsky, Research Associate, Department of Paleobiology, National Museum of Natural History, Smithsonian Institution, Wash­ ington, D.C. 20560-0121, United States. Dan Grigorescu, Faculty of Geography-Geology, Laboratory of Paleontology-Stratigraphy, Uni­ versity of Bucharest, Romania. sidered to be late Tertiary by Eichwald (1850) and von Nord­ mann (1858-1860), whereas Andrusov (1893) specified the age more precisely as Upper Miocene (Sarmatian). Further studies on Sarmatian marine mammals were made early in the last century. Alekseev (1924, 1926) described two new species, Phoca sarmatica and Phoca novorossica. About the same time, in his study of true seals of the northern littoral region of the Black Sea, Simionescu (1925) also described two other species—P. maeotica and Ppontica. As can be judged by the illustrations of the femur in his study (pl. 1: fig. 2), Simio­ nescu included in "P." pontica the seal described previously by Alekseev (1924) as P. sarmatica and noted that these finds date from the Sarmatian period. Following Simionescu (1925), Macarovici and Oescu (1942) and Macarovici (1942) pub­ lished short reviews of information on fossil seals of the Euro­ pean Sarmatian, which only confused the picture. Meanwhile, Kretzoi (1941), who tried to classify the Neo­ gene seals of this region, proposed three new genera: Prae- pusa, Pontophoca, and Monachopsis. The importance of Kret- zoi's work is that he attempted for the first time to make comparisons of previously known fossil material. Unfortu­ nately, he did not succeed completely. King (1956), in her monographic review of monk seals, pre­ sented the first description of bones of the postcranial skeleton as well as descriptions and measurements of the skulls and mandibles of modern species of Monachinae. Of special inter­ est in this context is an article by McLaren (1960), who, on the basis of previous publications, revised the two subfamilies of true seals of the northern Black Sea coastal region of the former USSR. King (1964), in the first edition of her mono­ graph on seals of the world, presented her conception of fossil seals of the Miocene of the northern Black Sea coastal region, separating them into four species of Phoca and two other spe­ cies referred to Monotherium and Pontophoca. Subsequently, King (1983) changed her views on classification of the true seals, but she considered only their classification above the rank of tribe. Grigorescu (1977), in his article on the seals of 149 150 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY Paratethys, presented a detailed description of postcranial bones from southern Romania. He also discussed the evolution and phylogeny of Sarmatian seals. Miocene seals in the North­ ern Hemisphere were reviewed by Ray (1977), who also con­ sidered routes of penetration of Phocidae into the North Atlan­ tic during the Neogene. ACKNOWLEDGMENTS.—The authors thank all persons and scientific organizations who permitted access to paleontologi­ cal and osteological materials, helped on expeditions, and aided in the writing of this study, especially Y. A. Semenov and S.I. Ivanova, both from the Paleontological Museum of the Ukrainian Academy of Sciences, Kiev, Ukraine, and A.A. An- toniuk from the Military Medical Academy, Saint Petersburg, Russia. We thank Ann Forsten and Mikael Fortelius, both of the Museum of Zoology, Helsinki, Finland, for loan of fossil seals from the Nordmann collection. D. Domning, of Howard University, Washington, D.C, and Diana Lipscomb, of George Washington University, Washington, D.C, reviewed the manu­ script at various stages and made many helpful suggestions. REPOSITORIES.—The following abbreviations are used for institutions housing specimens used in this study: IZUAN Institute of Zoology (named after I.I. Shmalhausen) of the Academy of Sciences of the Ukraine, Kiev JaU Department of Geology, University of Jassy, Romania MZHF Museum of Zoology, Helsinki, Finland OGUM Paleontological Museum (named after I.I. Mechnikov), Odessa State University, Ukraine PIN Paleontological Institute of the Academy of Sciences of Rus­ sia, Moscow TGPI Tiraspol State Pedagogical Institute, Moldavia UBFG Faculty of Geography-Geology, University of Bucharest, Ro­ mania USNM Collections of the National Museum of Natural History, Smithsonian Institution, Washington, D.C. (includes collec­ tions of the former United States National Museum) ZIN Zoological Institute of the Academy of Sciences of Russia, Saint Petersburg ZKM Zaporozhye Museum of Regional Studies, Zaporozhye, Ukraine MATERIAL AND METHODS For solving diagnostic problems, we used the methods of As- tanin (1936), Chapskii (1952, 1967), and Antoniuk (1970, 1972). Morphometric analysis of skulls and mandibles was based upon the methods of Chapskii (1955, 1974), Semenov (1981), Andreescu and Murariu (1985), and the methods de­ scribed below. Bones of the postcranial skeleton and skulls of both Miocene and extant Monachinae were measured accord­ ing to the schemes of Marcoci and Popa (1957), Dornesco and Marcoci (1958), Sergienko (1967), Pierard (1971), Driesch (1976), Muizon (1981), Antoniuk and Koretsky (1984), Ko­ retsky (1987), and Koretsky and Ray (1994). Anatomical ter­ minology follows the International Anatomical Nomenclature edited by Michaylov (1980), the Anatomical Atlas by Sinelni- kov (1963), and Pierard (1971). The information presented below on geographic location and geologic age of the finds as well as on collectors and institu­ tional repositories is compiled from data published by Kellogg (1922), Pidoplichko (1956), Gromova et al. (1962), Godina (1973), Dubrovo and Kapelist (1979), Korotkevich et al. (1985), and Semenenko (1987) and from our unpublished data. In this study we use the stratigraphic scheme of eastern Paratethys published by Chepalyga et al. (1985). This cladistic study is intended to clarify the phylogenetic re­ lationships among modern and fossil species of monachine seals. Six species of Monachinae were analyzed, along with one species of Cystophorinae (Cystophora cristata) and one species of Phocinae (Leptophoca lenis) as outgroups, using 48 cranial and postcranial morphological characters. Originally, 62 characters were examined, but 14 were eliminated because they could not be examined in the fossil taxa. The 48 informa­ tive characters were analyzed with the Hennig86 computer pro­ gram (Farris, 1988). Systematic Paleontology Superfamily PHOCOIDEA Smirnov, 1908 Family PHOCIDAE Gray, 1825 Subfamily MONACHINAE Trouessart, 1897 TYPE GENUS.—Monachus Fleming, 1822; present distribu­ tion: Mediterranean Basin, Atlantic Ocean (North Africa), Ha­ waiian Islands, Gulf of Mexico, Caribbean Sea (probably ex­ tinct in the last two areas). DISTRIBUTION.—Middle Miocene to the present; Mediterra­ nean Basin, North Atlantic, North Pacific, Antarctic region. EMENDED DIAGNOSIS.—Large seals with eight incisors (1= 2/2). Mastoid with wide convexity; convexity not strongly lat­ erally protruding and not turned abruptly downward behind mastoid process (King, 1966). Maxilla anterior to orbits slightly concave. Anterior palatal foramen tending to disap­ pear, according to Chapskii (1974). Mandibular chin promi­ nence present; posterior symphysis border reaching at least to middle of alveolus for p3. Middle of internal crest of humeral trochlea raised arch-like over coronoid fossa; width of distal epiphysis exceeding width of proximal epiphysis by one-fifth to one-sixth. Width of distal femoral epiphysis greater than that of proximal epiphysis by one-quarter to one-fifth; minimum width of femoral shaft more than two-thirds of the proximal ep­ iphysis width; intertrochanteric crest weakly developed. INCLUDED TRIBES.—Monachini Scheffer, 1958; Lobodontini Scheffer, 1958. COMPARISONS.—The interorbital width at the frontal bones in Monachinae is wider than in Phocinae, although not as wide as in Cystophorinae. The anterior part of the frontal has a fossa that is directed medially. On the midline at the fronto-nasal su­ ture is the origin of the very low sagittal crest, which is com­ pletely absent in the other two subfamilies. In contrast to the NUMBER 93 15: condition in Cystophorinae, the part of the maxilla located be­ tween the nares and orbits is wide. Just as in Phocinae, an inter­ val is present between the external auditory meatus and the postglenoid process. The jugal bone, as in Cystophorinae, has an antero-orbital process. The bolster-like convexity of the mastoid is strongly compressed and is directed laterally. The anterior palatal foramina tend to disappear. The symphyseal part of the mandible is very strongly devel­ oped; it is straight and its posterior edge is considerably dis­ placed posteriorly relative to its position in the other subfami­ lies. The middle of the crest of the humeral trochlea, unlike that in Phocinae, is arch-like in shape and raised over the coronal fora­ men. Both distal condyles of the femur are of equal dimensions, unlike those in Phocinae. The difference in width of the distal and proximal epiphyses is significant. DISCUSSION.—Up to the present there has been no clear con­ ception of the relationships within this subfamily. Previously, in accordance with the classification of Trouessart (1897), the subfamily Monachinae contained the genera Lobodon Gray, 1844; Ommatophoca Gray, 1844; Hydrurga Gistel, 1848; and Leptonychotes Gill, 1872. Simpson (1945), however, placed Lobodon and the other three genera mentioned above in the subfamily Lobodontinae. Placement of these four genera is still a controversial problem. One group of investigators has as­ signed them to a single subfamily (Ognev, 1935; Grasse, 1955; King, 1964, 1983; Tedford, 1977; Muizon, 1982), whereas oth­ ers (Wyss, 1988; Muizon, 1992) have separated them into two subfamilies. Finally, some investigators (Sokolov, 1979; Pavli- nov and Rossolimo, 1987; Wozencraft, 1989) have chosen not to separate true seals (Phocidae) into subfamilies at all. Chapskii (1955, 1961, 1971, 1974) presented a comprehen­ sive series of analyses of the suprageneric systematics of pinni­ peds. Analyzing the crania, he clearly described diagnostic characters that separate true seals into three subfamilies: Phoci­ nae, Monachinae, and Cystophorinae, which he in turn divided into tribes and subtribes. This is also true for the tribes Mona- chini and Lobodontini. Chapskii's (1974) detailed analysis proved King's (1966) hypothesis to be untenable; however, King (1983) persisted in her hypothesis that the genus Cysto- phora should be transferred from the subfamily Cystophorinae into the subfamily Phocinae, and that the genus Mirounga should be moved into the subfamily Monachinae. It may be as­ sumed that Muizon (1982) was unaware of the study by Chapskii (1974) because he accepted the systematics of King (1964, 1966) without any reservations, and thus he returned to the concept of subdivision of the subfamily Cystophorinae. Chapskii's concept is supported by the conclusion of Robin- ette and Stains (1970) in their comparative study of the pin­ niped calcaneus. These authors emphasized that it is inadmissi­ ble to separate the hooded seal and the elephant seal taxonomically. This point of view was supported by Anbinder (1980:76) who noted that "modern analytical methods of chro­ mosome investigations actually do not permit the separation of genera Cystophora and Mirounga, and this contrasts with the concept of their separate taxonomic status and of inclusion of Cystophora in Phocinae." In our view, the problem of the status of the Cystophorinae is solved, and we support Chapskii's point of view that the subfamily Cystophorinae is valid. Tribe MONACHINI Scheffer, 1958 TYPE GENUS.—Monachus Fleming, 1822; present distribu­ tion: Mediterranean Basin, southern North Atlantic, North Pa­ cific. DISTRIBUTION.—Middle Miocene to the present; Europe, southern North Atlantic, North Pacific. DIAGNOSIS.—Maxillary process of jugal bone clearly out­ lined. Lower edge (masseteric margin) of jugal bone arched up­ ward, elevated to greatest degree in middle part of bone. An- tero-upper process of jugal bone reaching level of infraorbital foramen, terminating almost over inferior edge of infraorbital foramen. Lower edge of orbit at same level as infraorbital fora­ men. Nasal bones not united with each other; frontal contact of na­ sals not longer than maxillary contact. Turbinals (fontanelles) in presphenoid region huge and round (see Chapskii , 1971:311). INCLUDED GENERA.—Monachus Fleming, 1822; Monathe- rium Van Beneden, 1877; Callophoca Van Beneden, 1877; Pliophoca Tavani, 1941; Pontophoca Kretzoi, 1941. COMPARISONS.—Representatives of the Monachini are dis­ tinguished from the Lobodontini by (1) an arch-like bending of the jugal bone, (2) the position of the infraorbital foramen on a level with the greatest deflection of the upper border of the ju­ gal bone, and (3) separate nasal bones. DISCUSSION.—Several other genera were earlier included in the Monachinae, but now these assignments are considered un­ certain or even wrong. We review the status of some of them below; Mesotaria Van Beneden (1877) and Pristiphoca Ger­ vais and Serres (1847) will be discussed in detail by Koretsky and Ray (in prep.). Although the author of "Miophoca" vetusta (Zapfe, 1937), as well as Simpson (1945), Thenius (1950, 1952), King (1964), and Holec et al. (1987), assigned this western European genus to the subfamily Monachinae and dated it to the middle Mi­ ocene, other investigators (Kirpichnikov, 1961; Ray, 1977; Muizon, 1982; Savage and Russell, 1983) did not mention this genus at all in their reviews of Tertiary seals of Europe. The­ nius (1950, 1952) assigned this species to Pristiphoca, and Holec et al. (1987) supported the opinion of Thenius with new cranial remains from Devinska Nova, Slovakia. The uncer­ tainty of the taxonomic position is because of the incompletely known morphology of "Miophoca,'1 which also precludes a de­ tailed comparison between "Miophoca" and Pontophoca. It should be pointed out, however, that in accord with the opinion of Zapfe (1937), the very distinctive, characteristic morphotype 152 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY of the mandible shows that representatives of the genus "Mo- phoca" are undoubtedly ancestral to Cystophora. The problem of whether "Miophoca" belongs in the subfamily Monachinae still remains open, however. Genus Monachus Fleming, 1822 EMENDED DIAGNOSIS.—Condylobasal length of skull not exceeding 200 mm. Facial part of skull markedly lower than occipital part, and with large infraorbital processes. Pl single rooted. Diastemata between teeth absent. Basal cingulum well developed. Main cusp on all cheek teeth sharp-triangular in form. Supplementary cusps weakly developed. Carotid canal displaced almost to bottom of triangular tympanic bulla. Bony blade of external auditory meatus relatively weakly developed. Foramen ovale more or less covered by hamular process of sphenoid bone. Transverse measurement of glenoid fossa of mandible equal to or slightly exceeding longitudinal dimension of tympanic bulla. Jugular process not conjoined with mastoid process, and with a convexity on anterolateral part. Deltoid crest of humerus short, not reaching middle of the shaft, and distended in proximal quarter; lesser tubercle oval shaped, with height exceeding height of head; head flattened proximodistally; ratio of width to height of head greater than 0.90; intertubercular sulcus narrow and deep; epicondyles strongly developed; distal epiphysis wider than the proximal epiphysis; coronoid fossa shallow. Greater trochanter of femur slightly higher than femur head and square in form; slightly distended in proximal part; tro­ chanteric fossa shallow, rounded, and ending in anterior one- third of trochanter. Minimal width of shaft in middle of femur. Distal condyles similar in size and placed widely apart; maxi­ mal distance across epicondyles 0.61 times length of femur; in­ tercondylar area flat; distal end of femur wider than proximal end by -115%. DISCUSSION.—This is the first time that characters of the hu­ merus and femur have been included in the diagnosis of Mona­ chus, except for the short general description by King (1956: 239, 241). Because the material of the other fossil taxa needed for analysis is totally lacking or is fragmentary, we cannot make a more detailed comparison with genera in the tribe Mo­ nachini. For example, there is no information on skull frag­ ments, mandibles, or femora of the genus Monatherium, and the fossils assigned to Paleophoca represent a cetacean (Ko­ retsky and Ray, in prep.). Genus Pontophoca Kretzoi, 1941 TYPE SPECIES.—Phoca sarmatica Alekseev, 1924; middle Sarmatian of Kishinev. DISTRIBUTION.—Middle Miocene of eastern Europe. EMENDED DIAGNOSIS.—p3 and p4 double rooted, placed parallel to the tooth-row axis. Protoconid on p4 triangular. Di­ astemata present. Metaconid and basal cingulum weakly devel­ oped. Deltoid crest of humerus terminating lower than middle of shaft; proximal part of crest recurved posterodorsally; distal end of bone considerably wider than proximal end; lesser tu­ bercle located higher than proximal end of deltoid crest and head; ratio of width to height of head about 1.00; supracondy­ lar crest strongly developed. Greater trochanter of femur slightly higher than head and very oblique; distal end of greater trochanter narrower than proximal end. Distal end of femur broader than proximal end by 111%-114%; trochanteric fossa shallow and elongated along femur axis; head of femur very small relative to massive bone and seated on narrow neck; minimal width of shaft lo­ cated in proximal part of femur between head and distal part of greater trochanter; distal condyles widely separated; maximal distance across epicondyles -70% of femur length. INCLUDED SPECIES.—In the middle Sarmatian of the north­ ern Black Sea littoral region, only one monachine species (the type species) is recorded—"Phoca" sarmatica from Moldavia and possibly from the middle part of the Don River (Alekseev, 1924). COMPARISONS.—Among the modern and fossil representa­ tives of the subfamily Monachinae, the extinct genus Ponto­ phoca can be compared only with Monachus, Pliophoca, Cal- lophoca, and partly with Monatherium. It differs from all of these taxa as follows: (1) The teeth are oriented parallel to the axis of tooth row, with diastemata between the teeth; cheek teeth have one anterior and one posterior accessory cusp (ex­ cept in M. monachus); basal cingulum is weakly developed (except in Monachus). (2) The deltoid crest of the humerus is shorter (except in Monachus); the distal part of the deltoid crest is narrow, and the proximal part is averted posterodorsally; the development of a coronoid fossa and intertubercular sulcus is rudimentary (except in Callophoca and Monatherium); the lesser tubercle is elongated; the head is more spherical; the de­ velopment of the medial supracondyles is relatively stronger. (3) The medial border of the femoral body is distended; the su­ pracondyles are of different heights and are thickened at the points of attachment of both heads of the gastrocnemius mus­ cle; the lateral supracondyle, however, is thickened only at the point of attachment of m. extensor digitorum longus. Apart from these characters, this genus differs from Monatherium as follows: the width of the distal end of the hu­ merus is greater than that of the proximal end; the lesser tuber­ cle extends above the head and above the proximal part of the deltoid crest; and development of the lateral epicondyloid crest is considerably stronger. It differs from Pliophoca as follows: The body of the mandi­ ble is higher. The lesser tubercle extends above the head of the humerus; the radial sulcus is absent; the deltoid crest is widest in its proximal part. The trochanteric fossa of the femur is shal­ low and wide; the least width of the shaft is in the proximal part of the bone; and the condyles are widely spaced and flat. NUMBER 93 153 FIGURE 1.—Mandible of Pontophoca sarmatica. Cast of fragment of R. mandible, collection of ZKM, unnum­ bered (with p3, p4) from village of Vasilyevka, Zaporozhye Region, bank of Kachovka Reservoir (Mount Lysaya), Ukraine. In labial (a), lingual (b), and occlusal (c) aspects. (Scale bar=l cm.) It differs from Callophoca in the following characteristics: the greater trochanter of the femur extends higher above the head, and the condyles are flat. DISCUSSION.—Like most taxa erected by Kretzoi, the genus Pontophoca was not diagnosed satisfactorily. For this reason, until recently this genus was recognized by only a few special­ ists, such as McLaren (1960) and Kirpichnikov (1961). Repre­ sentatives of this genus were usually assigned either to Phoca or to Monachus. Some species included in these genera un­ doubtedly belong to Pontophoca. Von Nordmann (1858) was the first investigator who ana­ lyzed the bones of the extremities of Pontophoca sarmatica, but he described them under the name Phoca maeotica, to­ gether with the remainder of his material. Alekseev (1924) sep­ arated these specimens from the Nordmann collection, supple­ mented them with material kept at Odessa University (Ukraine), and described them as Phoca sarmatica. Some au­ thors (Grigorescu, 1977; Trelea and Simionescu, 1985; Mui­ zon, 1992) still use the name Phoca when referring to them. We view this assignment as incorrect, because the femur and humerus differ so distinctly from those of Phoca that the possi­ bility of confusion is practically excluded. The importance of Grigorescu's (1977) work is that he recognized characters diag­ nostic of the Monachinae in bones of Pontophoca sarmatica. In his opinion, the strong development of the gastrocnemius mus­ cle and the location of its attachment stimulated considerable expansion of the distal part of the femur. Our material confirms this, but all femora that we studied also have a very well devel­ oped plantar fossa, the place for attachment of the plantaris muscle, so we can add that not only the gastrocnemius muscle but also the plantaris contributed to expansion of the distal part of the femur. To remove some doubts on the correctness of the assignment of Pontophoca to the subfamily Monachinae, we compared its femora with those of young individuals of Monachus mona­ chus. These femora are very similar to those of Pontophoca sarmatica in the form of the bone, the obliqueness of the greater trochanter, the narrow neck, and the relatively small head. For these reasons we consider the assignment of"Phoca" sarmatica to the Monachinae to be well founded. In the "'Comparisons" section above we drew attention to the common structural features of the mandibles in P. sarmatica, M. monachus, M. tropicalis, and M. schauinslandi. These com­ mon features are evidence of the common origin of these four species (Repenning and Ray, 1977) and also suggest that Pon­ tophoca might be ancestral to the genus Monachus. The cladis­ tic analysis below confirms this (Figure 5). Pontophoca sarmatica (Alekseev, 1924) McLaren, 1960 FIGURES 1-4; TABLES 1-3 Phoca maeotica Nordmann, 1860:356-357, pl. 23: figs. 3, 7; pl. 24: fig. 1. Phoca moeotica [sic].-—Nordmann, 1860:317. Phoca pontica—Kellogg, 1922:120 [in part].—Simionescu, 1925:180, 188, 190-191, fig. 5P; pl. 1: fig. 2.—Macarovici and Oescu, 1942:351-352, 363- 367, 378-379, figs. 7, 8; pl. 2: figs. 18, 19.—Macarovici, 1942:262-263.— McLaren, 1960:51. Phoca sarmatica.—Alekseev, 1924:203, figs. 4-7.—Friant, 1947:12.—Pidop- lichko, 1956:142.—McLaren, 1960:57.—Kirpichnikov, 1961:29,32,34,36.— Aslanova, 1965:52.—Grigorescu, 1977:407^11, 413—418, fig. 5D.—Dubro­ vo and Kapelist, 1979:36.—Trelea and Simionescu, 1985:19.—Muizon, 1992: 35. Phocapontica var. sarmatica.—Macarovici, 1942:263, 267, pl. 2: fig. 18.1. Pontophoca simionescui.—Kretzoi, 1941:354, fig. 3.2. Monachus [sic].—Friant, 1947:6, 16, 47-50, pl. 1: fig. la-c. Pontophoca sarmatica.—McLaren, 1960:47, 52, 57, fig. lg,h,i.—King, 1964: 131. HOLOTYPE.—Femur described and illustrated by Alekseev (1924:202, fig. 6). Phoca sarmatica (McLaren, 1960:57); col­ lection of OGUM, Moldavia, middle Sarmatian. DISTRIBUTION.—Middle Sarmatian of northern Black Sea coastal region (southern Ukraine and possibly Russia). 154 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY TABLE 1.—Measurements (mm) of mandible and lower dentition of Ponto­ phoca sarmatica, collection of ZKM, unnumbered, incomplete right ramus with p3 and p4. TABLE 2.—Measurements (in mm) of humeri of Pontophoca sarmatica. Character p3 length p4 length Ramus depth under p4 p3 height p4 height p3 width p4 width Diastema length between p3 and p4 Measurement 14.0 20.0 36.5 8.7 15.0 7.0 10.0 4.5 REFERRED MATERIAL.—Kishinev: PIN 1713/10, R. humerus and L. femur from one individual; PIN 1713/23, femur from another individual; OGUM 23, two humeri, eight femora (R. and L.); OGUM 53-57, one scapula, two fused tibiae and fibu­ lae, five isolated tibiae, three isolated fibulae, and two pelvic bones (this material was described by Alekseev (1926)); ZIN 4, one femur, collection of von Nordmann; ZIN 8, one femur, col­ lection of von Nordmann; UBFG 259, L. and R. femora, col­ lection of Simionescu; USNM 214980, cast of R. femur (origi­ nal UBFG 259, collection of Simionescu); JaU MS20L, femur, collection of Simionescu (this material was described and illus­ trated as Phoca pontica by Macarovici and Oescu (1942: 351-352, 363-367, 378-379, figs. 7, 8, pl. 2: figs. 18, 19); Character Total length Length of deltoid crest Height of head Height of trochlea Width of head Width of deltoid crest Width of proximal epiphysis Width of distal epiphysis Width of trochlea distally Width of trochlea, anteriorly Transverse width of diaphysis Thickness of proximal epiphysis Thickness of medial condyle Thickness of lateral condyle Diameter of diaphysis with deltoid crest n 1 7 5 5 5 5 7 7 5 5 8 7 7 5 8 X 111.1 67.0 25.8 20.1 25.1 30.8 33.3 36.1 20.1 17.0 16.9 38.7 18.1 18.2 35.2 Range 100.5-120.0 64.0-73.0 25.5-28.0 19.5-23.0 23.0-27.5 27.0-33.0 28.0-38.0 30.1-47.0 20.0-24.0 16.5-18.0 15.3-19.1 37.3-43.0 15.0-22.0 16.5-22.0 31.0-45.0 MZHF 1811, R. humerus, illustrated by von Nordmann (1858, pl. 23: fig. 3) and described by him (1860:317, 356-357) as Phoca maeotica; UBFG unnumbered, two L. and one R. fem­ ora; UBFG 249, L. humerus; MZHF unnumbered, five R. fem­ ora, collection of von Nordmann. Vicinity of Tiraspol, Moldavia: OGUM 6, 23-25, six humeri (two of which are unnumbered). Stanitsa Tsymlyansk on the Don River: OGUM 9, humerus. FIGURE 2.—Humerus of Pontophoca sarmatica. Cast of R. humerus, collection of MZHF 1811, from Kishinev, Moldavia. This was illustrated by von Nordmann (1858, pl. 23: fig. 3) and described by him (1860:317, 356-357) as Phoca maeotica. In cranial (a), caudal (b), medial (c), and lateral (d) aspects. (Scale bar=l cm.) NUMBER 93 155 FIGURE 3.—Femur of Pontophoca sarmatica. Cast of R. femur, USNM 214980 (original UBFG 259, collection of Simionescu), from Kishinev, Moldavia. In cranial (a) and caudal (b) aspects. (Scale bar=l cm.) Like the material from the foregoing localities, this material was described by Alekseev (1924). Kerch Peninsula (Kamysh-Burun): IZUAN 64-357, 64-361, 64-362, three femora. Village of Vasilyevka, Zaporozhye region; bank of Kachovka Reservoir (Mount Lysaya): ZKM unnumbered, incomplete R. ramus of mandible with p3 and p4. DIAGNOSIS.—Same as for the genus. DESCRIPTION.—Mandible (Figure 1, Table 1): The long axes of the teeth are parallel to the tooth-row axis. The di­ astema between p3 and the alveolus of p2 is longer than the di­ astema between p3 and p4 Lower p4 is considerably larger than p3. The height of p3 exceeds by only 3.0 mm the para­ conid on p4. The basal cingulum and metaconid are weakly de­ veloped on both teeth. Scapula: "Its distinctive features are the strong develop­ ment of the muscular spine, thickened summit and thick, mas­ sive acromion. The articular surface is very narrow; the tuber­ culum supraglenoidale and cervix scapulae are very weakly pronounced" (Alekseev, 1924:202). Humerus (Figures 2, 4, Table 2): The lesser tubercle is large, elongate parallel to the bone's axis, and proximally higher than the head, practically on the same level as the proxi­ mal border of the deltoid crest. The intertubercular sulcus is ab­ sent. The head is spherical. In well-preserved specimens, the ratio between the length and the width of the head is almost 1.00. The deltoid crista ends in the distal one-third of the bone. This crest is markedly convex, and the deltoid tuberosity is strongly swollen. The coronoid fossa is barely outlined in smaller (i.e., juvenile) bones; in larger individuals it is absent. The radial sulcus is absent. In well-preserved bones, significant distention of the two epicondyles is clearly seen. The lateral ep- icondyle is wide; in height it practically reaches the distal part of the deltoid crest. The medial epicondyle in a large individual (evidently an adult) barely reaches the height of the lateral su- pracondyle. The entepicondylar fossa is seen on all specimens, but in sexually mature individuals this fossa is covered on its medial side by a thicker and wider wall. Pelvis: The ilia are considerably thickened and distended. "The acetabulum is very deep. Its diameter is rather small, less than in "Phoca" maeotica and with very pronounced borders" (Alekseev, 1924:202; 1926:138-143). Femur (Figures 3, 4, Table 3): The femora are closely sim­ ilar in size to those of the living seal of the genus Histriophoca. The greater trochanter is higher than the head, very obliquely oriented, and elongated along the bone's long axis. In smaller individuals from the Kerch Peninsula (these individuals proba­ bly were younger; see Astanin, 1936; Heptner, 1947), it is less oblique and, consequently, more protruding. In its distal part the greater trochanter has a V-shaped end. The length of the greater trochanter varies considerably (from 19.5 mm to 52.0 mm), depending upon the individual's age. The trochanteric fossa is shallow and wide and reaches the middle of the greater trochanter's length. This fossa is open on the proximal side of the greater trochanter. Relative to the bone's mass, the femoral head is very small and is placed on a narrow, long neck. The least width of the shaft is located in the proximal part of the bone. The distal condyles are flat and very widely spaced. The maximum distance between them (10 .0 -15 .5 mm) is 60%-80% of the bone's length. In young individuals a plantar fossa is present on top of the lateral epicondyle. In adult indi­ viduals a considerable swelling of the bone is seen at this loca- 156 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY FIGURE 4.—Humerus and femur of Pontophoca sarmatica from one individual, originals, collection of PIN 1713/10, from Kishinev, Moldavia. Illustrations of R. humerus in medial (a) and cranial (b) aspects; L. femur in cranial (c) and caudal (d) aspects. (Scale bar=l cm.) tion. The characteristic, distinctive feature of this species is sig­ nificant distention (expansion) of the distal part of the bone, which in adult individuals is 1.4—1.5 times wider than the prox­ imal part (in juveniles 1.2 times). Tibia and Fibula: Distinctive features are a narrower dia­ physis and markedly thickened epiphysis; "the distal condyles of the femur are, consequently, widely placed; the articular sur­ face bordering this distal end of the femur is significantly dis­ tended" (Alekseev, 1924:202). DISCUSSION.—According to published data, this type species is a widely known representative of the Monachinae. It has not, however, been adequately described. In particular, the limits of its dimensions are unknown, as are most of the bones of its skeleton. McLaren (1960) thus stated that it is of smaller di­ mensions than Cryptophoca maeotica (formerly Phoca maeot­ ica), which was revised by Koretsky and Ray (1994); but Alek­ seev (1924) stated that some bones (pelvis) are of equal dimensions with those of the latter species. Kretzoi (1941) il­ lustrated the femur of Pontophoca sarmatica as being smaller than that of "Phoca'1 pontica. According to our information, however, P. sarmatica is considerably larger than C. maeotica, and even more so than Phoca pontica. Although detailed stud­ ies were made of femora of P. sarmatica, no one except Alek­ seev has studied other skeletal parts. We have described herein for the first time a humerus and femur belonging to one indi­ vidual (PIN 1713/10 from Kishinev). The fact that these bones have the same pathology (not described) and were found not far from each other in one bed supports our opinion that they belong to the same individual. The skeletal parts (the scapula, pelvis, tibia, and fibula) described and illustrated by Alekseev (1924, 1926) were not cited as associated and were never stud­ ied by other authors (e.g., Kretzoi, 1941; Friant, 1947; McLaren, 1960); therefore, at present we cannot supplement their descriptions by our materials or confirm their species as­ signment. To describe this species more completely, we used the description of Alekseev for these missing skeletal parts of Pontophoca sarmatica. Future analysis of supplementary mate­ rials may show this hypodigm to be a composite, particularly in view of the considerable intraspecific variability of the hu­ merus and femur, and of the rather narrow geographic distribu­ tion (eastern Europe) of this seal as herein recognized. Also, there are no reliable records from the banks of the Don River, although Alekseev reported finding this species in the Lower Don. In addition, more than one large seal is known from the Sarmatian deposits. Alekseev's scapula, pelvis, tibia, and fibula can be questionably assigned to the same species. Cladistic Analysis CHARACTERS USED FOR SUBFAMILY MONACHINAE The data matrix for the 48 included characters is shown in Table 4. Characters and character states for Monachinae are listed below; 0 designates the most primitive state among the taxa studied; 1 and 2 are derived states; - indicates unknown or missing data. Skull 1. Mastoid process: (0) not strongly pronounced; (1) pro­ nounced. NUMBER 93 157 TABLE 3.—Measurements (in mm) of femora of Pontophoca sarmatica. Measurement Total length Medial length Lateral length Length of medial condyle Length of lateral condyle Length of greater trochanter Height of head Height of articular area of patella surface Intertrochanteric width Width of proximal epiphysis Width of distal epiphysis Width of condyles Width of greater trochanter Width of head Width of shaft Anteroposterior thickness of shaft Thickness of medial condyle Thickness of lateral condyle Distance between condyles Diameter of head n 16 8 9 13 17 15 6 6 9 13 19 13 16 11 15 12 8 13 16 13 Adult X 82.5 78.3 75.3 16.6 18.0 28.0 18.5 17.2 29.0 40.8 59.2 46.8 16.1 18.9 26.0 11.7 17.5 25.3 12.7 15.9 Range 76.5-96.0 74.0-88.0 71.5-80.5 15.0-21.0 16.0-22.0 25.0-30.0 17.2-23.0 15.9-22.0 28.5-38.0 36.0-44.0 54.5-66.0 42.6-50.5 14.5-18.0 17.5-20.7 24.5-29.5 11.0-12.5 16.0-20.0 24.0-27.0 12.0-15.5 14.0-17.5 n 2 2 3 2 3 3 2 2 3 2 2 2 3 2 3 3 2 2 2 2 Juvenile X - - 60.7 - 16.8 21.3 - - 25.5 - - - 13.7 - 18.7 11.2 - - - - Range 68.0; 68.5 64.5; 68.0 58.0-62.5 12.0; 12.0 16.0-18.0 20.5-22.0 15.0; 15.5 16.5; 20.0 25.0-26.0 34.0; 35.0 39.0; 39.5 33.0; 35.0 13.0-14.0 15.0; 15.5 18.0-19.0 10.5-12.0 15.0; 15.5 17.0; 23.0 11.0; 11.0 12.5; 14.0 2. Mastoid process: (0) narrow; (1) wide (Chapskii, 1974:301; polarity opposite to Berta and Wyss, 1994:48). 3. Maxilla: (0) convexity anterior to the orbits; (1) short concavity; (2) long concavity (polarity opposite that of Berta and Wyss, 1994:46). 4. Anterior palatine foramina: (0) faintly marked; (1) oval and shallow (Burns and Fay, 1970:372). 5. Interorbital width: (0) less than 25.0% width of skull across mastoids; (1) less than 30.0%, but equal to or greater than 25.0% of mastoid width; (2) equal to or greater than 30.0% of mastoid width (Burns and Fay, 1970:370; Chapskii, 1974:299). 6. Jugular process: (0) poorly developed; (1) well devel­ oped. 7. Rostrum: (0) elongate; (1) short, compared with cra­ nium (Chapskii, 1974:300). 8. Diameter of infraorbital foramen: (0) less than diame­ ter of alveolus of maxillary canine; (1) approximately equal to diameter of alveolus of maxillary canine; (2) greater than diam­ eter of alveolus of maxillary canine (polarity opposite that of Berta and Wyss, 1994:47). 9. Anteroposterior length of auditory bullae: (0) greater than distance between them; (1) less than distance between them; (2) about equal to distance between them (Burns and Fay, 1970:382; Chapskii, 1974:300) (unordered character). Mandible 10. Symphyseal part: (0) continues at least to the middle of the alveolus of p3; (1) reaches only to the alveolus of p2; (2) reaches only to the alveolus of pl . 11. Lateral outline of symphyseal region: (0) square, sym­ physis thin; (1) rounded, symphysis thick; (2) straight, sym­ physis thick. 12. Chin prominence: (0) pronounced; (1) absent or weakly outlined. 13. Chin prominence: (0) extends from the anterior or pos­ terior alveolus of p2 to the posterior or anterior alveolus of p4; (1) extends from the anterior alveolus of p2 to anterior alveolus ofp3. 14. Maximum height of body of mandible: (0) between p2 and p3; (1) in the middle or at the posterior portion of p2 (Ko­ retsky and Ray, 1994:21); (2) situated between alveoli of p4- ml. 15. Diastemata and tooth alveoli: (0) alveoli small, with equal diastemata; (1) alveoli round and large, with equal di­ astemata between them; (2) alveoli shallow, and diastemata un­ equal. 16. Alveoli of p4 and ml: (0) alveoli similar in size; (1) al­ veoli of p4 smaller than alveoli of ml; (2) alveoli of p4 larger than alveoli of ml (unordered character). 17. Retromandibular space: (0) long; (1) short. Teeth 18. Number of incisors: (0) 3/2; (1) 2/2; (2) 2/1 (Chapskii, 1974:289; polarity opposite that of Burns and Fay, 1970:380). 19. Roots of postcanine teeth (P,p 2-P,p 4): (0) one root, divided partially at the base; (1) two (polarity opposite that of Berta and Wyss, 1994:51). 20. Crowns of postcanine teeth: (0) single cusped; (1) multicusped (polarity opposite that of Berta and Wyss, 1994:51) (reversal to primitive condition, unordered character). 21. Relative dimensions of postcanine teeth: (0) large; (1) small. 22. Relative dimensions of canine: (0) large; (1) small. 23. Basal cingulum of postcanine teeth: (0) well devel­ oped; (1) not developed. 158 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY Species Leptophoca lenis* Cystophora cristata* Pontophoca sarmatica Monachus monachus Monachus schauinslandi Monachus tropicalis Pliophoca etrusca Callophoca obscura Species Leptophoca lenis* Cystophora cristata* Pontophoca sarmatica Monachus monachus Monachus schauinslandi Monachus tropicalis Pliophoca etrusca Callophoca obscura TABLE 4 .—Matrix of character-state data for monachine taxa and outgroups ana lyzed. (*= outgroup.) Character 1 0 1 1 1 1 0 0 25 0 0 0 1 1 0 1 1 2 0 1 1 1 0 0 0 26 0 1 - 0 0 0 0 0 3 0 1 2 2 2 2 2 27 0 1 - 1 1 1 1 2 4 1 0 0 0 1 0 0 28 0 1 0 5 0 1 0 0 0 2 1 29 1 1 0 0 0 0 0 0 6 0 0 1 0 1 0 0 30 0 0 0 7 0 0 1 1 1 0 0 31 0 2 1 0 0 0 1 1 8 1 0 0 1 2 1 0 32 0 1 1 0 0 0 2 1 9 1 0 2 1 2 0 0 33 0 1 0 2 1 1 2 1 10 1 0 1 2 2 1 1 34 1 0 0 1 1 1 1 2 11 1 1 2 1 1 2 0 35 1 0 0 1 1 1 0 0 12 1 1 1 1 0 0 1 13 1 0 1 1 1 0 1 Character 36 1 1 1 2 1 2 0 1 37 0 0 0 1 1 1 0 0 14 0 2 2 1 1 0 0 38 0 0 0 1 1 1 0 0 15 0 1 1 2 2 0 1 2 39 1 2 0 0 0 0 0 0 16 0 1 2 2 2 0 0 40 0 1 0 0 1 0 0 0 17 0 0 - 1 1 1 0 0 41 0 0 0 18 0 2 - 42 1 1 0 1 1 1 1 0 19 0 0 43 0 2 0 0 0 0 1 1 20 1 0 0 44 0 0 0 1 1 1 0 1 21 1 1 0 1 1 0 0 0 45 1 0 0 1 0 1 0 1 22 1 0 - 0 1 0 1 0 46 1 1 1 0 1 1 0 0 23 0 0 0 0 0 0 0 0 47 0 1 1 0 1 1 0 0 24 1 2 0 0 1 1 1 1 48 0 1 0 24. Number of additional cusps of premolars: (0) two; (1) more than two; (2) no additional cusps (unordered character). 25. Premolars: (0) seated parallel to axis of tooth row; (1) seated obliquely. 26. Upper incisors: (0) form a curved line; (1) form a straight line. 27. Second and third (or first) upper incisors: (0) third larger than second; (1) second larger than third (or first); (2) in­ cisors equal in size (unordered character). Humerus 28. Lesser tubercle: (0) pronounced; (0) not pronounced (polarity opposite that of Berta and Wyss, 1994:52). 29. Trochlear crest: (0) raised arch-like over coronoid fossa; (1) not separated from coronoid fossa by a distinct lip. 30. Lesser tubercle and head: (0) equal in height or tuber­ cle insignificantly higher than head; (1) tubercle very much higher than head. 31. Lesser tubercle: (0) rounded; (1) extended along the bone's axis; (2) oval. 32. Head: (0) mediolaterally compressed; (1) rounded; (2) flattened proximodistally. 33. Deltoid crest: (0) maximal enlargement is in its proxi­ mal part; (1) neither part noticeably enlarged; (2) maximal en­ largement is in its middle part. 34. Deltoid crest: (0) shorter than one-half length of the bone, confined to the proximal half of the bone; (1) longer than one-half length of the bone but not reaching coronoid fossa; (2) reaches coronoid fossa (in contrast to that of Berta and Wyss, 1994:52). 35. Coronoid fossa: (0) deep; (1) shallow. 36. Head and trochlea: (0) head wider than trochlea; (1) head almost equal in width to trochlea; (2) trochlea wider than head (polarity opposite that of Berta and Wyss, 1994:53). Femur 37. Lesser trochanter: (0) present; (1) absent (Berta and Wyss, 1994:54). 38. Condyles: (0) different in size; (1) similar in size. 39. Epiphyses: (0) distal epiphysis wider than proximal by one-fourth to one-fifth; (1) widths of proximal and distal epiphy­ ses about equal; (2) proximal epiphysis wider than distal one. 40. Shaft: (0) minimum width less than or about equal to two-thirds width of proximal epiphysis; (1) minimum width more than two-thirds width of proximal epiphysis. 41. Intertrochanteric crest: (0) well developed; (1) absent or poorly developed. 42. Intertrochanteric crest: (0) reaches lower than head; (1) short, ends on same level as distal edge of head or fovea capitis. 43. Head: (0) rounded; (1) flattened in proximodistal di­ rection; (2) compressed in mediolateral direction. 44. Intercondylar area: (0) narrow, deep; (1) wide, flat­ tened. 45. Greater trochanter: (0) maximum width in its middle part; (1) maximum width in its proximal part (Koretsky, 1987: 75). 46. Head and greater trochanter: (0) same height; (1) greater trochanter higher than head. 47. Neck: (0) long, slender; (1) short, wide. 48. Shaft: (0) minimum width in its proximal part; (1) minimum width in its middle part. RESULTS The analysis of these taxa using Hennig86 and the 48 un­ weighted characters shown above produced a single tree, 45 steps long, with consistency index of 0.84 and retention index NUMBER 93 159 of 0.83 (Figure 5). The matrix of character-state data for six species of fossil and living monachine seals is given in Table 4, together with data on two outgroup taxa. Leptophoca lenis is representative of the oldest known Phocinae (Repenning and Ray, 1977), and Cystophora cristata represents the Cystophori­ nae (as mentioned above). The nodes of the tree corresponding to taxa recognized by us are indicated; only one new name, Pontophoca sarmatica, has been added to those previously recognized: inclusion of Ponto­ phoca sarmatica within the Monachinae supports the recogni­ tion of the monotypic genus Pontophoca. The nodes of the cla­ dogram shown in Figure 5 are supported by the following character transformations: Node 1. (Subfamily Monachinae) one branch forms the pos­ sibly paraphyletic subfamily: 3(2); 18(1); 32(1.2). Node 2. 41(1) (intertrochanteric crest of the femur poorly developed). Node 3. 43(1) (head of the femur flattened in dorsoventral direction). Node 4. (Genus Monachus): 7(1); 16(2); 17(1); 37(1); 38(1). Also, character 10(2) is homoplasious in M. tropicalis and M. schauinslandi; character 35(1) shares the homoplasy of L. le­ nis. Node 5. 6(1); 9(2); 36(2). Also, character 45(1) shares the homoplasy of Callophoca obscura and Leptophoca lenis. As can be seen from the tree (Figure 5), the assignment of Pontophoca to the Monachinae is confirmed. Thus, the charac­ ters and their polarities that we used on every level of taxonomy are working. Moreover, Pontophoca sarmatica (consisting of a new description of the mandible and humerus) is a closely re­ lated sister group to Monachus and other Monachinea. In contrast to Berta and Wyss (1994:43), we treated Mona­ chus schauinslandi as a not-so-close sister taxon (Flynn, 1988). At the same time, we agree with their conclusion that the sub­ family Monachinae is monophyletic (tribe Lobodontini not considered). In this subfamily we include just the three Ho­ locene species of Monachus, as well as two Pliocene fossil gen­ era (Callophoca and Pliophoca) and one middle Miocene ge­ nus (Pontophoca). Conclusions The family Phocidae separated from Carnivora, probably in the early Oligocene (Davies, 1958), became widely distributed during the middle and (especially) late Miocene, and practi­ cally ceased to exist in the European part of the former USSR (Black Sea region) in the early Pliocene (Chapskii, 1955, 1970, 1971,1975; Repenning et al., 1979). Shared derived features of the morphology of the skull, and the relatively early geological age of these Miocene seals, allow the conclusion that these ani­ mals are broadly ancestral forms of presently living true seals. A direct relationship between known Miocene seals and mod­ em genera is very uncertain, however. Probably, together with other Phocidae, monachine seals are descendants of some Oli­ gocene or even Eocene semiaquatic mustelid. It should be - © Cystophora cristata " © Pontophoca sarmatica 5) Pliophoca etrusca 6) Callophoca obscura \ D Monachus schauinslandi Monachus monachus 4) Monachus tropicalis FIGURE 5.—Single most-parsimonious tree of monachine taxa and two out­ groups, generated by Hennig86 using 48 unweighted characters. Tree length, 45 steps; consistency index, 0.84; retention index, 0.83. pointed out, however, that the only Oligocene representatives of this family yet found are two fragments of femora from the late Oligocene of South Carolina (see Koretsky and Sanders, 2002). The most ancient fossil phocids known from good, in­ formative material are from the early middle Miocene (Holec et al., 1987; Koretsky and Holec, 2002). By that time these ani­ mals were fully recognizable members of subfamilies to which modern pinnipeds belong. Middle Miocene seals did not differ very much from modern species, and in most morphological characteristics they are not especially similar to any of the terrestrial or semiaquatic car- nivorans that might have been the ancestors of phocids. The analysis of the dentition allows us to conclude that monachines have long been separated from the common stem of pinnipeds and formed a separate phylogenetic branch that has existed un­ til now. Clearly, the geographical and geological distributions of taxa are of considerable interest for biostratigraphy and for correla­ tions of middle Miocene to early Pliocene marine deposits of Eurasia. At present, however, in view of insufficient investiga­ tion of Western European and Asiatic materials, these findings may be used mainly for more precise control of the geological age of the true seals in the European part of the former USSR, and of the stratigraphic distribution of Monachinae in the mid­ dle Sarmatian to Pontian of this region. Moreover, the majority of problems of systematics and mor­ phology of Monachinae are not solved completely (Barnes et al., 1985). In this respect the present study is of a preliminary character that opens up new perspectives in investigations of the groups of predators analyzed. We hope that the results pre­ sented herein will draw the specialist's attention and will allow the investigation, from new and different viewpoints, of many problems of classification of both ancient and modern repre­ sentatives of the subfamily Monachinae. 160 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY Literature Cited Alekseev, A.K. 1924. [Seals in the Sarmatian Deposits of Southern USSR.] Journal Nauchno-Issledovatelskich Kafedr v Odesse, 1(10-11):201—205. [In Russian.] 1926. [Seals in the Sarmatian Deposits of the South of the USSR.] Journal Nauchno-Issledovatelskich Kafedr v Odesse, 11 (4): 138—143. [In Russian.] Anbinder, E.M. 1980. Karyologia e Evolucia Laslonogich [Karyology and Evolution of the Pinnipeds]. 452 pages. Moscow: Nauka. [In Russian.] Andreescu, I., and D. Murariu 1985. Contribution a l'etude du crane de quelques representants des fami­ lies des Otariidae et Phocidae (Mammalia: Pinnipedia). Travaux du Museum d'Histoire Naturelle "Grigore Antipa" (Bucharest), 27: 317-324. Andrusov, N.I. 1893. Geotektonika Kerchenskogo poluostrova [Geotectonics of the Kerch Peninsula]. In Materialy dlia Geologii Rossii [Material on the Geology of Russia], 16:91-113. [In Russian.] Antoniuk, A.A. 1970. [Comparative Morphological Research on the Vertebrae of Pinnipe­ dia.] Trudy Tichookeanskogo Instituta Rybnogo Khoziaislva (TINRO), 17:87-95. [In Russian.] 1972. [The Morphology of the Vertebral Column of Genus Pusa.] Fifth All-Union Conference on Marine Mammals, Machatchkala, 2:24-26. [In Russian.] Antoniuk, A.A., and I.A. Koretskaya [Koretsky] 1984. [A New Seal from the Middle Sarmatian Deposits of the Crimean District of the Ukraine.] Vestnik Zoologii, 4:26-31. [In Russian.] Aslanova, S.M. 1965. [Seal in Lower Miocene Deposits in Azerbaijan.] Doklady Akademii Nauk, Azerbaizhanskoy SSR, 21(6)46^18, 4 figures. [In Russian.] Astanin, L.P. 1936. [Ontogenetic Dimorphism of the Skeleton of Mammals in Connec­ tion with the Question of Age Determination.] Izvestia Nauchnogo Instituta, 19(2):203-231. [In Russian.] Barnes, L.G., D.P. Domning, and CE. Ray 1985. Status of Studies on Fossil Marine Mammals. Marine Mammal Sci­ ence, 1(1): 15-53. Berta, A., and A.R. Wyss 1994. Pinniped Phylogeny. In A. Berta and T.A. Demere, editors, Contri­ butions in Marine Mammal Paleontology Honoring Frank C. Whit­ more, Jr. Proceedings of the San Diego Society of Natural History, 29:33-57, 6 figures, 3 tables, 2 appendices. Burns, J.J., and F.H. Fay 1970. Comparative Morphology of the Skull of the Ribbon Seal, Histrio- phoca Fasciala. with Remarks on Systematics of Phocidae. Journal of Zoology (London), 161 (3):363—394, 3 figures, 4 plates. Chapskii, K.K. 1952. [Age-Sexual Dimorphism of the Cranial Features and Its Effect on the Diagnoses of Some Pinnipedia.] Izvestia Instituta Eslestvenich Nauk, Moscow, 25:78-96. [In Russian.] 1955. [Contribution to the Problem of the History of Development of the Caspian and Baikal Seals.] Trudy Zoologicheskogo Instituta Aka­ demii Nauk SSSR, 17:200-216. [In Russian. English translation by Fisheries Research Board of Canada, Translation Series, 174, 26 pages, 1958.] 1961. [Current Status and Problems in the Systematics of Pinnipeds.] Trudy Soveschanii Ikhliologicheskoi Komissii, Akademii Nauk SSSR (Moscow), 12:138-149. [In Russian. English translation by Sayer, U.S. Fish and Wildlife Service, Seattle, Washington.] 1967. [Morphology-Taxonomic Structure of Phoca largha of the Bering Sea.] In Research in Marine Mammals. Trudy Poliarnogo Nauchno- Issledovatelskogo Instituta Rybnogo Khoziaistva e Okeanography (PINRO), 21:147-177. [In Russian.] 1970. [The Concept of the Arctic Origin of Pinnipeds and Another Solu­ tion to This Problem.] In A.I. Tolmachev, editor, Severnyi Ledovityi Okean i ego poberezh'e v Kainozoe [Northern Arctic Ocean and Its Coast in the Cenozoic], pages 166-173. Leningrad: Hydrometeoro- logical Publishers. [In Russian.] 1971. [The Systematic Rank and Suprageneric Differentiation of the Eight-Incisor Seals (Monachinae).] Trudy Atlantic Nauchno-Issle- dovatelskogo Instituta Rybnogo Khoziaistva e Okeanography (AT- LANTNIRO) (Kaliningrad), 39:305-316. [In Russian. English translation, Fisheries Research Board of Canada, Translation Series, 3185, 1974). 1974. [In Defense of Classical Taxonomy of the Seals of the Family Pho­ cidae.] In [Theoretical Questions on the Systematics and Phylogeny of Animals]. Trudy Zoologicheskogo Instituta Akademii Nauk SSSR, 53:282-334. [In Russian.] 1975. Taxonomy of the Seals of the Genus Phoca Sensu Stricto. In Marine Mammals. Materials for the Sixth All-Union Conference, Naukova Dumka.Kiev. 2:159-162. Chepalyga, A.L., E.L. Korotkevich, and YA. Semenov 1985. Paratethys Regional Stages and Hipparion Faunas According to Pa- leomagnetic Data. In Abstracts of the Seventh Congress of the Re­ gional Committee on Mediterranean Neogene Stratigraphy, Buda­ pest, pages 137-139. Davies, J.L. 1958. The Pinnipedia: An Essay in Zoogeography. Geographical Review, 48:474-493. Dornesco, G.-T., and G.-V. Marcoci 1958. Etude comparative du crane des pinnipedes. Annates des Sciences Naturelles, Zoologie, 11: xx + 158-182, 9 figures, 2 tables. Driesch, A. von den 1976. A Guide to the Measurement of Animal Bones from Archeological Sites. Bulletin of the Peabody Museum of Archaeology and Ethnol­ ogy, Harvard University, 1:1-137. Dubrovo, I.A., and K.V. Kapelist 1979. Katalog mestonachozdeniy tretichnuch pozvonochnuch Ykrainu [Catalog of the Localities of the Tertiary Vertebrates of the Ukrai­ nian SSR]. 159 pages. Moscow: Nauka. [In Russian.] Eichwald, E. 1850. [Paleontology of Russia.] 520 pages, atlas, 13 plates. Saint Peters­ burg: Edward Pratz. [In Russian.] Farris, J.S. 1988. Hennig86, Version 1.5. New York: Port Jefferson Station. Fleming, J. 1822. Philosophy of Zoology, or a General View of the Structure, Func­ tion, and Classification of Animals. Volume 2, 187 pages. Edin­ burgh. Flynn, J.J. 1988. Ancestry of Sea Mammals. Nature, 334:383-384. Friant, M. 1947. Recherches sur la femur des Phocidae. Bulletin du Musee Royal d'Histoire Naturelle de Belgique, 23(2): 1—51, 4 plates. Gervais, P., and M. de Serres 1847. Sur les mammiferes fossiles des sables marins tertiaires de Montpel­ lier. Comptes Rendus des Seances de I'Academie des Sciences (Paris), 28(8):799-801. Gill, T.N. 1872. Arrangement of the Families of Mammals, with Analytical Tables. Smithsonian Miscellaneous Collections, 11(1): vi + 98 pages. NUMBER 93 161 Gistel, J. 1848. Naturgeschichte des Tierreichs. xvi + 216 pages, 32 leaves of color plates. Stuttgart: Scheitlin & Krais. Godina, A.Y. 1973. Pozvonochnue neogenovuch otlozheniy na terrilorii Moldavii [Neo­ gene Vertebrate Localities in the Territory of Moldavia]. 107 pages. Kishinev: "Shtiintsa." [In Russian.] Grasse, P.-P. 1955. Mammiferes: Les Ordres—Anatomie, Ethologie, Systematique. In Trade de Zoologie: Anatomie, Systematique, Biologie, 17:568-668. Paris: Masson. Gray, J.E. 1825. Outline of an Attempt at the Disposition of the Mammalia into Tribes and Families with a List of the Genera Apparently Pertaining to Each Tribe. Annals of Philosophy, new series, 10:337-344. 1844-1875. The Zoology of the Voyage of H.M.S. Erebus & Terror, under the Command of Captain Sir James Clark Ross, R.N., F.R.S., during the Years 1839 to 1843, Part 1: Mammalia. 12 pages, 17 plates. London: Longman, Brown, Green, and Longmans. [See Scheffer, 1958, for dates.] Grigorescu, D. 1977 ("1976"). Paratethyan Seals. Systematic Zoology, 25(4)407-419. [Date on title page is 1976; actually published 1977.] Gromova, V.I., I.A. Dubrovo, and N.M. Janowskaya 1962. [Order Pinnipedia: Seals.] In VI. Gromova, editor, Osnovy Paleon- tologii [Fundamentals of Paleontology], 13 [Mammals]: pages 234-235. Moscow. [In Russian. English translation by Israel Pro­ gram for Scientific Translations, Jerusalem, 1968] Heptner, V.G, K.K. Chapskii, B.A. Arsen'ev, and V.E. Sokolov 1996. Pinnipeds and Toothed Whales. In V.G. Heptner and N.P. Naumov, editors, Mammals of the Soviet Union, 2(3): xxx + 995 pages. Wash­ ington, D.C. [Originally published in Russian as Mlekopitayushchie Sovetskogo Soyuza. Moscow: Vysshaya Shkola, 1976.] Holec, P., J. Klembara, and S. Meszaros 1987. Discovery of New Fauna of Marine and Terrestrial Vertebrates in Devinska Nova Ves. Geologicky Zbornik, Geologica Carpathica (Bratislava), 38(3):349-356. Kellogg, R. 1922. Pinnipeds from Miocene and Pleistocene Deposits of California. University of California Publications. Bulletin of the Department of Geological Sciences. 13(4):23—132, 19 figures, folding table. King, J. 1956. The Monk Seals (Genus Monachus). Bulletin of the British Museum (Natural History), Zoology, 3:201-256, 5 plates. 1964. Seals of the World. 154 pages, 17 figures, 14 plates, frontispiece, 30 maps. London: Trustees of the British Museum (Natural History). 1966. Relationships of the Hooded and Elephant Seals (Genera Cysto­ phora and Mirounga). Journal of Zoology (London), 148(4): 385-398, 7 figures. 1983. Seals of the World. Second edition, 240 pages. London: British Mu­ seum (Natural History). Kirpichnikov, A.A. 1961. [Sketch of the History of Study of Marine Mammals from Sarmatian Localities in the USSR and Neighboring Countries.] Trudy Ikhtio- logicheskoi Komissii Akademii Nauk Ukrainskoi SSR, 12:25-39. [In Russian.] Koretskaya [Koretsky], I.A. 1987. [Sexual Dimorphism in the Structure of the Humerus and Femur of Monachopsis pontica (Pinnipedia: Phocinae).] Vestnik Zoologii, 4:77-82. [In Russian.] Koretsky, I.A., and P. Holec 2002. A Primitive Seal (Mammalia: Phocidae) from the Early Middle Mi­ ocene of Central Paratethys. In R. Emry, editor, Cenozoic Mammals of Land and Sea: Tributes to the Career of Clayton E. Ray. Smithso­ nian Contributions to Paleobiology, 93:163-178. Koretsky, I.A., and C E . Ray 1994. Cryptophoca, New Genus for Phoca Maeotica (Mammalia: Pinni­ pedia: Phocinae) from Upper Miocene Deposits in the Northern Black Sea Region. Proceedings of the Biological Society of Wash­ ington, 107(1): 17-26, 4 figures, 3 tables. In prep. Phocidae of the Pliocene of Eastern North America. Smithsonian Contributions to Paleobiology. Koretsky, I.A., and A.E. Sanders 2002. Paleontology of the Late Oligocene Ashley and Chandler Bridge Formations of South Carolina, 1: Paleogene Pinniped Remains; The Oldest Known Seal (Carnivora: Phocidae). In R. Emry, editor, Cen­ ozoic Mammals of Land and Sea: Tributes to the Career of Clayton E. Ray. Smithsonian Contributions to Paleobiology, 93:179-183. Korotkevich, E.L., V.N. Kushniruk, Y.A. Semenov, and A.P. Chepalyga 1985. [New Localities of Middle Sarmatian Vertebrates in the Ukraine.] Vestnik Zoologii, 3:81-82. [In Russian.] Kretzoi, M. 1941. Seehund-Reste aus dem Sarmat von Erd bei Budapest. Foldtani Koz­ lony, 71(7-12):350-356. Macarovici, N. 1942. Sur les phoques fossiles du Sarmatien de I'Europe. Annates Scienti- fiques de I'Universite de Jassy, section 2, 28(10)260—268. Macarovici, N., and C.V. Oescu 1942. Quelques vertebres fossiles trouves dans les calcaires recifales de Chisinau (Bessarabia). Analele Academiei Romdne, Memoriile Sect iunii Stiin(ifice, series 3, 17(6):351—382, 7 plates. Marcoci, G.V, and M. Popa 1957. Contribution a la description du crane de Monachus monachus Herm., le phoque de la mer Noire. Revue de Biologie, Academie de la Republique Populaire Roumaine, 2(2):277-295. McLaren, I.A. 1960. On the Origin of the Caspian and Baikal Seals and Their Paleocli- matological Implications. American Journal of Science, 258:47-65. Michaylov, S.S. 1980. Mezdynarodnaya anatomicheskaya nomenklatura [International Anatomical Nomenclature]. Fourth edition, 239 pages. Moscow: Medicinskaya Literature [Medical Literature]. [In Russian.] Muizon, C. de 1981. Les vertebres fossiles de la formation Pisco (Perou), Premiere par- tie: Deux nouveaux Monachinae (Phocidae: Mammalia) du Pliocene de Sud-Sacaco. Travaux de I'Institut Francais d'Etudes Andines, 22: 161 pages. 1982. Phocid Phylogeny and Dispersal. Annals of the South African Mu­ seum (Cape Town), 89(2): 175-213, 9 figures, 1 table. 1992. Palaontologie. In Raymond Duguy and Daniel Robineau, editors, Meeressauger: Robbe^Pinnipedia. In Jochen Niethammer and Franz Krapp, editors. Handbuch der Saugetiere Europas, 6(2): 34-41. Wiesbaden: AULA-Verlag. Nordmann, A.D. von 1860 [1858], Palaeontologie Sudrusslands. 360 pages, atlas of 24 plates. Helsingfors: H.C. Friic. [Plates published 1858. Text to accompany plates published I860.] Ognev, S.I. 1935. [Fissipedia and Pinnipedia.] In [Mammals of the USSR and Adja­ cent Countries], 3: 752 pages. Moscow, Leningrad: Glavpushnina. [In Russian. English translation by A. Birron and Z.S. Coles for Is­ rael Program for Scientific Translations, Washington, D.C, 1962.] Pavlinov, I.J., and O.L. Rossolimo 1987. [Systematics of Mammalia of the Soviet Union.] 282 pages. Mos­ cow: Moscow University. [In Russian.] Pidoplichko, I.G. 1956. [Materials for the Study of the Faunas of the Ukrainian SSR.] Sec­ ond edition, 225 pages. Kiev: Naukova Dumka. [In Ukrainian.] Pierard, J. 1971. Osteology and Myology of the Weddell Seal (Leptophoca weddelli 162 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY (Lesson, 1826). In W.H. Burt, editor, Antarctic Pinnipedia. Antarctic Research Series, 18:53-108. Washington, D.C: National Academy of Sciences, National Research Council. Ray, C E . 1977 ("1976"). Geography of Phocid Evolution. Systematic Zoology, 25(4): 391-406. [Date on title page is 1976; actually published 1977.] Repenning, C.A., and C E . Ray 1977. The Origin of the Hawaiian Monk Seal. Proceedings of the Biologi­ cal Society of Washington, 89(58):667-688. Repenning, C.A., C E . Ray, and D. Grigorescu 1979. Pinniped Biogeography. In J. Gray and A.J. Boucot, editors. Histor­ ical Biogeography, Plate Tectonics, and the Changing Environment: Proceedings of the 37 th Annual Biology Colloquium and Selected Papers, pages 357-369. Corvallis: Oregon State University Press. Robinette, H., and H.J. Stains 1970. Comparative Study of the Calcanea of the Pinnipedia. Journal of Mammalogy, 51(3): 527 pages. Savage, D.E., and D.E. Russell 1983. Mammalian Paleofaunas of the World. 432 pages. Reading, Massa­ chusetts: Addison-Wesley. Scheffer, V.B. 1958. Seals, Sea Lions and Walruses: A Review of the Pinnipedia. 179 pages, 32 plates, 15 figures. Stanford, California: Stanford Univer­ sity Press. Semenenko, V.N. 1987. Sytratigraficheskaya korrelacia verchnego Miocena i Pliocena, Vos- tochnuy Paratetis i Tetis [Stratigraphical Correlation of Upper Mi­ ocene and Pliocene, Eastern Paratethys and Tethys]. 230 pages. Kiev: Naukova Dumka. [In Russian.] Semenov, Y.A. 1981. [On the Question of the Systematic Situation of the Genus Pal- hyaena Gervais, 1859 (Carnivora, Hyaenida).] In [Ecological-Mor­ phological Specialization of the Animals and Their Environmental Habitats], pages 58-61. Sbornik Nauchnykh Trudov. Kiev: Naukova Dumka. [In Russian.] Sergienko, N.I. 1967. [On the Taxonomic Significance of Some Parts of the Postcranial Skeleton of the Genus Pusa.] Trudy Poliarnogo Institute Morskogo Rybnogo Khoziaistva e Okeanography (PINRO) (Murmansk), 21:185-193. [In Russian.] Simionescu, I. 1925. Foci fosile din Sarmatecul de la Chisinau. Academia Romana Memoriile, SecUunii Stiinfifice, 3(3)4:179-191, 4 plates. Simpson, G.G. 1945. The Principles of Classification and a Classification of Mammals. Bulletin of the American Museum of Natural History, 85: 350 pages. Sinelnikov, R.D. 1963. Atlas analomii cheloveka [Atlas of the Anatomy of Man]. Second edition, volume 1, 477 pages. Moscow: Medicinskaya Literatura [Medical Literature]. [In Russian.] Sokolov, V.E. 1979. [Cetacea, Pinnipedia, Carnivora, Tubulidentata, Tylopoda, Perisso- dactyla]. In Sistematika mlekonitauzchich [Systematics of Mam­ mals], 3: 528 pages. Moscow: "Vischaya Schkola" [High School], [In Russian.] Tavani, G. 1941. Revisione dei resti del pinnipede conservato nel Museo di Geologia di Pisa. Paleontographica Italica, Memorie di Paleontologia, 40: 97-113, 15 plates, 23 figures. [Volume 40 dated 1942.] Tedford, R.H. 1977 ("1976"). Relationship of Pinnipeds to Other Carnivores (Mamma­ lia). Systematic Zoology, 25(4):363-374. [Date on title page is 1976; actually published 1977.] Thenius, E. 1950. Ergebnisse der Neuuntersuchung von Miophoca vetusta (Zapfe) (Phocidae, Mammal.) aus dem Torton des Wiener Beckens. An- zeiger der Osterreich Akademie der Wissenchaften, Vienna, Mathe- matisch-Naturwissenschaftliche Klasse. 5:1-9. 1952. Die Saugetierfauna aus dem Torton von Neudorf an der March (CSR). Neues Jahrbuch fiir Geologie und Paldontologie, Abhand­ lungen (Stuttgart), 96(1)27-136, 70 figures. Trelea, N., and T. Simionescu 1985. Au sujet de quelques formes de vertebres des formations sarmati- ennes de Schelia du departement Jassy. Analele Stiinnfice ale Uni- versitatii Jasi, Geologie-Geografic, 31:18-20. Trouessart, E.-L. 1897. Carnivora, Pinnipedia, Rodentia I. In Catalogus mammalium tarn viventium quam fossilium. New edition, volume 2, pages 219-452. Berlin: R. Friedlander and Son. Van Beneden, P.-J. 1877. Description des ossements fossiles des environs d'Anvers, premiere partie: Pinnipedes ou Amphitheriens. Annates du Musee Royal d'Histoire Naturelle de Belgique, 1: 88 pages, 18 plates, 17 figures, 9 maps. Wozencraft, W.C. 1989. The Phylogeny of the Recent Carnivora. In John L. Gittleman, edi­ tor, Carnivore Behavior, Ecology, and Evolution, pages 495-593. Ithaca, New York: Cornell University Press. Wyss, A.R. 1988. On "Retrogression" in the Evolution of the Phocinae and Phyloge­ netic Affinities of the Monk Seals. American Museum Novitates, 2924: 38 pages. Zapfe, H. 1937. Ein bemerkenswerter Phocidenfund aus dem Torton des Wiener- Beckens. Verhandlungen der K.K. Zoologisch-Botanischen Gesell- schaft (Vienna), 86/87:271-276, 2 figures. A Primitive Seal (Mammalia: Phocidae) from the Early Middle Miocene of Central Paratethys Irina A. Koretsky and Peter Holec A B S T R A C T A well-preserved skull from the early middle Miocene (approxi­ mately 15 Ma) at Devinska Nova Ves (formerly Neudorf an der March), Slovakia, herein named Devinophoca claytoni, new genus and new species, is morphologically the closest common ancestor of all true seals. It shows a mixture of subfamilial characters. Fea­ tures shared with Phocinae are the number of incisors and the lack of a strongly pronounced mastoid process. Characters similar to Monachinae are the shape of maxillae and the ratio between fron­ tal and maxillary contacts of nasal bones. Characters shared with Cystophorinae are the ratio between interorbital width and mastoid width and (also shared with Phocinae) the ratio between length of auditory bullae and distance between them. Moreover, this skull has primitive features that are not known in any of the three sub­ families: Ml is triangular, with three cusps and three roots; in P2-P4 the larger posterior roots are clearly made up of two fused roots; the incisors form a curved line; the anterior palatal foramina are deep and oval; and the sagittal crest is very well developed. The traditional separation of the family Phocidae into the sub­ families Phocinae, Monachinae, and Cystophorinae has been intensively debated during the past 40 years and debate continues today, but we herein follow the traditional classification. The ple­ siomorphic D. claytoni is thus considered a sister taxon to the three extant subfamilies of Phocidae and is referred to a new subfamily, Devinophocinae. Because of its late age, D. claytoni cannot be ancestral to the more advanced phocids. Its primitive characters in combination with the characters it shares with the other subfami­ lies suggest, however, that it might approximate the common ancestral morphotype. Introduction Around Devinska Kobyla Hill near Bratislava, Slovakia (Figure 1), fossil mammals are found at eight localities. A brief Irina A. Koretsky, Research Associate, Department of Paleobiology, National Museum of Natural History, Smithsonian Institution, Wash­ ington, D.C. 20560-0121, United States. Peter Holec, Department of Geology and Paleontology, Faculty of Natural Sciences, Comenius University, Mlynskd Dolina, 84215 Bratislava, Slovakia. description and a list of all vertebrate species from these locali­ ties were given by Holec and Sabol (1996). These assemblages contain a mixture of marine and terrestrial vertebrates. Fossil seals are found at three of the localities at Devinska Nova Ves: Sandberg, Bonanza (a newly discovered fissure-filling at the quarry of the former Stokerau limekiln), and Wait's Quarry. The first locality discovered was Sandberg, the history of which was discussed in detail by Thenius (1952). The original independent township Devinska Nova Ves (Neudorf an der March in the earlier literature) is now a suburb of Bratislava. In the quarry of the former Stokerau limekiln, two localities with mammalian faunas are found. The first is Zapfe's classical locality, consisting of two fissures in the western part of the quarry filled with terra fusca with many fossil bones of terres­ trial mammals; marine mammals have never been found there. At present, only the more western of the two fissures can be seen. At the eastern end of the limestone quarry a new, second locality, called Bonanza, was discovered more recently; there, abundant fossil bones of terrestrial and marine vertebrates (fishes and seals) are mixed in the same deposit. The knowledge of the mammalian assemblage found in the two western fissures at the Stokerau limekiln is the result of the efforts of Helmuth Zapfe. His father, LB. Zapfe, saved the fos­ sils that were discovered in the fissure-filling during exploita­ tion of the limestones in 1943. From the large-mammal fauna, Zapfe (1953) determined the age of the filling to be "Hel­ vetian." Fejfar (1974) also carried out research there, focused mainly on the small mammals. He identified five species of in­ sectivores and fifteen species of rodents, on the basis of which he determined the age of the assemblage to be early Badenian. Some species of rodents were mentioned also by Holec (1986). The Bonanza locality was discovered in 1984 by the amateur paleontologist S. Meszaros. Its vertebrate fauna was described by Holec et al. (1987). In addition to the mammals, abundant remains of fishes and frogs also were found there. Remains of seals also were found at Wait's Quarry; a short description of the locality by Holec and Sabol (1996) men­ tioned the occurrence. 163 164 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY FIGURE 1.—Map of Central Europe. Study area indicated by black square. Zapfe's (1937) description of Miophoca vetusta from the Sandberg locality (early middle Miocene; Badenian) was based upon a mandible. He noted that this specimen may be related to Pristiphoca Gervais, 1848-1852, and also is very close to the living Monachinae. Later, Thenius (1950), using Zapfe's hy­ pothesis, transferred this material to the genus Pristiphoca. As explained below, we do not accept this transfer because this ge­ nus is a nomen dubium. A preliminary description of the pinniped skull described herein has already been published by Holec with a list of the other fossil mammals from the early Miocene of Slovakia (Holec et al., 1987). When Koretsky visited the Slovak Na­ tional Museum in Bratislava in 1995, she recognized the un­ usual morphology of a this skull and its importance to phocid evolution. In this paper, we accept a traditional classification (Simpson, 1945; Chapskii, 1955, 1974; Scheffer, 1958; King, 1966; Hept­ ner et al., 1976) in which the family Phocidae is divided into three subfamilies: Phocinae, Monachinae, and Cystophorinae. For our morphometric analysis of the skull, we used the methods of Dornesco and Marcoci (1958), Pierard (1971), von den Driesch (1976), and Trelea and Simionescu (1985). Osteo­ logical terminology in this paper follows the Illustrated Veteri­ nary Anatomical Nomenclature (Schaller, 1992). ACKNOWLEDGMENTS.—We thank Clayton E. Ray and Rob­ ert J. Emry, National Museum of Natural History (NMNH), Smithsonian Institution, and Daryl P. Domning, Howard Uni­ versity, for their critical reviews and discussion of several drafts of this work; Steven J. Jabo and Victor Krantz, NMNH, for the photographs presented in this paper; Jennifer Emry for illustrations; Peter Kroehler, NMNH, for providing casts for study; James G. Mead and Charles W. Potter, NMNH, for per­ mission to study the modern pinnipeds in the National Museum collections; and R. Ewan Fordyce, Department of Geology, University of Otago, Dunedin, New Zealand, for his help with the cladistic analysis. The authors express their sincere thanks to I. Iablonsky for critical reading of the manuscript and for en­ couraging remarks; to V Matlakova for the geological map; and to Anna Durisova and Branislav Matousek, both from the Slovak National Museum of Natural History, Bratislava, and Michael Kova, Department of Geology and Paleontology, Fac­ ulty of Sciences, Comenius University, Bratislava, for their co­ operation and permission to study material under their care. We would like to thank two anonymous reviewers and Lawrence G. Barnes from the Natural History Museum of Los Angeles County for constructive critical reviews of the manu­ script. Most of the financial support for the travel and research for this project was provided by the Remington and Marguerite Kellogg Fund of the Smithsonian Institution. GEOLOGY Rising to 514 m above sea level, Mt. Devinska Kobyla is the highest peak in the Devinske Karpaty Mountains, which consti­ tute the southernmost tip of the Male Karpaty Mountains in NUMBER 93 165 Slovakia. Devinska Kobyla lies near the southern edge of the mountains, close to the Vienna Basin. The Devinske Karpaty range is bounded by the Danube River on the south and south­ east, the Morava River on the west, Diibravsky Potok creek on the north, and the Lamae Fault, which separates these moun­ tains from the rest of the Male Karpaty, on the north-northeast. The whole area lies in the territory of greater Bratislava City, which includes once-separate villages such as Devinska Nova Ves, near which are several important fossil vertebrate locali­ ties, and the village of Devin (Figures 1, 2). The Devinska Kobyla area was emergent at the beginning of the late Miocene, and terrestrial sediments being produced by erosion were sometimes trapped and preserved in fissures and caverns in the subjacent limestones. This is exemplified by loamy fillings in a quarry of the former Stockerau lime plant. The early middle Miocene (early Badenian, 16.0-13.6 Ma) fill­ ings contain many bones of various vertebrates, described mainly by Zapfe between 1949 and 1983. The terrestrial loam in these fissures is overlain by transgressive sandy marine sedi­ ments of late Badenian age. Late Badenian sediments occur (along with early and middle Badenian sediments) in drillhole DNV-1 (Bonanza locality), as well as in outcrops. The lithostratigraphic unit, Sandberg Sands, had been defined there by Barath et al. (1994). The base of this formation consists of gravels with pebbles of variable size intercalated with sands. The sands reflect a transgressive stage of deposition on the eastern edge of the Vienna Basin. Calcareous clays and friable siltstones, which account for the entire late Badenian in the basin are found in a brick plant in Devinska Nova Ves. These sediments are typically rich in mi­ cro- and macrofauna and include terrestrial flora (Holec, 1987; Holec and Sabol, 1997). The late Badenian marine transgression reached the slopes of Devinska Kobyla, whose summit at that time protruded as an island or peninsula. Littoral deposits (carbonate breccias) of late Badenian age are found in numerous outcrops: sands, sand­ stones, algal limestones, and carbonate breccia. Coarse breccia and conglomerates, with clasts as much as 1 m in size, are ex­ posed in the former Wait's Quarry on the western slope of Devinska Kobyla, where they were deposited on the inundated ridge (Andrusov, 1969). The signs of a seacoast also are very clearly visible at the renowned Sandberg locality, where almost 300 species of fossils including marine algae, invertebrates, marine vertebrates, and terrestrial mammals have been found (Holec, 1985; Holec and Sabol, 1996; Svagrovsky, 1981). The presence of red algal sandy limestones at Sandberg proves that the sea here was shallow during the time of late Badenian dep­ osition. The whole Devinska Kobyla area is rich in fossil remains of invertebrates and vertebrates alike. The three Badenian locali­ ties yielding fossil seals (Figure 2) are (1) Wait's Quarry, an abandoned quarry on the western slope of Devinska Kobyla, (2) Sandberg, a sand pit located about 700 m north of Wait's Quarry, and (3) the Bonanza sublocality of the Stockerau limekiln. FIGURE 2.—Geological map of Devinska Kobyla: l=Quaternary, undifferenti­ ated: fluvial, proluvial, and man-made sediments. 2=Sarmatian: Bryozoan-ser- pulid limestones, oolitic sands, calcareous and variegated clays, sands. 3=Stu- dienka Formation, late Badenian: algal limestones, sands with sandstone beds and gravel intercalations, conglomerates dominated by granite pebbles. 4= Devinska Nova Ves, middle Badenian and undifferentiated Badenian: granite conglomerates, carbonate breccias cemented by sinter, conglomerates. 5 = Mesozoic, Albian?: shales; Neocomian: mildly marly, cherry limestones; Dog­ ger-Malm: siliceous limestones and silicites; Lias: brecciated limestones with carbonate extraclasts. 6=Envelope unit, Middle Triassic: massive, locally dolo­ mitic limestones, massive intraclast breccia dolomites interlayered with lime­ stones. 7 = Early Triassic: bedded siliceous limestones and dolomites. 8=Late Devonian-Early Carboniferous: aplite and pegmatite dikes. 9=Pre-Devonian: crystalline schists, primarily metapelites. 10=Fossil occurrences. Localities (circled numbers): Wait's Quarry (1), Sandberg (2), Bonanza (3). The former Stockerau lime plant and its quarry are situated on the northern slope of Devinska Kobyla. Dark gray, fairly strongly recrystallized limestone of Jurassic age is exposed there. It is dissected by north-south trending fissures, many of them filled with sinter, terra rossa, and terra fusca. Overlying all of these is late Badenian marine sand. A rich vertebrate fauna was described largely by Zapfe (1937-1983) from two of these fissures in the western part of the Stockerau quarry. Bo­ nanza is a third fissure located nearby on the quarry's eastern edge. Unlike the preceding two fissures, which yielded only terrestrial vertebrates, this one also contains a wealth of re­ mains offish, seals, and marine invertebrates (Holec et al., 166 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY 1987). The holotype skull of Devinophoca claytoni described below was found there. Systematic Paleontology Superfamily PHOCOIDEA Smirnov, 1908 Family PHOCIDAE Gray, 1825 Subfamily DEVINOPHOCINAE, new subfamily DIAGNOSIS.—Upper dental formula 13, CI, P4, Ml (shared with Phocinae). Incisors forming U-shaped arcade; P2-P4 dou­ ble rooted (as in Phocinae and Monachinae, in contrast to Cys­ tophorinae), with posterior root larger than anterior; P4 with two fused posterior roots and with carnassial notch on meta- style blade; Ml triangular with three cusps and three roots (no other phocid has teeth with three cusps and three roots in trian­ gular arrangement); preorbital part of maxilla with very short, pronounced concavity (similar to Monachinae); antorbital pro­ cess well defined; frontal contact of nasal bones much shorter than maxillary contact (shared with Phocinae); interorbital space slightly broader anteriorly than posteriorly; interorbital width less than 25% of width of skull at mastoid processes (as in Cystophorinae); sagittal crest very well developed (more than in Monachinae) and not forming triangle with lambdoidal crests; diameter of infraorbital foramen less than diameter of alveolus of upper canine (as in Monachus schauinslandi, unlike Phocinae and Cystophorinae); anterior palatal foramina oval- shaped and deep (as in Cystophorinae), with well-pronounced palatal groove; anteroposterior length of tympanic bullae greater than distance between them (similar to Phocinae and Cystophorinae); width of mastoid process less than one-half length of tympanic bulla; mastoid convexity not turned down behind mastoid process (as in Phocinae). TYPE AND ONLY INCLUDED GENUS.— Devinophoca, new genus Genus Devinophoca, new genus Pristiphoca.—sensu Holec et al., 1987:350, fig. 3.—[Not Pristiphoca Gervais, 1852:272 (nomen dubium)].—[Not Pristiphoca Thenius, 1950:1-9; 1952: 60-63; 1969:402, figs. 420,421; 1979:171; 1992:73]. TYPE SPECIES.—Devinophoca claytoni, new species. ETYMOLOGY.—From Devin, in reference to Devinska Nova Ves, Slovakia; and phoca (Latin), seal. DIAGNOSIS.—As for the subfamily until other genera are de­ scribed. REMARKS.—According to Koretsky and Ray (in prep.), the name Pristiphoca occitana Gervais and Serres, 1847, in Ger­ vais, 1848-1852 is a nomen dubium because it was based upon an illustration of the canine tooth of an indeterminate carni- Devinophoca claytoni, new species FIGURES 3-8, TABLES 1, 2 Miophoca vetusta Zapfe, 1937.—Holec et al., 1987:350, fig. 3 [assignment to Pristiphoca]. HOLOTYPE.—Incomplete skull with right Ml and left P2-M1; collected by the amateur paleontologist S. Meszaros in 1984; catalog number Z14523 of the Museum of Natural His­ tory, Slovak National Museum, Bratislava, Slovak Republic. TYPE Locality.—Stokerau lime quarry, Bonanza Hill, Devin­ ska Kobyla, Slovak Republic. ETYMOLOGY.—Named in honor of Clayton E. Ray, in recog­ nition of his numerous contributions to the study of fossil pin­ nipeds. AGE AND DISTRIBUTION.—Badenian, early middle Miocene, 14.8 Ma (Steininger and Nevesskaya, 1975; Steininger et al., 1989; Rogl and Daxner-Hock, 1996); Vienna Basin of Slova­ kia. DIAGNOSIS.—As for the subfamily and genus. DESCRIPTION AND COMPARISON.—The obliterated sutures and heavily worn premolars indicate that the holotype cranium of D. claytoni represents an adult individual. All of the incisors and both canines have fallen out. On the right side of the skull all teeth have fallen out except Ml. The anterior part of the pre­ maxillae around the incisors, the palatine, parts of the palatine processes of the maxilla, and much of the orbital parts of the frontal bones are missing. Part of the basicranium is broken away also, and the vomer, pterygoid, presphenoid, and ba­ sisphenoid bones are missing. Both jugal (=zygomatic) bones are missing, the paroccipital (=jugular) processes are broken away, and the supraoccipital part of the occipital shield is largely absent. The pre- and postorbital parts of the skull are approximately equal in length (Figure 3). The lateral outline of the braincase is rounded. In lateral profile, the top of the skull is slightly con­ cave. The braincase is widest (84.5 mm) above the external au­ ditory meatus. The preorbital parts of the maxillae, between the nasal aperture and the orbits, are short and concave, the same shape as in the Monachinae (Chapskii, 1974). The palatal parts of the premaxilla-maxillary sutures are fused and completely obliterated, but lateral to the nares, near the nasal bones, the su­ tures are clear. The tongue-like ascending process (pnp) of the premaxilla has a short (at least 4.0 mm) contact with the lateral edge of the nasal bones (Figure 4A,B); it cannot be determined with certainty whether this process of the premaxilla is com­ plete. The supraorbital process (sop) of the frontal bone is repre­ sented only by a small tubercle (Figure 3A,B). On the maxilla at the anterior margin of the orbit is a small but distinct antorbital process (aop). The fronto-maxillary suture is slightly forward of the anterior rim of the orbit. NUMBER 93 167 FIGURE 3.—Devinophoca claytoni. new genus and species, holotype skull Z14523: A, dorsal view; scale bar= 0.75 cm; B, outline drawing of the photograph of dorsal view. (Abbreviations for Figures 3-7: aop=antorbital process; Bo=basioccipital; cc=carotid canal; cf=carotid foramen; ch=hypoglossal canal; cma=anterior muscu- lotubular canal; coc=condyloid canal with transverse venous sinus; eam=external acoustic meatus; ec=ectotym- panic; en=entotympanic; fi=incisive foramen; fio=infraorbital foramen; flp=posterior lacerate foramen; fml= median lacerate foramen; fpal=palatine foramen; fpg=postgelenoid foramen; Fr=frontal; fsm=stylomastoid foramen; g=glenoid fossa; ju=jugal process of squamosal; lc = lambdoidal crest; mp=mastoid process; mt= meatal tube of the bulla; Mx=maxilla; Na=nasal; Oc=occipital; occ=occipital condyle; ocs = supraoccipital crest; Pa=parietal; Pal=palatine; Pmx = premaxilla; pnp=nasal process of the premaxilla; pp=paroccipital pro­ cess; Sq=squamosal; sips=sulcus for inferior petrosal sinus; sop = supraorbital process; spt=transverse palatal suture; tb=tympanic bulla.) Neither jugal bone is present, but the maxillary-jugal suture is complete on the left side; it slopes posterolaterally from the anteroventral edge of the orbit. The nasal bones (Figures 3, 4A) are very short and com­ pletely fused to each other along the midline; their frontal con­ tact is much shorter than their maxillary contact. Posteriorly the nasal bones together form a V-shaped projection about 5 mm long inserted between the frontal bones. The posterior limit of the nasal bones is immediately behind the broad frontal-maxil­ lary contact. The width of the nasal bones is approximately constant for their whole length except for the small pointed frontal process. The nasal opening is circular (Figure 4A,B). The lateral edge of each nasal bone projects farthest anteriorly; the anterior edge of each nasal bone is concave, and together the two nasals form a short median anterior projection. This shape of the anterior border of the nasal bones is more similar to that of the Phocinae (especially Erignathus) than to that of other members of the Phocidae. The ratio between the lengths of the frontal and maxillary contacts of the nasal bones, how­ ever, is more similar to that of Monachinae. The interorbital region is slightly wider anteriorly and nar­ rows posteriorly to where the braincase begins. The least inter­ orbital width occurs in the most posterior portion of the interor­ bital area. Berta and Wyss (1994) noted that this very primitive 168 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY B — FIGURE 4.—Devinophoca claytoni, new genus and species, holotype skull Z14523: A, anterior view; B, outline drawing of the photograph of anterior view; scale bar=0.75 cm; c, posterior view; D, outline drawing of the pho­ tograph of posterior view, scale bar=1.0 cm. For explanation of abbreviations see Figure 3. feature is typical for terrestrial carnivorans. The widest part of the interorbital area (24.0 mm) is about 21% of the width of the skull across the mastoid processes (113.0 mm; Table 1), which is very similar to the ratio in Cystophorinae (including Cysto­ phora and Mirounga). The sagittal crest is much better developed than in other Pho­ cidae; it begins at the narrowest part of the interorbital area, even with the posterior ends of the orbit, and continues to the lambdoidal crest. At its anterior end the sagittal crest divides into two much smaller, indistinct temporal crests that disappear at the dorsal edge of the orbit. The depressions on either side of the sagittal crest become deeper where they meet the lambdoi­ dal crests. The sagittal and lambdoidal crests join at a point, without the triangle seen in the monachines. The maximum height of the sagittal crest is about 2 mm. The infraorbital foramen (Figures 4A,B, 6B) is located above the posterior P4-anterior Ml; it is circular and relatively small. The diameter of the alveolus of the upper canine (13.2 mm) is greater than the diameter of the infraorbital foramen, in con­ trast to Phocinae. When the skull is viewed dorsally, the poste­ rior opening of the infraorbital canal cannot be seen in the or­ bits. The ventral root of the zygomatic process of the maxilla originates posterolateral to Ml, at the same level as the palatal process of the maxilla. The medial wall of the infraorbital canal is continued posteriorly by the vertical interorbital plate of the maxilla, which is very similar to the condition in Enaliarctos mealsi Mitchell and Tedford, 1973. The palatine process of the maxilla (Figure 6) is a flat plate. The anterior palatal foramina (=fissurae palatinae) are located between the canines and are oval and deep, in contrast to the condition stated by Wozencraft (1989) for other phocids. Be­ tween the canines the palate is narrow and concave (18.5 mm wide and 11 mm high) and descends abruptly to the posterior margin of the incisor arcade. The lingual alveolar margins of the canines and incisors are at the same level as those of the cheek teeth (Figure 5). From the anterior palatal foramina to the level of P2 the palate is broken; posterior to P2 the palate is flattened transversely. The maxillary-palatine suture (=sutura palatina transversa, spt) lies medial to Ml ; the palatine bones are missing. The posterior palatine foramina (=canali palatine major, fpal) are medial to P4 (Figure 6), anterior to the palatine/ maxilla suture; each is divided into two openings by a tiny bridge. The anterior and posterior palatine foramina are con­ nected by a relatively shallow anteroposteriorly aligned groove (= sulcus palatinus) (Figure 6A). Wyss and Flynn (1993) theo­ rized that this primitive condition characterized phocid ances- NUMBER 93 169 TABLE 1.—Cranial measurements (in mm) in the holotype skull Z14523 of Devinophoca claytoni, new genus and new species. Character 1. Total length 2. Condylobasal length 3. Length of processus palatinus 4. Length of rostral part, measured from antero-upper corner of orbit 5. Length of braincase, measured from anterior corner of orbit 6. Length of bulla tympanica 7. Length of tooth row, Pl to Ml 8. Length of tooth row, P2 to P4 9. Maximum diameter of infraorbital foramen 10. Length of temporal fossa 11. Width of rostrum across canines 12. Maximal infraorbital width 13. Minimal infraorbital width 14. Width of skull across of processus zygomaticus of squamosal 15. Width of braincase 16. Mastoid width 17. Width of processus palatinus between Pis 18. Maximum width of processus palatinus 19. Maximum width of foramen infraorbitale 20. Width of bulla tympanica 21. Width of rostrum 22. Height of skull in region of bulla tympanica 23. Distance from center of stylomastoid foramen to center of postglenoid foramen Ratios of measurements character 14/ character 1 character 4/ character 5 character 18/ character 9 Measurement 119.9 119.3 71.0 49.0 93.5 38.5 49.0 (L), 53.2 (R) 32.5 (L), 34.5 (R) 7.5 (L), 11.2 (R) 61.5 40.0 25.5 14.0 124.0 88.0 113.0 10.5 55.5 9.0 (L), 10.0(R) 49.3 37.0 80.0 15.1 103.4 52.4 160.9 tors. The posterior border of the horizontal plate of the maxilla is sharp edged, turning ventrally about 2 mm behind M1. Pos­ terior to Ml, the ventrolateral border of the maxilla (inside the orbits) is flattened, and a very distinct separation can be de­ fined between the horizontal and vertical surfaces of the pa­ latine bones. The thickness of the maxilla between the posterior alveolus of Ml and the ventral border of the infraorbital canal is about 12.5 mm. The anterior edge of the orbit is above the middle of P4 (Fig­ ure 5). The jugal process (ju) of the squamosal ascends anteri­ orly, as a long tapered process; the length of this process in front of the glenoid fossa is at least 36 mm. The glenoid fossa (g) measures 11 mm anteroposteriorly and 23.5 mm trans­ versely. The postglenoid process itself is located about 4 mm forward of the meatal tube. A postglenoid foramen (fpg) is present in the fossa between the postglenoid process and the meatal tube (Figures 6, 7B). In contrast to the opinion of Wyss and Flynn (1993), we consider the presence of this foramen to be primitive in Phocidae; it is present and large in terrestrial camivorans. Between the meatal tube and the postglenoid fora­ men is a deep, narrow groove in the tympanic bone; this groove is parallel to the meatal tube and floored by the tympanic. The presence or absence of a suprameatal fossa cannot be deter­ mined. Laterally, the bulla (Figure 6) is extended as a long tube, with a prominent ventral lip forming the ventral margin of the exter­ nal auditory meatus (earn); this opening is slightly oval (Fig­ ures 5, 6). The rim of the external auditory meatus is separated by a shallow notch from the mastoid process on the left side (as in other camivorans), whereas on the right side these are in contact, with no notch present. The middle part of the auditory tube (on both sides) is completely fused to the anteromedial portion of the mastoid. A distinct shallow groove extends ante- rolaterally from the stylomastoid foramen (fsm) between the meatal tube of the bulla and the mastoid process (mp); this groove is present in all phocids and absent in all otarioids, in­ cluding the enaliarctines. The groove has a different prolonga­ tion on the two sides of the skull: it starts from the vagina pro­ cessus hyoidei (for this terminology see Mitchell, 1966, and Burns and Fay, 1970); on the right side of the skull this groove disappears at the middle of the meatal tube (mt), whereas on the left side, the groove continues to the lateral one-third of the tube, where it disappears. A deep cleft continues from the hy­ oid fossa around the posterolateral border of the bulla; this is the typical phocid structure, unlike any otarioid pinniped. The pit for the tympanohyal ligament is separated from the stylo­ mastoid foramen (Figure 6B) and is medial to the latter (a prim­ itive character). In ventral view (Figure 6), the tympanic bulla (tb) is roughly triangular in outline, has a smoothly convex ventral surface, is slightly inflated in its anterolateral (=ectotympanic) parts, and slopes uniformly toward the external auditory meatus (earn). 170 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY earn FIGURE 5.—Devinophoca claytoni, new genus and species, holotype skull Z14523: A, left lateral view, scale bar= 0.75 cm; B, outline drawing of the photograph of left lateral view. For explanation of abbreviations see Figure 3. The anteroposterior length of the tympanic bulla is 1.38 times greater than the distance between the bullae (27 mm); this size relationship is similar to that of Cystophorinae and Phocinae. The length of the auditory bulla (37.5 mm) is 3.4 times the an­ teroposterior width of the glenoid fossa (11 mm), whereas in phocines it is 2.5-3.0 times this width. The long axis of the bulla is slightly oblique to the midline of the skull. The median lacerate foramen (frnl) and musculotubular canal (=canalis musculotubaris, cma) are internally separated by a septum above the anteromedial corner of the bulla (Figures 6, 7B). The entotympanic part is much larger (more than two-thirds of the bulla) than the ectotympanic (ec). Caudally, the entotympanic (en) is more flattened than the ectotympanic along an antero­ posterior axis, and it is separated from the ectotympanic part of the bulla by a distinct ridge instead of a sulcus. The flatter ento­ tympanic is in contrast to the more inflated entotympanic of Mustelinae and other Phocidae, as noted by Wozencraft (1989). The medial portion of the entotympanic close to the petrosal forms a deep, long fissure around the medial side of the bulla, and the carotid foramen (cf) is separated from the posterior lacerate foramen by a tiny lip. The carotid canal (cc) (Figures 6, 7) is partially concealed in the posteromedial wall of the bulla, considerably anterior to the posterior lacerate foramen (the primitive state; see Tedford, 1977). In contrast to phocines (Berta and Wyss, 1994), in Devinophoca claytoni the posterior opening and the posteromedial process of the carotid canal (cc) are visible in ventral view. In D. claytoni the posterior aperture of the carotid canal is horizontal by lateral to the basioccipital, opens in a posteroventral direction, and has a fully formed mar­ gin on the right bulla at its medial side. This is the primitive condition, in contrast to that of other phocids, in which the pos­ terior opening of the carotid canal is directed ventrally. It can be seen clearly on an X-ray (Figure 7) that the carotid canal is quite long; on the right side of the skull it is parallel to the sag­ ittal plane of the skull, but on the left side the canal curves in an anterolateral direction. In the wall of bone dividing the carotid canal from the basioccipital bone (Bo) on the left bulla is a small canal (= inferior tympanic canaliculus), whereas on the right bulla this canaliculus is found inside (medial to) the pos­ teromedial process of the carotid foramen. The posterior lacerate foramen (tip) does not reach the base of the paroccipital process (pp) as it does in other phocids (Mitchell and Tedford, 1973). The posterior lacerate foramen (flp) at the posteromedial corner of the bulla is transversely bi­ lobed and formed of two fenestrae; through the posterior lacer­ ate foramen a septum is visible (Figure 6A). The posterior ca­ rotid foramen (cf) (Figure 6B) does not open into a common fossa with the posterior lacerate foramen (such a common opening is characteristic of ursids, otariids, and also of primi­ tive musteloids) (see Mitchell and Tedford, 1973; Tedford, NUMBER 93 171 B FIGURE 6.—Devinophoca claytoni, new genus and species, holotype skull Z14523: A, in ventral view scale bar= 0.75 cm; B, outline drawing of the photograph of ventral view. For explanation of abbreviations see Figure 3. 1977; Wolsan, 1993). The anterior part of the posterior lacerate foramen is expanded anteroposteriorly, whereas its posterior part extends mediolaterally (=transversely), a condition un­ known in any other phocid (Wyss, 1988; Barnes, 1989; Berta and Wyss, 1994) (Figure 6). The posterior extremity of the pet­ rosal is visible inside the posterior lacerate foramen, behind the bulla (King, 1966; Burns and Fay, 1970; Ray, 1976; Berta and Wyss, 1994) (Figure 6A). On the lateral margin of the basioc­ cipital is a sulcus for the inferior petrosal sinus (sips) (see out­ line drawing from the X-ray on Figure 7B). The mastoid process (mp) does not extend far laterally as it does in Monachinae, but it does form a pronounced promi­ nence lateral to the auditory bulla. The mastoid is not so in­ flated that it obscures the bulla in lateral view; this is the condi­ tion described by Chapskii (1974) and King (1983) for pho­ cines. According to Mitchell and Tedford (1973), a unique pho­ cid feature, especially well developed in the Phocinae, is the in­ flation of the lateral side of the squamosal between the paroc­ cipital and mastoid processes, joining the two in a crest (Figure 6A). This is present also in Devinophoca claytoni. A continu­ ous crest extends from the mastoid process over the paroccipi­ tal process (pp; jugular process in the terminology of Burns and Fay, 1970) to the supramastoid crest and connects to the lamb­ doidal crest (lc); it is rounded and weakly developed. Posteri­ orly the bulla is separated from the base of the paroccipital pro­ cess by a distance of 10 mm, whereas it contacts both mastoid and exoccipital bones. The bases of the paroccipital (= jugular) processes are small, but the processes themselves are broken away. In his analysis of the functional morphology of the inner and middle ear regions, Repenning (1972) concluded that phocids, otariids, and odobenids have different patterns of environmen­ tal adaptation in their auditory apparatus. Unfortunately, the middle ear cavity of D. claytoni is impossible to examine di­ rectly without destroying the tympanic bulla. An X-ray (Figure 7A) does not adequately display the middle ear cavity, but it 172 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY FIGURE 7.—Devinophoca claytoni, new genus and species, holotype skull Z14523: A (left), X-ray of skull in ventral view; B (below), outline drawing based upon Figure 7A and other X-rays of skull in ventral view. Scale bars= 1.0 cm. For explanation of abbreviations see Figure 3. does allow us to learn that in Devinophoca claytoni the inner ear has a phocid construction; i.e., the oval window is a small, slim fissure, not round as in otariids. Along its ventral path to the oval window, the basal whorl of the cochlea extends trans­ verse to the orientation of the skull as in other phocids, rather than posterolaterally as in otariids and odobenids. The fossa of the round window is partially shielded by the bony lip of the entotympanic. The occipital shield (oc) is partially broken away (Figure 4c,D). A thin, median supraoccipital crest (ocs), only the dorsal part of which is preserved, extends dorsally to meet the thick lambdoidal crest. Lateral to the occipital condyles, the lamb­ doidal crests become very thin and continue as sharp ridges to the posterodorsal corners of the mastoid processes. The occipi­ tal condyles are 27 mm apart in the upper part of the foramen magnum and are approximately 18 mm apart below the fora­ men. The border of the intercondylar notch is thick (4.5 mm). The dorsal border of the foramen magnum is semicircular. The foramen for the transverse venous sinus is located medially in­ side of the occipital condyle, and the condyloid canal runs ante- rolaterally (Figure 7B). The upper dental formula is 13, CI, P4, Ml (Figures 5B, 8). The incisor alveoli form a narrow U-shaped arcade. Cingula traverse the lingual sides of the crowns of the maxillary teeth and extend around to the buccal sides both anteriorly and pos- TABLE 2.—Measurements (in mm) of the upper dentition in the holotype skull Z14523 of Devinophoca claytoni, new genus and new species. Teeth 11 alveolus 12 alveolus 13 alveolus C alveolus Pl alveolus P2 crown P3 crown P4 crown Ml crown length - - - 13.2 6.5 10.0 10.0 10.6 9.3 width 3.0 3.0 5.7 10.5 5.6 6.5 8.0 8.2 7.5 NUMBER 93 173 FIGURE 8.—Devinophoca claytoni, new genus and species, holotype skull Z14523: Drawing of teeth in occlusal view. Scale bar= 1.5 cm. teriorly. As a result, the cingula nearly encircle the crowns of the teeth, as in the Enaliarctinae (Barnes, 1992), in Kolpono- mos (Tedford et al., 1994), and in other primitive arctoid car- nivorans (Wolsan, 1993). The number of upper incisors is six, the same as in Phocinae, whereas Monachinae and Cystophori­ nae have four. Preserved alveoli indicate that 13 is much larger than both II and 12, and that II and 12 are approximately equal in size (Table 2). The canines were relatively small and pro­ jected ventrally more than forward, judging from the curvature of the anterior surfaces of their alveoli. Pl has a single root, cir­ cular in cross section. P2 and P3 each have two roots; the ante­ rior roots are round in cross section, but the posterior roots are expanded transversely and bilobed in cross section. The crowns of P2 and P3 are oval in occlusal view and heavily worn. P3 has a distinct lingual cusp (protocone). The crown of P4 is subtriangular in occlusal view, and the buccal side is convex. Its metacone is present as a distinct cusp; the paracone and protocone are heavily worn. P4 has two roots, with the posterior alveolus bigger than the anterior and bilobate in cross section, indicating that the posterior root is made up of two fused roots, one above the protocone and one above the metacone. The bilobed posterior roots of P4 are very similar to the primitive condition in terrestrial carnivores (Berta and Wyss, 1994), unlike most pinniped teeth. A shallow notch (car­ nassial notch) is present between the paracone and metacone (Figure 8). The protocone of P4 and Ml is posteromedial to the paracone, and its buccal side is longer than the lingual, as in terrestrial arctoid camivorans (Wolsan, 1993; Wyss and Flynn, 1993). The Ml of D. claytoni is unlike that of any other known pin­ niped, but some similarity can be seen with the same tooth in the Enaliarctinae (Mitchell and Tedford, 1973; Barnes, 1979, 1992). The crown is triangular in occlusal view and is com­ posed principally of the paracone, which is flanked by the much smaller metacone and protocone (Figure 8). These three cusps are supported respectively by the three roots. A three- rooted Ml has not been reported previously in any phocid. The paracone is long anteroposteriorly and compressed trans­ versely, and the small metacone lies posterolateral to it. The protocone is smaller than on the P4 but equal in size to those on P2 and P3. It is separated from the paracone by a trigon basin (as in enaliarctines, according to Barnes, 1989). The metacone and protocone connect to the paracone by sharp ridges, but a stronger ridge connects the apex of the paracone to the anterior angle of the tooth where it joins the anterior end of the buccal cingulum without a parastyle. The axis of this ridge is parallel to the sagittal plane. Cladistic Analysis CHARACTERS USED IN FAMILY PHOCIDAE For this cladistic analysis, 32 characters were scored (Table 3). 0 designates the most primitive state among the taxa stud­ ied; 1 and 2 are alternate derived states; - indicates unknown or missing data. 1. Tympanic bulla: (0) small; (1) large (Chapski i , 1974:300, fig. 12). 2. External auditory meatus: (0) inframeatal lip well devel­ oped; (1) poorly developed. 3. Mastoid process: (0) not united with paroccipital pro­ cess; (1) united with paroccipital process. 4. Mastoid process: (0) axis of mastoid convexity not di­ rected ventrally; (1) directed ventrally. 5. Mastoid process: (0) prominence lateral to auditory bul­ lae not strongly pronounced; (1) pronounced. 6. Mastoid process: (0) narrow (width of process less than length of process itself); (1) wide (Chapskii, 1974). 7. Mastoid process: (0) bulbous; (1) cylindrical. 8. Mastoid process: (0) width less than or equal to one-half of length of tympanic bulla; (1) width greater than one-half of length of tympanic bulla. 9. Mastoid convexity: (0) not turned down; (1) moderately turned down behind mastoid process; (2) directed sharply downward behind mastoid process. 174 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY TABLE 3.- Species Allodesmus kelloggi* Lutra canadensis* Devinophoca claytoni Cystophora cristata Mirounga leonina Monachus monachus Monachus schauinslandi Callophoca obscura Phoca vitulina Erignathus barbatus Leptophoca lenis —Matrix of character-state data for Phocidae taxa and outgroups analyzed. (* = ^outgroup taxa.) Character number 1 0 1 0 1 1 0 0 1 1 1 0 2 1 0 0 0 0 0 0 1 0 1 0 3 0 0 - 4 1 0 0 1 1 0 0 0 0 1 0 5 0 1 0 0 1 1 1 0 0 0 0 6 0 1 0 1 1 1 1 0 0 1 0 7 1 1 0 1 0 0 0 1 0 0 1 8 1 0 0 1 0 1 1 1 0 0 1 9 1 2 0 2 2 1 0 1 1 0 1 10 1 0 0 0 1 1 1 0 0 0 0 11 0 1 1 0 0 1 1 1 0 0 0 12 2 0 2 1 1 2 2 2 0 0 0 13 2 0 1 1 0 2 2 2 1 1 0 14 1 1 0 1 1 0 0 0 0 0 0 15 0 0 0 0 0 1 1 0 0 1 0 16 1 1 - 0 0 0 0 0 0 0 1 17 0 2 1 1 2 0 0 1 0 0 0 18 0 1 1 1 1 0 0 0 1 1 1 19 1 0 0 0 0 0 0 1 0 0 1 20 0 0 0 0 0 0 1 0 2 2 1 21 0 0 0 0 0 0 1 0 1 1 1 22 0 0 0 2 2 1 1 1 0 0 0 23 0 1 1 0 0 1 1 1 1 1 1 24 2 0 0 25 1 0 0 1 1 0 0 0 0 1 0 26 0 1 1 0 0 1 0 0 0 1 1 27 0 1 0 1 0 0 0 1 1 1 0 28 1 0 0 1 1 0 0 0 1 1 0 29 1 0 0 1 1 0 0 0 0 0 0 30 0 0 0 0 0 1 1 1 0 1 0 31 1 0 0 1 1 0 0 0 0 0 0 32 0 0 0 1 1 0 0 2 0 0 0 10. Nasal bones: (0) anterior ends form one common termi­ nation; (1) anterior ends separated. 11. Nasal bone: (0) maxillary contact longer than frontal contact; (1) frontal and maxillary contacts almost equal in length. 12. Maxilla: (0) very pronounced convexity anterior to or­ bits; (1) short concavity; (2) long concavity (Chapskii, 1974). 13. Anterior palatine foramina: (0) round and deep; (1) oval and shallow; (2) indistinctly marked (Burns and Fay, 1970). 14. Palatal groove: (0) present; (1) absent. 15. Palatal process of maxillary bone: (0) flat; (1) convex. 16. Foramen ovale: (0) hidden under hamular process of pterygoid bone; (1) exposed. 17. Interorbital width: (0) less than 25.0% of mastoid width of skull; (1) less than 30.0% but equal to or greater than 25.0% of mastoid width; (2) equal to or greater than 30.0% of mastoid width (Bums and Fay, 1970; Chapskii, 1974). 18. Paroccipital process: (0) well developed, large, hook- shaped; (1) poorly developed or absent, small conical pro­ jection. 19. Rostrum: (0) short, relative to cranium; (1) elongated (Chapskii, 1974). 20. Diameter of infraorbital foramen: (0) less than diameter of alveolus of upper canine; (1) equal to diameter of alveo­ lus of upper canine; (2) greater than diameter of alveolus of upper canine. 21. Length of auditory bullae: (0) equal to or greater than distance between them; (1) less than distance between them (Bums and Fay, 1970; Chapskii, 1974). 22. Number of incisors: (0) 3/2; (1) 2/2; (2) 2/1 (Chapskii, 1974). 23. Roots of postcanine teeth (P2, P3): (0) one (fused); (1) two (according to Berta and Wyss, 1994). 24. Roots of P4: (0) three; (1) two; (2) one. 25. Crowns of postcanine teeth: (0) multicusped; (1) single cusped. 26. Relative dimensions of postcanine teeth as compared with longitudinal diameter of alveolus of upper canine: (0) more than 60% of longitudinal diameter of upper canine; (1) less than 60% or subequal. 27. Longitudinal diameter of alveolus of upper canine com­ pared with maximal width of infraorbital foramen: (0) subequal in size; (1) more than one-half of maximal width. 28. Basal cingulum of postcanine teeth: (0) well developed; (1) not developed. 29. Number of additional cusps of premolars: (0) more than two; (1) no additional cusps. 30. Premolar crown: (0) aligned parallel to axis of tooth row; (1) seated obliquely. 31. Upper incisors: (0) arranged in a curved arcade; (1) ar­ ranged in a straight line. 32. Second and third upper incisors: (0) third larger than sec­ ond; (1) second larger than third, (2) all upper incisors sub- equal in size. The matrix of character-state data for nine species of fossil and modem phocids is given in Table 3; in addition, these char­ acters were scored for two outgroup taxa. These outgroup taxa are the fossil otarioid Allodesmus and the living mustelid Lutra, reflecting the competing hypotheses of pinniped rela­ tionships, monophyly (Wyss and Flynn, 1993; Berta and Wyss, 1994) or diphyly (McLaren, 1960; Mitchell, 1966; Tedford, 1977). Allodesmines are highly evolved otariids, widely diver­ sified in the middle Miocene, and possess many derived marine carnivore features (Barnes and Hirota, 1995). The analysis of the phocid taxa using these 32 unweighted cranial and dental characters and the bb*; ie* routines in Hennig86 (Farris, 1988) produced two maximally parsimoni­ ous trees, each 79 steps long with a consistency index of 0.51 and a retention index of 0.68. Use of Hennig86's successive-weighting option reduced the number of trees from two to one, leaving this part of the tree much better resolved (Figure 9). Points where the nodes of the present tree correspond, the clades, traditionally recognized as subfamilies, are indicated. Only one new higher taxon name is introduced here: inclusion of Devinophoca within the Phocidae requires recognition of the new subfamily Devinophocinae. The nodes of the cladogram shown in Figure 9 are supported by the following character transformations: NUMBER 93 175 § -©• ^I> MS- "-©• h Allodesmus kernensis — Lutra canadensis + Devinophoca claytoni — Cystophora cristata Mirounga leonina + Callophoca obscura Monachusmonachus Monachus schauinslandi -r Leptophoca lenis Erignathus barbatus Phoca vitulina FIGURE 9.—Nelson consensus tree of two trees of the hypothesized phyloge­ netic relationships among taxa of true seals and two outgroups, generated by Hennig86 using 32 cranial characters and the successive weighting option. Tree length, 79 steps; consistency index, 0.51; retention index, 0.68. Character states are given in Table 3. Node 1 (Family Phocidae): 3(1); 23(1); 24(1). These apo- morphic characters (relative size of the mastoid process and its lack of union with the paroccipital process; two roots on post­ canine teeth) are treated as synapomorphies of the family Pho­ cidae. Node 2 (subfamily Devinophocinae, possibly paraphyletic): 24(0). Autapomorphy of D. claytoni: three roots of the poste­ rior postcanine teeth fused. Node 3 (subfamily Cystophorinae): 9(2); 22(2); 29(1). The mastoid convexity directed sharply downward behind the mas­ toid process; reduced number of incisors; no additional cusps on premolars. Node 4 (subfamily Monachinae): 11(1); 13(1). The relative dimensions of the frontal and maxillary parts of the nasal bones; the shape of the anterior palatine foramina. Also, char­ acter 13(2) is homoplasious in Phoca vitulina and Erignathus barbatus. Node 5 (subfamily Phocinae): 20(1, 2); 21(1). The diame­ ter of the infraorbital foramen is equal to or greater than the di­ ameter of the alveolus of the maxillary canine; the length of the auditory bullae is less than the distance between them. Also, character 20(1) is homoplasious in Monachus schauinslandi and Leptophoca lenis. Discussion and Conclusions The traditional separation of Phocidae into subfamilies Phocinae, Monachinae, and Cystophorinae has been inten­ sively debated during the past 40 years, but overall it is the sys­ tem still used today. The well-preserved skull of Devinophoca claytoni from the early middle Miocene (16.4 Ma) shows a mix of subfamilial characters. The features it shares with Phocinae are weakly pronounced mastoid process, 5(0); six incisors, 22(0); and poorly developed paroccipital process, 18(1) (also shared with Cystophorinae). The characters similar to Monachinae are the shape of maxillae, 12(2) and the ratio between frontal and max­ illary contacts of the nasal bones, 11(0) (shared with Phocinae also). Characters shared with Cystophorinae are ratio between interorbital space and mastoid width, 17(1); ratio between the length of auditory bullae and distance between them, 21(0); and oval-shaped and deep anterior palatine foramina, 13(1). Moreover, this skull has primitive features that are not known in any of the three subfamilies: Ml is triangular, with three cusps and three roots; in P2-P4 the larger posterior roots are clearly made up of two fused roots; and the sagittal crest is very well developed for a phocid, though weak compared with some camivorans. This plesiomorphic pinniped is therefore considered a sister taxon to the other three subfamilies of Pho­ cidae. Because of its late geologic age, Devinophoca might represent a very primitive relict of the common ancestor of the groups of Phocidae. Wozencraft (1989) identified the shortened and narrowed contact between the premaxilla and the nasal, as seen in phoc­ ids and lutrines, as a primitive condition. Barnes (1989), Berta and Wyss (1994), and Wyss and Flynn (1993) had the opposite opinion and suggested that this character is derived. The su­ praorbital process of the frontal is a small, rounded knob that is the primitive condition seen in terrestrial camivorans (Berta and Wyss, 1994). In Devinophoca claytoni, the nasals bones end near the fron­ tal-maxillary contact. According to Berta and Wyss (1994), this is a primitive condition among terrestrial camivorans, and among otariids and odobenids, as exemplified by the enaliarc­ tine otarioids Enaliarctos and Pteronarctos. In the more-de­ rived phocids, the nasals extend far posterior to the frontal- maxillary suture between the frontals, and they share this fea­ ture convergently with the otarioids Desmatophoca and Al­ lodesmus. Our material supports that statement; this plesiomor­ phic representative of phocids, i.e., Devinophoca, shares the primitive form of the nasals with terrestrial camivorans, espe­ cially with Mustelidae. Computer-assisted phylogenetic analysis of some fossil and living phocids supports the monophyly of the family and rec­ ognizes four principal clades, including the new subfamily Devinophocinae. This phylogenetic framework supports the classification used by Chapskii (1955, 1974), Scheffer (1958), and Koretsky and Grigorescu (2002), and it conflicts with the various arrangements proposed by King (1966), Bums and Fay (1970), Wyss (1994), and McKenna and Bell (1997). We are reluctant to name a new higher-level taxon, but the very un­ usual morphology of Devinophoca does not allow us to assign 176 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY it to any known subfamily. We thus erect a new taxon at the subfamilial level. The interorbital construction in Devinophoca claytoni, the large size of the orbits, and the loss of the lacrimal bone also are found in other aquatic species, including the Enaliarctinae among the otarioids, as advanced aquatic adaptations (Howell, 1928; Savage, 1957; Mitchell and Tedford, 1973). The devel­ opment of the dorsal sagittal sinus (as clearly seen in the X- ray; Figure 7) for venous drainage of the brain is suggested by Mitchell and Tedford (1973) to be related to diving adapta­ tions. Striking similarities between the skulls of Devinophoca and Enaliarctos include a short rostrum. Although the ventral part of the skull is not preserved in Devinophoca, we presumed that the interorbital part of the skull has the shape of an I-beam, as seen in Enaliarctos and other Phocidae. Mitchell and Tedford (1973) suggested that this construction is an adaptation for structural rigidity, while accommodating the large eyes (as in desmatophocines also). The presence of a sagittal crest and strong lambdoidal crests, the complexity of the teeth, and the short rostrum all indicate that the feeding mechanism of Devi­ nophoca claytoni was adapted for slower and more powerful closing of the jaws, rather than the rapid action of the fish-eat­ ing pinnipeds. We conclude that Devinophoca claytoni was adapted to eating mollusks and crustaceans. Some of the features mentioned above, such as the paroccip­ ital process separated from the mastoid and a single-chambered bulla with large entotympanic and much smaller ectotympanic, are more similar to the primitive Mustelidae or Procyonidae than to Ursidae (in accord with Tedford, 1977). Literature Cited Andrusov, D. 1969. Chaines enterrees, meganticlinaux et horst dans la region des Car- pathes occidentales. Geologicky Zbornik SAV, Bratislava, 20(1): 39-45. Barath, I., A. Nagy, and M. Kova 1994. [Sandberg Member—Late Badenian Marginal Sediments on the Eastern Margin of the Vienna Basin.] Geologicke Prdce, Sprdvy (Bratislava), 99:59-66. [In Slovak.] Bames, L.G. 1979. Fossil Enaliarctine Pinnipeds (Mammalia: Otariidae) from Pyramid Hill, Kern County, California. Contributions in Science, Natural History Museum of Los Angeles County, 318:1—41. 1989. A New Enaliarctine Pinniped from the Astoria Formation, Oregon, and Classification of the Otariidae (Mammalia: Carnivora). Contri­ butions in Science, Natural History Museum of Los Angeles County, 403:1-26. 1992. A New Genus and Species of Middle Miocene Enaliarctine Pin­ niped (Mammalia, Carnivora, Otariidae) from the Astoria Forma­ tion in Coastal Oregon. Contributions in Science, Natural History Museum of Los Angeles County, 431:1-27. Bames, L.G., and K. Hirota 1995. Miocene Pinnipeds of the Otariid Subfamily Allodesminae in the North Pacific Ocean: Systematics and Relationships. In L.G. Barnes, N. Inuzuka, and Y. Hasegawa, editors. Evolution and Bioge­ ography of Fossil Marine Vertebrates in the North Pacific. Island Arc. 3(4):329-360. Berta, A., and A.R. Wyss 1994. Pinniped Phylogeny. In A. Berta and T.A. Demere, editors, Contri­ butions in Marine Mammal Paleontology Honoring Frank C. Whit­ more, Jr. Proceedings of the San Diego Society of Natural History, 29:33-57. Burns, J.J, and F.H. Fay 1970. Comparative Morphology of the Skull of the Ribbon Seal, Histrio- phoca Fasciaia, with Remarks on Systematics of Phocidae. Journal of Zoology (London), 161(3):363-394. Chapskii, K.K. 1955. [An Attempt at Revision of the Systematics and Diagnostics of Seals of the Subfamily Phocinae.] Trudy Zoologicheskogo Instituta Akademii Nauk SSSR, 17:160-199. Leningrad. [In Russian. English translation by T.F. Jeletzky, Fisheries Research Board of Canada, Translation Series, 114, 57 pages, 1957.] 1974. [In Defense of Classical Taxonomy of the Seals of the Family Pho­ cidae.] In [Theoretical Questions on the Systematics and Phylogeny of Animals]. Trudy Zoologicheskogo Instituta Akademii Nauk SSSR, 53:282-334. [In Russian.] Domesco, G.-T., and G.-V Marcoci 1958. Etude comparative du crane des pinnipedes. Annates des Sciences Naturelles, Zoologie, 11: xx+158-182. Driesch, Angela von den 1976. A Guide to the Measurement of Animal Bones from Archeological Sites. Bulletin of the Peabody Museum of Archaeology and Ethnol­ ogy [Harvard University], 1:1-137. Farris, J.S. 1988. Hennig86, Version 1.5. New York: Port Jefferson Station. Fejfar, 0 . 1974. Die Eomyiden und Cricetiden (Rodentia, Mammalia) des Miozans der Tschechoslowakei. Palaeontographica, Part A, 16: 146 pages. Gervais, P. 1848-1852. Zoologie et paleontologie Francoises (animaux vertebres) ou nouvelles recherches sur les animaux vivants et fossiles de la France, 1 [text]: 18-141. Gervais, P., and M. de Serres 1847. Sur les mammiferes fossiles des sables marins tertiaires de Montpel­ lier. Comptes Rendus de I'Academie des Sciences (Paris), 28(8): 799-801. Heptner, V.G, K.K. Chapskii, B.A. Arsen'ev, and V.E. Sokolov 1996. Pinnipeds and Toothed Whales. In V.G. Heptner and N.P. Naumov, editors, Mammals of the Soviet Union, 2(3): xxx + 995 pages. Wash­ ington, D.C. [Originally published in Russian as Mlekopitayushchie Sovetskogo Soyuza. Moscow: Vysshaya Shkola, 1976.] Holec, P. 1985. Finds of Mastodon (Proboscidea, Mammalia) Relics in Neogene and Quaternary Sediments of Slovakia (SSR). Zdpadny Karpaty, Pale­ ontologia, 10:13-53. 1986. Neueste Resultate der Untersuchungen von neogenen und quar- taeren Nashornen, Baeren und Kleinsaeugern im Bereich der West- karpaten (Slowakei). Acta Universitatis Carolinae, Geologica (Prague), 2:223-231. Holec, P., J. Klembara, and §. MeszaroS 1987. Discovery of New Fauna of Marine and Terrestrial Vertebrates in Devinska Nova Ves. Geologicky Zbornik, Geologica Carpathica, 38(3):349-356. Bratislava. NUMBER 93 177 Holec, P., and M. Sabol 1996. [The Tertiary Vertebrates from Devinska Kobyla.] Mineralia Slo- vaca, 28:519-522. [In Slovak.] 1997. [Fossils of the Devinska Kobyla Hill.] In V Ferakova, editor, Flora of the Devinska Kobyla Hill. 639 pages. Bratislava: APOP-Edition. [In Slovak.] Howell, A.B. 1928. Contribution to the Comparative Anatomy of the Eared and Earless Seals (Genera Zalophus and Phoca). Proceedings of the United States National Museum, 3(15):1—42. King, J. 1966. Relationships of the Hooded and Elephant Seals (Genera Cystophora and Mirounga). Journal of Zoology, London, 148(4):385—398. 1983. Seals of the World. Second edition, 240 pages. Ithaca, New York: Cornell University Press. Koretsky, I.A., and D. Grigorescu 2002. The Fossil Monk Seal Pontophoca sarmatica (Alekseev) (Mamma­ lia: Phocidae: Monachinae) from the Miocene of Eastern Europe. In R. Emry, editor, Cenozoic Mammals of Land and Sea: Tributes to the Career of Clayton E. Ray. Smithsonian Contributions to Paleo­ biology. 93:149-162. McKenna, M.C, and S.K. Bell 1997. Classification of Mammals above the Species Level. 631 pages. New York: Columbia University Press. McLaren, I.A. 1960. Are the Pinnipedia Biphyletic? Systematic Zoology, 9:18-28. Mitchell, E.D. 1966. The Miocene Pinniped Allodesmus. University of California Publi­ cations in Geological Sciences, 61:1-105. Mitchell, E.D., and R.H. Tedford 1973. The Enaliarctinae, a New Group of Extinct Aquatic Carnivora and Consideration of the Origin of the Otariidae. Bulletin of the Ameri­ can Museum of Natural History, 151(3):203—284. Pierard, J. 1971. Osteology and Myology of the Weddell Seal, Leptonychotes wed- delli (Lesson, 1826). In W.H. Burt, editor, Antarctic Pinnipedia. Ant­ arctic Research Series, 18:53-108. Ray, CE. 1976. Phoca wymani and Other Tertiary Seals (Mammalia: Phocidae) De­ scribed from the Eastern Seaboard of North America. Smithsonian Contributions to Paleobiology, 28: 36 pages. Repenning, CA. 1972. Underwater Hearing in Seals: Functional Morphology. In R.J. Harri­ son, editor, Functional Anatomy of Marine Mammals, pages 307- 331. London and New York: Academic Press. Rogl, F, and G. Daxner-Hock 1996. Late Miocene Paratethys Correlations. In R.L. Bernor, V Fahlbusch, and H.-W. Mittmann, editors, Evolution of Western Eurasia Neo­ gene Mammal Faunas, pages 46-56. New York: Columbia Univer­ sity Press. Savage, R.J.G. 1957. The Anatomy of Potamotherium, an Oligocene Lutrine. Proceed­ ings of the Zoological Society of London, 129(2): 151-244. Schaller, O., editor 1992. Illustrated Veterinary Anatomical Nomenclature. 617 pages. Stutt­ gart: Ferdinand Enke Verlag. Scheffer, V.B. 1958. Seals, Sea Lions and Walruses: Review of the Pinnipedia. 179 pages. Stanford, California: Stanford University Press. Simpson, G.G. 1945. The Principles of Classification and a Classification of Mammals. Bulletin of the American Museum of Natural History, 85: xvi + 350 pages. Steininger, F.F., and L.A. Nevesskaya 1975. Stratotypes of Mediterranean Neogene Stages. Volume 2, 364 pages. Bratislava: Committee on Mediterranean Neogene Stratigraphy. Steininger, F.F., R.L. Bernor, and V Fahlbusch 1989. European Neogene Marine/Continental Chronologic Correlation. In E.H. Lindsay, V Fahlbusch, and P. Mein, editors, European Neo­ gene Mammal Chronology, pages 15-46. New York: Plenum Press. Svagrovsky, J. 1981. Lithofazielle Entwicklung und Molluskenfauna des oberen Bade- nien (Miozan M4d) in dem Gebiet Bratislava—Devinska Nova Ves. Zdpadny Karpaty, Paleontologia, 7: 203 pages. Bratislava: GUDS. Tedford, R.H. 1977. Relationship of Pinnipeds to Other Carnivores (Mammalia). System­ atic Zoology, 25(4):363-374. Tedford, R.H., L.G. Barnes, and CE. Ray 1994. The Early Miocene Littoral Ursoid Carnivoran Kolponomos: Sys­ tematics and Mode of Life. In A. Berta and T.A. Demere, editors, Contributions in Marine Mammal Paleontology Honoring Frank C. Whitmore, Jr. Proceedings of the San Diego Society of Natural His­ tory, 29:1-32. Thenius, E. 1950. Ergebnisse der Neuuntersuchung von Miophoca vetusta Zapfe (Pho­ cidae, Mammal.) aus dem Torton des Wiener Beckens. Anzeiger der Osterreichischen Akademie der Wissenschaften, Vienna, Mathema- tisch-Naturwissenschaftlichen Klasse, 5:1-9. 1952. Die Saugetierfauna aus dem Torton von Neudorf an der March (SR). Neues Jahrbuch fiir Geologie und Paldontologie, Abhandlungen, 96(1):27-136. 1969. Stammesgeschichte der Saugetiere (einschliesslich der Hominiden). In J.-G. Helmcke, D. Starer, and H. Wermuth, editors, Handbuch der Zoologie, pages 393—105. Berlin: Walter De Gruyter & Co. 1979. Die Evolution der Saugetiere: Eine Ubersicht iiber Ergebnisse und Probleme. 270 pages. Stuttgart, New York: Gustav Fischer Verlag. 1992. Neue Befunde und Erkenntnisse zur Verbreitungsgeschichte der Robben (Mammalia: Carnivora: Pinnipedia). Anzeiger der Oster­ reichischen Akademie der Wissenschaften, Sitzungsberichte der Mathematisch-Naturwissenschaftlichen Klasse, 129:67-73. Trelea, N., and T. Simionescu 1985. Au sujet de quelques formes de vertebres des formations sarmati- ennes de chelia du departement Jassy. Analele Stiintifice ale Univer- sitatii "Al. I. Cuza" din lasi, Geologie-Geografie, 31(2): 18-20. Wolsan, M. 1993. Phylogeny and Classification of Early European Mustelidae (Mam­ malia: Carnivora). Acta Theriologica, 38(4):345-384. Wozencraft, W.C. 1989. The Phylogeny of the Recent Carnivora. In J.L. Gittleman, editor, Carnivore Behavior, Ecology, and Evolution, pages 495-593. Ith­ aca, New York: Cornell University Press. Wyss, A.R. 1988. On "Retrogression" in the Evolution of the Phocinae and Phyloge­ netic Affinities of the Monk Seals. American Museum Novitates, 2924: 38 pages. 1994. The Evolution of Body Size in Phocids: Some Ontogenetic and Phy­ logenetic Observations. In A. Berta and T.A. Demere, editors, Con­ tributions in Marine Mammal Paleontology Honoring Frank C. Whitmore, Jr. Proceedings of the San Diego Society of Natural His­ tory, 29:69-75. Wyss, A.R., and J.J. Flynn 1993. A Phylogenetic Analysis and Definition of the Carnivora. In F.S. Szalay, M.J. Novacek, and M.C. McKenna, editors, Mammal Phy­ logeny: Placentals, pages 32-52. New York: Springer-Verlag. Zapfe, H. 1937. Ein bemerkenswerter Phocidenfund aus dem Torton des Wiener- Beckens. Verhandlungen der Zoologisch-Botanischen Gesellschaft in Wien, 86/87:271-276. 1949. Eine mittelmiozane Saugetierfauna aus einer Spaltenfullung bei 178 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY Neudorf an der March (SR). Anzeiger der Osterreichischen Akade­ mie der Wissenschaften, Sitzungsberichte der Mathematisch-Natur- wissenschaftlichen Klasse, 86(7): 173-181. 1950. Die Fauna der miozanen Spaltenfullung von Neudorf an der March (CSR), Chiroptera, Carnivora. Anzeiger der Osterreichischen Akad­ emie der Wissenchaften, Sitzungsberichte der Mathematisch-Natur- wissenschaftlichen Klasse, 159:51-64, 109-141. 1952. Die Pliopithecus—Funde aus der Spaltenfullung von Neudorf an der March. Verhandlungen der Geologoschen Bundesanstadt (Wien), special issue C: 126-130. 1953. Das geologische Alter der Spaltenfullung von Neudorf an der March. Verhandlungen der Geologischen Bundesanstalt (Wien), 3: 195-202. 1960. Die Primatenfunde aus der Miozan Spaltenfullung von Neudorf an der March, Tschechoslowakei. Schweizerische Paleontologische Abhandlungen, 78:1-293. 1976. Die Fauna der miozanen Spaltenfullung von Neudorf a.d. March (SSR), Chalicotherium grande (BLV). Anzeiger der Osterreichi­ schen Akademie der Wissenschaften, Sitzungsberichte der Mathe- matisch-Naturwissenschaftlichen Klasse, Part 1, 185(7):91-112. 1979. Chalicotherium grande (Blainv.) aus der miozanen Spaltenfullung von Neudorf an der March (Devinska Nova Ves), Tschecho­ slowakei. Neue Denkschriften des Naturhistorischen Museums in Wien, 2: 282 pages. 1983. Die Fauna der miozanen Spaltenfullung von Neudorf an der March (SSR). Suidae. Anzeiger der Osterreichischen Akademie der Wis­ senschaften, Sitzungsberichte der Mathematisch-Naturwissenschaft- lichen Klasse, Part 1, 192(7): 167-182. Paleontology of the Late Oligocene Ashley and Chandler Bridge Formations of South Carolina, 1: Paleogene Pinniped Remains; The Oldest Known Seal (Carnivora: Phocidae) Irina A. Koretsky and Albert E. Sanders ABSTRACT The proximal halves of two femora from the Chandler Bridge and Ashley Formations (early Chattian, late Oligocene) near Charleston, South Carolina, provide the earliest evidence to date of true seals. They are clearly referable to the Phocidae and furnish information regarding osteological and myological features that had evolved in early phocids by early Chattian time. Although not determinate to the generic level, these specimens represent a taxon closely comparable to the most specialized phocid, the modem genus Cystophora. Introduction The long-standing question of phylogenetic relationships among the Odobenidae, Otariidae, and Phocidae is still open. The debates about the monophyletic or diphyletic origin of pin­ nipeds from a series of ancestors (mustelid- or ursid-like ani­ mals) are not yet resolved. The new material reported herein, the proximal parts of two femora from the late Oligocene of South Carolina, now in The Charleston Museum (ChM), helps little in solving the problems of phylogeny, but we have for in­ vestigation the earliest and most primitive remnants of Pho­ cidae. Herein we refer to the two partial femora from the late Oligocene as the "Oligocene seal." The two new specimens ap­ pear to represent the same taxon, but they are so incomplete that any taxonomic assignment below the family level would be mere speculation. This material nonetheless does provide in­ formation concerning some derived (=apomorphic) and primi­ tive (=plesiomorphic) characters of the Phocidae. Irina A. Koretsky, Research Associate, Department of Paleobiology, National Museum of Natural History, Smithsonian Institution, Wash­ ington, D.C. 20560-0121. Albert E. Sanders, Department of Natural Sciences, The Charleston Museum, 360 Meeting Street, Charleston, South Carolina 29430. For comparisons we used postcranial material from several living and fossil representatives of aquatic and semiaquatic car­ nivores, including Mustelidae (the genera Lutra, Enhydra, and Potamotherium); Otariidae (the genera Zalophus and Allodes­ mus, the latter the most controversial eared seal (Berta and Wyss, 1994; Barnes and Hirota, 1994)); and Phocidae (for ex­ ample, the genus Cystophora as the most specialized represen­ tative of the family). ACKNOWLEDGMENTS.—The authors are grateful to James Mead and Charles Potter for access to modern comparative ma­ terial, to Clayton E. Ray for access to fossil pinniped collec­ tions, to Victor Krantz and Steven Jabo for specimen photogra­ phy (all National Museum of Natural History (NMNH), Smithsonian Institution, Washington, D.C), and to Robert J. Emry (NMNH), Daryl Domning (Howard University, Wash­ ington, D.C), and Mieczyslaw Wolsan (Institute of Paleobiol­ ogy, Warsaw, Poland) for their helpful comments on several versions of this manuscript. MATERIAL AND LOCALITY.—Proximal portion of right femur (ChM PV5712), from a ditch in the Irongate subdivision, ap­ proximately 3.9 km (2.4 mi.) SW of Summerville, Dorchester County, South Carolina, collected by Vance McCollum, sum­ mer 1978; cast of ChM PV5712 (USNM 299831) in the NMNH (which subsumed the collections of the former United States National Museum). Proximal portion of right femur (ChM PV5713), from a ditch on Trolley Road (County Road 199), approximately 3.0 mi. (4.8 km) SW of Summerville, Dorchester County, South Carolina, collected by Chris Kenney, 6 December 1978; cast of ChM PV5713 (USNM 306508). FORMATION AND AGE.—ChM PV5712: Chandler Bridge Formation, early Chattian (late Oligocene). ChM PV5713: Ashley Formation, early Chattian (late Oligocene). The Chandler Bridge Formation, an unconsolidated noncal- careous marine unit rich in vertebrate fossil remains, occurs at numerous disjunct localities in Berkeley, Charleston, and Dorchester Counties, South Carolina. It unconformably over- 179 180 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY FIGURE 1.—Femora of Oligocene seals (xl). Right femur, original ChM PV 5712, cast USNM 299831, from Chandler Bridge Formation, early Chattian, late Oligocene: a, caudal view; b, cranial view. Right femur, original ChM PV 5713, cast USNM 306508, from Ashley Formation, lower Chattian, late Oligocene: c, caudal view; d, cranial view. lies the Ashley Formation, a highly indurated calcarenite un­ derlying most of the area encompassed by the aforementioned counties (Sanders, 1980; Sanders et al., 1982). E. Martini has found the nannoplankton of the Ashley Formation to be refer­ able to zone NP24 (Sanders and Barnes, 2002). Sanders et al. (1982) have estimated the Chandler Bridge Formation to be ap­ proximately 28 million years old and the Ashley Formation to be about 30 million years in age. Both units have yielded abun­ dant remains of sharks; bony fishes; sea turtles (Carolin- achelys, Procolpochelys, Syllomus, two dermochelyids); three nonmarine turtle taxa; a large, newly described estuarine croco­ dile, Gavialosuchus carolinensis Erickson and Sawyer (1996); marine birds (including a large pseudodontorn with a wingspan of about six meters); sirenians; and more than 20 taxa of primi­ tive cetaceans. Now a pinniped is added to the impressive fau­ nas of marine vertebrates from these two formations. PHOCIDAE Genus and Species Indeterminate DESCRIPTION.—The femora (Figure 1, Table 1) are similar in size to those of the modern harp seal, Pagophilus groenlandi- cus. The shaft of the bone is flattened in the cranial-caudal di­ rection. The greater trochanter extends proximally slightly higher than the head and is slender, although not well devel­ oped. The trochanteric fossa is shallow and wide open, actually not reaching the middle of the greater trochanter. The intertro­ chanteric crest is not expanded. The lesser trochanter is very well developed and is located on the posteromedial side of the bone at the same level as the distal border of the greater tro­ chanter. The cervix is short but narrow relative to the bone's mass. The head is damaged, but it clearly has an indentation for attachment of the ligamentum teres in the center of the head. NUMBER 93 181 TABLE 1.—Measurements (in mm) of the late Oligocene seal femora from the Chandler Bridge Formation (ChM PV5712) and the Ashley Formation (ChM PV5713), Dorchester County, South Carolina. • ' • Character Length of greater trochanter Intertrochanteric length Height of head Width of proximal epiphysis Width of head Narrowest part of diaphysis Maximum thickness of greater trochanter Anteroposterior thickness of diaphysis Diameter of cervix Measurements ChM PV57I2 29.5 42.0 20.5 54.0 18.0 27.0 13.5 16.0 17.0 ChMPV5713 21.5 56.5 19.0 28.0 18.0 20.0 This insertion of the ligamentum teres is not a completely de­ veloped fovea as it is in most terrestrial mammals. In phocids this fovea is absent. The slight difference in bone morphology of the two Oli­ gocene femora perhaps could be ascribed to sexual dimorphism (Koretsky, 1987), but the limitations of the material preclude a more precise interpretation. According to authors who have studied the functional mor­ phology and myology of the limbs of different carnivores (Howell, 1928, 1930; Savage, 1957; Mori, 1958; Mitchell, 1966; Pierard, 1971; Tarasoff, 1972; Sokolov et al., 1974), the lesser trochanter is the point of insertion of the iliopsoas mus­ cle. These authors also noted, however, that if the lesser tro­ chanter is absent, then the pectineus and adductor muscles in­ sert on this place; that is true for all comparatively aquatic and semiaquatic mammals. The iliopsoas muscle is a large, heavy muscle connecting the lumbar spine and ilium with the femur. The actions of this muscle are to flex and laterally rotate the fe­ mur, tilt the pelvis forward, flex the thigh upon the pelvis, and adduct the femur (working as a synergist with the adductor brevis) (Gordon, 1983). The quadratus femoris muscle is absent in extant phocids. In the otter, this reduced and weakened muscle inserts on the dis­ tal part of the greater trochanter, on the caudal border of the fe­ mur (Sokolov et al., 1974). In sea lions (otariids), the insertion of the m. quadratus femoris is along the entire laterocaudal bor­ der of the femur (Howell, 1928). The action of the quadratus femoris muscle is to extend the hip joint. Howell (1928, 1930), Pierard (1971), and other researchers described the trochanteric fossa as a place of attachment of the obturator internus and extemus muscles, together with the two gemelli muscles, the superior and inferior. The gemelli and ob­ turator extemus muscles arise from the lateral border of the ob­ turator foramen and its membrane. The two gemelli join with the obturator intemus muscle to form a common tendon for in­ sertion. (Pierard at the same time (1971:68) mistakenly stated that the origin of the tendon of the obturator intemus muscle is a shallow groove in the middle one-third of the pubic edge.) The action of the obturator extemus muscle is to rotate the fe­ mur laterally. The actions of the gemelli and obturator intemus are to abduct the femur and also to rotate it laterally. COMPARISONS.—The Oligocene seal differs from other car­ nivores we examined (except for Cystophora and Zalophus) in its shallow and poorly defined trochanteric fossa, from all ex­ cept Cystophora by its slight extension of the greater trochanter above the head, and from all except Lutra and Potamotherium in having a pit for attachment of the ligamentum teres. Other distinct differences between this seal and other genera also exist. It differs from Cystophora (the hooded seal) in hav­ ing a lesser trochanter (but this is always absent in females of Cystophora); a narrower and less prominent greater trochanter; and a more delicate, not massive, overall bone structure. So far as they can be determined, the muscle attachments for the Oli­ gocene seal are in general the same as for Cystophora, but with one exception: in female hooded seals, which lack a lesser tro­ chanter, the tendons of the pectineus and adductor cranialis muscles are inserted on the same spot on the bone, and the ili­ opsoas muscle inserts upon the medial tuberosity of the tibia (Howell, 1928). From Lutra this seal differs in the presence of a well-devel­ oped and distinct lesser trochanter and a much broader shaft. In general, there are no differences between these two taxa in in­ sertion of the described muscles; gemelli, obturator extemus, and intemus muscles all insert into the trochanteric fossa, and the iliopsoas muscle inserts onto the lesser trochanter (Sokolov etal., 1974). The Oligocene seal differs from Enhydra in the shorter cer­ vix and smaller size of the head compared with the rest of the bone. Insertions of the muscles into the fossa trochanterica are almost the same in Enhydra and Lutra, but the mass of these muscles is apparently greater in Enhydra. In contrast to the Oli­ gocene seal, Enhydra lacks the ligamentum teres (Sokolov et al., 1974; Howard, 1975). Its differences from Potamotherium are the more medial lo­ cation of the lesser trochanter and the absence of the intertro­ chanteric crest. In Potamotherium the tendon of the obturator intemus did not merge with the common tendon of the obtura­ tor extemus and gemelli; the quadratus femoris was located on the intertrochanteric crest (Savage, 1957). The Oligocene seal differs from Zalophus in having a more medially located lesser trochanter and in lacking an intertro­ chanteric crest. Presumably the insertion of the obturator exter- nus in the Oligocene seal is located distal to the trochanteric fossa on the posteromedial side of the greater trochanter; the tendons of the two gemelli muscles were not joined together. In Zalophus, the tendon of the gemellus superior muscle does not join with the tendon of the obturator intemus muscle; the ten­ don of the adductor brevis muscle inserts into the intertrochan­ teric crest (Howell, 1928; Lyon, 1937); and the ligamentum teres is absent. From Allodesmus, the Oligocene seal differs in its smaller and thinner greater trochanter and its lack of the intertrochant­ eric crest. By analogy to the living otariids, the adductor brevis 182 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY TABLE 2.—Comparative diagnostic characters of the femora of the Oligocene seal and some other members of Car­ nivora. (+=character present; —character absent; +/-=character variable.) Character 1. Shallow and open trochanteric fossa 2. Greater trochanter slightly extended over head 3. Lesser trochanter present 4. Location of lesser trochanter on posteromedial side 5. Cervix short and narrow 6. Intertrochanteric ridge absent 7. Fovea for ligamentum teres present Olig. seal + + + + + + + Cystophora + + +/- + - + - Lutra - - - - + + + Enhydra - - - + + + - Potamotherium - - + + + - + Zalophus + - + - - - - Allodesmus - - + + + - - muscle in Allodesmus inserted on the intertrochanteric crest (Mitchell, 1966; Hirota, 1983), which is not present in the Oli­ gocene seal. The fovea capitis, and hence presumably the liga­ mentum teres, is absent also in Allodesmus. Discussion The osteology and interpreted myology of these proximal femora indicate that these remains are definitely more closely related to the Phocidae than to other carnivores (Table 2). In addition, the functional morphology of even these proximal femora can suggest whether this Oligocene seal used its hind limb predominantly for propulsion in water or to some degree for terrestrial movement as well. For example, the diaphysis of the femur of Lutra is almost circular in cross-section and is better prepared for terrestrial lo­ comotion than for aquatic movement. Increasing the surface area for insertion of the adductor muscle makes the shaft of the femur broader. The enlarged adductor muscle also helps pre­ vent rotation and movement of the femur. The diaphysis in the Oligocene seal is flatter and wider than in Lutra, Potamothe­ rium, and Enhydra; similar to that of Allodesmus; and narrower than in Cystophora and Zalophus. The literature and observation of living animals both indicate that terrestrial locomotion in Lutra (and presumably in Pota­ motherium) is only slightly impaired. The increased surface area of the femur suggests that Enhydra, Zalophus, and the Oli­ gocene seal have more limitations on land. Cystophora has lost the ability to turn its hind limbs forward on solid ground, as have other modem Phocidae, in contrast to the Otariidae and Odobenidae. When the animal is on land, the ligamentum teres allows rotational movement of the femur and helps to fix the head of the femur to the acetabular fossa. This ligament disap­ pears when the animal spends less time on land (Tarasoff, 1972) and has greater freedom of movement of the femur. In the Oligocene seal there is morphological evidence for a higher degree of terrestriality than in Cystophora and Zalophus but much less than in Lutra and Potamotherium. The distribution among carnivore taxa of osteological fea­ tures associated with lower-limb propulsion can assist in inter­ preting the specific method of locomotion in both aquatic and terrestrial movement. The main effect of the flattening and ex­ pansion of the greater trochanter is thus to increase the area of insertion of the muscles that rotate the femur (Howell, 1928; Tarasoff, 1972; Gordon, 1983). The Oligocene seal, with its greater trochanter only slightly expanded, used its hind limb in lateral movement and with some flexion and adduction (similar to Cystophora) for aquatic movement. This involved little rota­ tion of the femur, suggesting that these Oligocene animals were not as good swimmers as later phocids. As the antagonistic muscle to those lateral rotators, the iliop­ soas muscle, located on the caudal side of the femur, becomes very much reduced and weakened. In parallel to the decrease in its stress on the lesser trochanter, the latter is reduced and fi­ nally disappears. In those mammals that do not bring their hind limbs forward, the lesser trochanter is less and less developed. We can observe this progressive reduction from Lutra, to the Oligocene seal, to Cystophora (where it is small in males and absent in females), and finally to Phoca, where it is completely absent. The lateral rotation of the lower limb results in turning the hind foot so that it can move in a horizontal plane. The anatomy of Enhydra and Zalophus shows that the feet of these animals are adapted predominantly for lateral movement, but that vertical and horizontal movement also is possible for body support and locomotor activity on land. On the other hand, the anatomy of the pelvic limb of Lutra is similar to that of completely terrestrial carnivores. Phocids have undergone major morphological changes that are closely correlated with their specialized methods of aquatic locomotion. In its patterns of aquatic locomotion, Enhydra shows analogy to both Lutra and Cystophora and to the Oligocene seal. Moreover, Enhydra has a convergent similarity to the seal in its elongated hind limb as the major organ of propulsion. The primary result of this study is that, from the viewpoint of adaptation to the water, the numerous modifications of the Oli­ gocene seal correspond most closely with Cystophora. From a taxonomic viewpoint, the relationship between the Oligocene seal and other phocids is supported both osteologically and, by inference, myologically. Within the range of taxa compared herein, the Oligocene seal is closer in several characters (see Table 2) to the semiaquatic Mustelidae than to the representative of the eared seals. These characters are primitive (e.g., the lesser trochanter), indicating its derivation from land carnivores. At the same time, the Oli­ gocene seal has derived characters (for example, the much en- NUMBER 93 83 larged greater trochanter) that place it within the Phocidae rather than as an ancestor of this taxon. The morphology of these partial femora, and the functional myology interpreted from the osteology, shows that the locomotor adaptations of the Oligocene seal were much closer to the pinniped type (espe­ cially the phocids) than to that of the semiaquatic mustelids. This, along with its occurrence in marine deposits, associated with other fully marine vertebrates, suggests that the Oligocene seal was well adapted to a marine environment. Savage (1957:235) concluded that the Phocidae "evolved from a lutra-like mustelid in the Oligocene. The Otariidae and Odobenidae may have had a separate origin." The material of this Oligocene seal is not sufficient to either support or falsify the hypothesis of mustelid ancestry, but it is sufficient to con­ clude that the ancestor of Phocidae must be sought in deposits older than the late Oligocene. Literature Cited Barnes, L.G., and K. Hirota 1994. Miocene Pinnipeds of the Otariid Subfamily Allodesminae in the North Pacific Ocean: Systematics and Relationships. Island Arc, 3(4):329-360, 13 figures. Berta, A., and A.R. Wyss 1994. Pinniped Phylogeny. In A. Berta and T.A. Demere, editors, Contri­ butions in Marine Mammal Paleontology Honoring Frank C. Whit­ more, Jr., Proceedings of the San Diego Society of Natural History, 29:33-57, 6 figures, 3 tables, 2 appendices. Erickson, B.R., and G.T. Sawyer 1996. The Estuarine Crocodile Gavialosuchus carolinensis n. sp. (Cro- codylia: Eusuchia) from the Late Oligocene of South Carolina, North America. Monograph of the Science Museum of Minnesota (St. Paul), 3: 47 pages, 30 figures. Gordon, K.R. 1983. Mechanics of the Limbs of the Walrus (Odobenus rosmarus) and the California Sea Lion (Zalophus californianus). Journal of Morphol­ ogy, 175:73-90, 7 figures, 2 appendices. Hirota, K.M. 1983. Notes on the Miocene Sea-Lion Allodesmus. 128 pages. Masters thesis, Department of Geology and Mineralogy, Faculty of Sciences, Kyoto University, Japan. Howard, L.D. 1975. Muscular Anatomy of the Hind Limb of the Sea Otter (Enhydra lu- tris). Proceedings of the California Academy of Sciences. 40(12): 335-416, 59 figures. Howell, A.B. 1928. Contribution to the Comparative Anatomy of the Eared and Earless Seals (Genera Zalophus and Phoca). Proceedings of the United Stales National Museum, 73(15): 1—42, 1 plate. 1930. Aquatic Mammals: Their Adaptations to Life in the Water, vii + 338 pages, 54 figures. Springfield, Illinois, and Baltimore, Maryland: Charles C. Thomas. Koretsky, I.A. 1987. [Sexual Dimorphism in the Structure of the Humerus and Femur of Monachopsispontica (Pinnipedia: Phocinae).] Vestnik Zoologii, 4: 77-82. [In Russian.] Lyon, G.M. 1937. Pinnipeds and a Sea Otter from the Point Mugu Shell Mound of Cal­ ifornia. Publications of the University of California at Los Angeles in Biological Sciences, 1 (8): 133—168, 11 figures. Mitchell, E.D. 1966. The Miocene Pinniped Allodesmus. University of California Publi­ cations in Geological Sciences, 61:1 —46, 29 plates. Mori, Masaru 1958. The Skeleton and Musculature of Zalophus. Folia Anatomica Japonica, 31(3-4):203-284, 54 figures, 4 plates. Pierard, Jean 1971. Osteology and Myology of the Weddell Seal Leptonychotes wed- delli (Lesson, 1826). In W.H. Burt, editor, Antarctic Pinnipedia. Ant­ arctic Research Series, 18:53-108, 59 figures. Sanders, A.E. 1980. Excavation of Oligocene Marine Fossil Beds near Charleston, South Carolina. National Geographic Society Research Reports, 12: 601-621. Sanders, A.E., and L.G. Barnes 2002. Paleontology of the Late Oligocene Ashley and Chandler Bridge Formations of South Carolina, 2: Micromysticetus rothauseni, a Primitive Cetotheriid Mysticete (Mammalia: Cetacea). In R. Emry, editor, Cenozoic Mammals of Land and Sea: Tributes to the Career of Clayton E. Ray Smithsonian Contributions to Paleobiology, 93:271-293. Sanders, A.E., R.E. Weems, and E.M. Lemon, Jr. 1982. Chandler Bridge Formation: A New Oligocene Stratigraphic Unit in the Lower Coastal Plain of South Carolina. U.S. Geological Sur\>ey Bulletin. 1529-H:105-124, figures 24-27. Savage, R.J.G. 1957. The Anatomy of Potamotherium, an Oligocene Lutrine. Proceed­ ings of the Zoological Society of London, 129(2): 151-244, 38 fig­ ures, 3 plates, 8 tables. Sokolov, 1.1., A.S. Sokolov, and E.A. Klebanov 1974. [The Morphological Specialization of the Organs of Movement of Some Mustelidae, in Connection with Their Life-Style.] In [Func­ tional Morphology of the Mammals.] Trudy Zoologicheskogo Insti­ tuta, Akademii Nauk SSSR (Leningrad), 54:4-103, 24 figures, 13 tables. [In Russian.] Tarasoff, F.J. 1972. Comparative Aspect of the Hind Limbs of the River Otter, Sea Otter and Seals. In R.J. Harrison, editor, Functional Anatomy of Marine Mammals, pages 333-359. London and New York: Academic Press. Simocetus rayi (Odontoceti: Simocetidae, New Family): A Bizarre New Archaic Oligocene Dolphin from the Eastern North Pacific R. Ewan Fordyce ABSTRACT Simocetus rayi (new genus, new species) is based upon a skull and mandible of a small archaic dolphin (Cetacea: Odontoceti) from the upper Oligocene Alsea Formation of Oregon, bordering the northeast Pacific. The species shows many primitive features reminiscent of the archaic odontocete family Agorophiidae: the cheek teeth appear nonpolydont, the nares and premaxillary sac fossae lie anteriorly, the orbit and facial fossa are elevated above the level of the rostrum, the ascending processes of premaxillae are narrow and long, the supraorbital processes of the maxillae are narrow, the intertemporal constriction is prominent, and the ptery­ goid sinus fossae are restricted to the basicranium. These features are consistent with a basal position among the odontocetes, but they do not justify placement in the paraphyletic- and probably polyphyletic-grade family Agorophiidae. Simocetus rayi shows some unusual autapomorphies (toothless premaxillae, anterior of rostrum and mandible downturned) that exclude it from described taxa of odontocetes, and for this reason it is placed in a new and currently monotypic family, Simocetidae. Broader relationships are uncertain; some cranial features hint at affinities with Eurhino- delphinidae. For now, S. rayi is regarded as a specialized archaic odon-tocete that lies stemward (more basal) to all extant groups of Odontoceti (namely, Physeteroidea, Ziphiidae, Platanistoidea, and Delphinida). Simocetus rayi was perhaps a bottom feeder that preyed through suction feeding on soft-bodied invertebrates. The inferred pres­ ence of nasal turbinals and a vomeronasal organ contrasts with the situation in living odontocetes. Features of the face and basicra­ nium point to echolocation abilities comparable to those of extant Odontoceti. Simocetus rayi and other contemporaneous archaic odontocetes from Oregon and Washington indicate that odonto­ cetes were taxonomically and ecologically diverse by the late Oli­ gocene. R. Ewan Fordyce, Department of Geology, University of Otago, Dune­ din, New Zealand; and Research Associate, Department of Vertebrate Zoology, National Museum of Natural History, Smithsonian Institu­ tion, Washington, D.C. 20560-0121, United States. Introduction Odontocetes, which include toothed whales, dolphins, and porpoises, have an extensive fossil record from the late Oli­ gocene to Holocene. In the last decade, sound progress has been made in understanding the relationships between many living groups, particularly those with a Miocene to Holocene record (Muizon, 1987, 1988a, 1988b, 1991). In contrast, the morphology and relationships of archaic odontocetes—hetero- dont dolphins that retain an intertemporal constriction—are un­ derstood poorly. The so-called primitive odontocetes have nonetheless been pivotal in helping develop basic concepts of odontocete evolution. Indeed, one species, Agorophius pyg- maeus (Muller, 1849), is the basis for the family Agorophiidae Abel, 1913, from which, according to many authors, later od­ ontocetes evolved. This article describes and discusses a new species and new genus of archaic odontocete that has been mentioned previously (Muizon, 1991:303) as a species of Ago­ rophiidae, but that is herein placed in a new group, Si­ mocetidae. The fossil, USNM 256517, shows many details of cranial structure not recorded previously for archaic odonto­ cetes. It is the first archaic odontocete from the northeastern Pacific margin to be described formally. Although the species shows a wide range of primitive features, consistent with its late Oligocene age and rather basal position in the Odontoceti, it is too specialized, particularly in terms of feeding apparatus, to have been directly ancestral to any other described odonto­ cete. Specimen USNM 256517 is one of several hundred fossil vertebrates assembled by the late Douglas R. Emlong, which now constitute the Emlong collection (Ray, 1977, 1980; Dom­ ning et al., 1986) in the National Museum of Natural History. The Emlong collection comprises fossils from a thin sequence of marine Oligocene and Neogene rocks along the eastern coast of the North Pacific Ocean, mainly from Oregon and Washing­ ton. Other unnamed species of archaic odontocete are repre- 185 186 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY sented by undescribed specimens in the Emlong collection; mentioned below are USNM 205491 ("non-squalodontid odon­ tocete" of Whitmore and Sanders, 1977, fig. 2b, from Alsea Formation, Oregon), USNM 243979 ("non-squalodontid odon­ tocete" of Whitmore and Sanders, 1977, fig. 2a, from ?Pysht Formation, Washington), and USNM 299482 (?Pysht Forma­ tion, Washington), all of which are of Oligocene age. Descriptions below are based upon the right or left side, whichever is more nearly complete. Some comparisons are of­ fered in the descriptive text to clarify homologies, but fuller systematic comparisons follow the description. Absences are noted for some cranial features that are present in archaeocete cetaceans but are enigmatic or missing in most odontocetes. Unreferenced statements about cetacean anatomy are based upon personal observations. Nomenclature follows that used by Kellogg (1936), Kasuya (1973), Fordyce (1994), and others, with modifications and synonyms (following Sisson and Grossman, 1953; Davis, 1964; Novacek, 1986; Evans, 1993) to better identify cetacean homologs with those of other mam­ mals. The few postcranial elements are rather uninformative and are not described. Line diagrams are from 35 mm photographs (Asahi Pentax camera, 50 mm macro lens) and are not corrected for parallax. Illustrations are by the author. F.V. Grady carried out most of the preparation using mechan­ ical tools. The author prepared fine details under a low-power binocular microscope using pneumatic scribe and an air-abra­ sive unit. Much of the matrix is a leached (formerly calcareous) mudstone that bonds tightly to bone, and still obscures some sutures and foramina. ABBREVIATIONS.—The following abbreviations are used: AMNH Departments of Mammalogy and of Vertebrate Paleontology, American Museum of Natural History, New York MCZ Museum of Comparative Zoology, Harvard University, Cam­ bridge, Massachusetts MNHN Museum Nationale d'Histoire Naturelle, Paris OU Geology Museum, University of Otago, Dunedin, New Zealand USGS United States Geological Survey USNM Collections of the National Museum of Natural History, Smith­ sonian Institution, Washington, D .C , including those of the former United States National Museum ACKNOWLEDGMENTS.—Work was started during the tenure of a Smithsonian Postdoctoral Fellowship in the Department of Paleobiology from early 1979 to mid-1980 and was continued with support from Smithsonian short-term visiting fellowships. The University of Otago supported visits to the Smithsonian In­ stitution in 1984, 1988-1989, 1994, and 1996.1 am indebted to Frank C. Whitmore, Jr., for his valuable guidance and advice over the years. I gratefully thank Clayton E. Ray for facilitating diverse aspects of this study, and Frederick V. Grady for prepa­ ration. Lawrence G. Barnes, Hiroto Ichishima, Richard Kohler, James G. Mead, Mark D. Uhen, and Frank C. Whitmore, Jr., re­ viewed different versions of the manuscript. Barnes, Mead, Whitmore, and Christian de Muizon also provided fruitful on­ going discussion of cetacean systematics. Lawrence G. Barnes, David J. Bohaska, Andrew P. Currant, John T. Darby, Anton van Helden, Jerry J. Hooker, James G. Mead, Clayton E. Ray, Tho­ mas H. Rich, Richard H. Tedford, and Albert E. Sanders kindly arranged access to specimens studied as part of this work. Class MAMMALIA Order CETACEA Suborder ODONTOCETI DIAGNOSIS.—Odontoceti are recognized herein as Cetacea with the following combination of osteological features: su­ praorbital process of maxilla telescoped posteriorly over fron­ tal; one or more dorsal infraorbital (=maxillary) foramina open posteriorly in supraorbital process of maxilla; premaxillary sac fossa, premaxillary sulci, and premaxillary foramen (or foram­ ina) present; infraorbital process of maxilla reduced; posterior- most teeth inserted anterior to antorbital notch; orbitotemporal crest displaced posterodorsally relative to postorbital ridge and partly roofing temporal fossa; distinct antorbital notch faces anteriorly to anterolaterally; pterygoid hamulus partly exca­ vated for the pterygoid sinus system; ossified mesethmoid present between bony nares; and parietal and squamosal dor- sointernally override periotic to occlude the cranial hiatus. REMARKS.—These features are seen in or inferred for S. rayi. They are modified in some other odontocetes. Many of the facial features listed appear to be linked to the presence of a posteriorly displaced and hypertrophied maxillo-naso-labialis (nasofacial) muscular complex, and thus to the ability to gener­ ate high-frequency sound that might be used in echolocation. See "Discussion," below. Superfamily Incertae Sedis SIMOCETIDAE, new family DIAGNOSIS.—As for Simocetus rayi, at present the only in­ cluded species; see remarks under species diagnosis. TYPE GENUS.—Simocetus new genus. INCLUDED GENERA.—Simocetus new genus only. Simocetus, new genus ETYMOLOGY.—From simus, pug-nosed, and cetus, whale (Greek). In allusion to the prominent medial "snout" posterior to the external nares. DIAGNOSIS.—As for Simocetus rayi, at present the only in­ cluded species; see remarks under species diagnosis. TYPE SPECIES.—Simocetus rayi new species. INCLUDED SPECIES.—Simocetus rayi only. Simocetus rayi, new species Agorophius sp., Muizon, 1991:303. HYPODIGM.—USNM 256517 (Emlong collection number NUMBER 93 187 FIGURE I.—Locality map of Pacific Northwest. The detailed map is based upon Snavely et al. (1975, 1976). E77-32), holotype, only. Almost-complete skull, right periotic in place on skull, incomplete right mandible, three anterior teeth, 10 cheek teeth (four in place in rostrum, four in place in mandible), two incomplete presumed lumbar vertebrae, one chevron, eight or more incomplete ribs. Collected by D.R. Em­ long, 3 August 1977. DIAGNOSIS.—Small odontocete. More primitive than de­ scribed species of Squalodontidae, Waipatiidae, Squalodel- phinidae, Platanistidae, Dalpiazinidae, Eoplatanistidae, Eurhin- odelphinidae, Lipotidae, Pontoporiidae, Iniidae, Albireonidae, Kentriodontidae, Delphinidae, Phocoenidae, Monodontidae, Ziphiidae, Physeteridae, and Kogiidae. Primitive features as follows: premaxillary sac fossa located well forward on ros­ trum; lateral margin of rostrum (rostral base) located well ven­ tral to level of elevated roof of orbit; rostrum with well-devel­ oped fossa for nasofrontal muscles; ventral infraorbital foramen and sphenopalatine foramen opening in a large in- fundibulum in anterior wall of orbit; external nares open well forward of level of antorbital notch; median facial elements (nasals plus frontals) form an elongate "snout"; multiple su­ praorbital foramina open on frontal; elongate palatine forms much of palate; and parietals and interparietal form elongate intertemporal constriction. Similar to Agorophius pygmaeus (Agorophiidae) only in ple­ siomorphic features including heterodont teeth; premaxillary sac fossa and associated foramina and sulci lying well forward on rostrum; lateral margin of rostrum (rostral base) lying ven­ tral to level of elevated roof of orbit; and parietals contributing to an elongate intertemporal constriction. Differs from Ago­ rophius pygmaeus in derived features as follows: rostrum more dorsoventrally compressed; supraorbital process of maxilla not expanded far laterally; braincase more inflated at base of zygo­ matic process, with deep cleft between process and braincase; supraoccipital more hemispherical; and cheek teeth smaller, asymmetrical, and with diastemata. Similar to Eurhinodelphinidae in having toothless premax­ illa, elongate conical pterygoid hamulus, and thick postglenoid process. Similar to some extant Delphinidae in having well-de­ veloped interparietal and unfused mandibular symphysis. Simi­ lar to Waipatiidae and Squalodelphinidae in having pronounced cleft, presumably representing foramen spinosum, arising near foramen ovale and trending posterolaterally toward periotic along or near parieto-squamosal suture. Similar to some Del­ phinidae and Squalodontidae in having palate with scattered multiple small palatine foramina. Presumed autapomorphies include thin spatulate edentulous rostral apex, formed by premaxillae, deflected ventrally, result­ ing in down-turned rostrum; relatively short rostrum; supraor­ bital process of maxilla not expanded laterally over supraor­ bital process of frontal; deep, narrow optic infundibulum; complex fine sutures between alisphenoid and squamosal; pos­ terior maxillary cheek teeth occluding into facing diastemata between mandibular cheek teeth; and vestigial lower II. ETYMOLOGY.—After Clayton E. Ray, in recognition of his enduring and influential contributions to the study of fossil ma­ rine mammals. TYPE LOCALITY.—Near intersection of North Yaquina Bay Road and Toledo Road, Lincoln County, Oregon, USA. Grid reference: SW 1/4, NE 1/4, Section 18, T US, R 10W on To­ ledo 15-minute quadrangle map (USGS) or about 44°37'06"N, 123°56'50"W. See Figure 1, and Snavely et al. (1976). HORIZON.—Gray calcareous to brown leached massive mud- stone of the Alsea Formation of Snavely et al. (1975). The specimen came from the most eastern exposure of Alsea For­ mation mapped by Snavely et al. (1975, 1976), near Toledo, Oregon. 188 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY TABLE 1.—Skull Measurements (in mm) of the standard skull measurements used by Perrin (1975), as well as measurements more appropriate for archaic odontocetes. Condylobasal length, from tip of rostrum to posterior of occipital condyles Length of rostrum, from tip to line across posterior limits of antorbital notches (right and left notch, respectively) Width of rostrum at base, along line across posterior limits of antorbital notches Width of rostrum at 60 mm anterior to line across posterior limits of antorbital notches Width of rostrum at mid-length, at -120 mm anterior to line across posterior limits of antorbital notches Width of premaxillae at mid-length of rostrum Width of rostrum at 3/4 length, measured from posterior end Distance from tip of rostrum to posterior of prenarial constriction Distance from tip of rostrum to anterior of bony naris Distance from tip of rostrum to most anterior portion of external wall of choana, immediately dorsal to ptery­ goid hamulus Distance from tip of rostrum to broken tip of pterygoid hamulus Preorbital width, at level of frontal-lacrimal suture Supraorbital width, at middle of orbit Postorbital width, across apices of postorbital processes Depth from vertex to roof of orbit Maximum width across narial aperture Maximum width across zygomatic processes of squamosals (width from right process to midline) Maximum width of premaxillae Maximum width between posterolateral sulci Distance from antorbital notch to apex of ascending process of right premaxilla Distance from antorbital notch to apex of supraorbital process of right maxilla Maximum distance between posterior apices of supraorbital processes of maxillae Distance from midline of right internal ventral infraorbital foramen Depth from vertex of right internal ventral infraorbital foramen Median length of nasals on vertex Maximum length of right nasal on vertex Distance anterior to antorbital notch of anterior border of nasals Median length of frontals on vertex Distance behind antorbital notch of anterior of frontals on vertex Median length of parietals/interparietals on vertex Minimum parietal/interparietal width, at level of temporal lines Parietal width, at dorsoventral midpoint of intertemporal region (35 mm ventral to vertex) and halfway along intertemporal region Distance from posterior of occipital condyle to anterior apex of supraoccipital Vertical external height of braincase, from midline of basioccipital to dorsal extremity of supraoccipital Maximum length of right temporal fossa, point to point from most anterior portion of orbitotemporal crest on frontal to most posterior boundary of fossa Minimum length of right temporal fossa, point to point from postorbital ridge on frontal to most anterior por­ tion of base of zygomatic process Maximum dorsoventral depth of right temporal fossa, from dorsal surface of intertemporal region to temporal angle Maximum transverse width, from apex of right zygomatic process to braincase wall Projection of premaxillae beyond maxillae, measured from tip of rostrum to line across anteriormost tips of maxillae in dorsal view Distance from posteriormost end of median suture between nasals to anteriormost apex of supraoccipital Length of right orbit, from anterior apex of orbit to ventral apex of postorbital process Median length of palatines Maximum transverse width across palatines Maximum length of left pterygoid Distance from tip of hamulus of left pterygoid to posteriormost portion of occipital condyle Maximum width of choanae Distance from most anterior portion of wall of left choana to most posterior portion of occipital condyle Distance from hindmost portion of vomer to hindmost portion of occipital condyle Width across paroccipital processes (width from right process to midline) Distance from most anterior portion of right pterygoid sinus fossa to most posterior portion of occipital condyles Length of upper right tooth-row, from hindmost margin of most posterior alveolus to tip of rostrum Distance from upper right canine to tip of rostrum Distance from position of posterior tooth to antorbital notch 449+ 235; 241 142 112 75 est. 61 51 +(est. 57) -198 -203 340 361 173 183 207 41 31 est. 238 119 89 71 63 80 125 36 19 23 + 46 -18 -81 86 -39 36 82 99 107 125 64 55 -24 139 51 72 91 67 + 78 + est. 48 105 49 est. 216 108 -143 206 est. 61 59 NUMBER 93 189 TABLE 2.—Measurements of periotic. Not all the standard measurements suggested by Kasuya (1973) for odon­ tocete periotics could be made. Measurements for the right periotic are to the nearest 0.5 mm. Maximum anteroposterior length of exposed portion of periotic, from anterior apex of anterior process to level of posterior of pars cochlearis Maximum anteroposterior length of pars cochlearis Maximum transverse width of pars cochlearis, from internal edge to fenestra ovalis Maximum dorsoventral depth of anterior process Length of anterior process, from anterior apex of anterior process to level of anterior of pars cochlearis Maximum transverse width of periotic, internal face of pars cochlearis to apex of lateral tuberosity Maximum width of anterior process at base 37.0 23.0 -9.5 -16.5 14.0 21.0 13.0 AGE AND CORRELATION.—Upper Zemorrian Stage, late Oli­ gocene. The Alsea Formation at the locality of USNM 256517 is faulted against the underlying upper Eocene Nestucca For­ mation, so that stratigraphic position of the specimen relative to the base of the Alsea Formation cannot be determined. Ellen J. Moore (USGS, pers comm., 1981) stated that foraminifera col­ lected about 200 m south of the odontocete locality indicate an upper Zemorrian or late Oligocene age, and that molluscs col­ lected about 300 m northwest of the odontocete locality indi­ cate "Blakeley" Stage or late Oligocene. Domning et al. (1986) gave more details of the stratigraphy of vertebrate-bearing strata of this part of Oregon. Absolute ages for international units have been refined since 1986, so that the age for the Alsea Formation is presumably in the range of 30-23 Ma. DESCRIPTION.—General Features of Skull (Figures 2-17; measurements in Tables 1-3): The skull is telescoped, with a plate-like supraorbital process of maxilla present dorsal to the supraorbital process of the frontal, the supraoccipital is thrust forward, and the short parietal is present in a distinct intertem­ poral constriction. Teeth are heterodont. Apart from the apex of the rostrum, there is little obvious postmortem distortion, and the asymmetry at the base of the rostrum appears to be real. Cranium: The cranial portion of the skull, behind the an­ torbital notches, is long relative to the condylobasal length (Figure 2A). In lateral view (Figure 8A,C), the dorsal profile of the cranium is roughly straight, without an obvious raised ver­ tex, and the dorsal and ventral profiles are more or less parallel. Viewed thus, the orbit clearly lies dorsal to the level of the lat­ eral border of the rostral part of the maxilla (base of rostrum), the remnant of infraorbital process of maxilla, and the jugal. Presumably, the nasofacial (maxillo-naso-labialis) muscle orig­ inated from the maxilla both on the rostrum and dorsal to the orbit. In dorsal view (Figures 2A, 3), the posterior limit of the origin of nasofacial muscles is marked by the ridge that runs from the apex of each postorbital process smoothly medially and dorsally onto the fronto-parietal suture. Seen thus, the pro­ file of the posterior of the face is anteriorly concave. In dorsal view (Figures 2A, 3), the maxilla and frontal cover little of the temporal fossa, and the intertemporal constriction, formed by parietal and interparietal, is prominent. A long snout of nasal and frontal roofs the nares and olfactory cavity. A prominent orbitotemporal crest (temporal crest of Fordyce, 1994) delimits the dorsal edge of the fossa. Within the fossa, the lateral wall of the braincase is not particularly inflated ex­ cept toward the base of the zygomatic process of the squamo­ sal. In lateral view (Figures 8A,C, 9), the external nares open more than 20 mm anterior to the level of the antorbital notch. The internal nares (choanae) open level with the apex of the su­ praoccipital and the middle of the pterygoid sinus fossa, indi­ cating inclined rather than subvertical narial passages. Antero­ posterior separation between internal and external nares also is emphasized by the long ventral exposure of the palatines (Fig­ ures 2D, 4). Rostrum: The triangular rostrum is short, about as long as the cranium, and is broad at the base. The apex is blunt, with its apical 10 mm formed only by premaxillae (Figures 2A, 3, 5c) that are dorsoventrally thin and, individually, spatulate. Farther posteriorly, the rostral edges are straight until 20-35 mm ante­ rior to the dorsoventrally deep antorbital notches. Here, at point of maximum width, the rostrum is asymmetrical, with the in­ complete right side flared out laterally more than the complete left side. The antorbital notches, which appear wide and shal­ low in dorsal view, are asymmetrical, with the left notch more open than the right. In lateral view (Figures 8A,C, 9), the ante­ rior half of the rostrum is thin, but rostral depth increases mark­ edly toward the cranium as the premaxillae and maxillae be­ come more elevated, and the rostrum is steep sided behind the external nares. As viewed laterally (Figure 8A) , the ventral profile of the rostrum is concave, with its tip depressed relative to the basic­ ranial axis. A postmortem break distorts the tip dorsally, but the mandible (Figure 8B,D) indicates the original profile of the ros­ trum. In transverse profile, the maxillary part of the palate is flat to slightly convex (Figure 8E), but for a slightly concave re­ gion posterolaterally. TABLE 3.—Measurements of mandible. Standard measurements of the right mandible, after Perrin (1975) are to the nearest 1 mm. Tooth-row length, from posterior margin of most posterior alveolus to tip of mandible Maximum length of mandible Maximum height of mandible, perpendicular to maximum length 221 334+ (est. 380) 90 + 190 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY FIGURE 2.—Simocetus rayi, USNM 256517, holotype skull: A, dorsal; B, ventral; C, anterior; D, posterior. (Scale bar=10 cm.) A prominent mesorostral groove dominates the rostrum ante- Farther posteriorly, the groove is narrow, deep, parallel-sided, rior to the nasals. The groove is wide and shallow anteriorly, and U-shaped, with a minimum width at an indistinct prenarial but it deepens and narrows behind the premaxillary foramina to constriction about 35 mm in front of the nasals (Figures 2A, 3). become about as wide as deep at the mid-length of the rostrum. (The prenarial constriction was inadvertently termed internarial NUMBER 93 191 10 cm major breaks or areas of matrix mesorostral groove anterior premaxillary foramen inferred fossa for nasal plug muscle posterior premaxillary foramen anteriorly-placed dorsal infraorbital foramen cluster of dorsal infraorbital foramina antorbital notch posterior dorsal infraorbital foramina ascending process of premaxilla orbitotemporal crest supraorbital foramina parietal foramina parasagittal crest nuchal crest external occipital crest premaxilla maxilla premaxillary sac fossa median premaxillary cleft posterolateral sulcus nasal ascending process of premaxilla postorbital process temporal fossa subtemporal crest zygomatic process of squamosal parietal interparietal supraoccipital — position of external acoustic meatus paroccipital process dorsal condyloid fossa occipital condyle FIGURE 3.—Simocetus rayi, USNM 256517, dorsal view of holotype skull, based upon Figure 2A, showing main features. (Scale bar=10 cm.) constriction by Fordyce (1994:149.) Premaxilla: Three regions of the premaxilla command at­ tention: the rostral apex, the premaxillary sac fossa, and the portion at and posterior to the nares. Most of the premaxilla lies on the rostrum, anterior to the antorbital notch. Anteriorly, the premaxillae lie close together for their apical 10 mm, separated by only a few millimeters at the open incisive suture (Figures 2A,B, 3, 4, 5c). Each premaxilla forms the thin spatulate ante­ rior about 27 mm of the dorsal surface and margin of the ros­ trum. Apically, the well-preserved ventral (or palatal) surface of the premaxilla lacks alveoli and is interpreted as edentulous (Figure 5c). Ventrally, about 5 mm from the medial edge, each premaxilla carries a long, shallow, narrow groove that origi­ nates posteriorly at a feature presumed to be the palatine fis­ sure. This fissure lies at the junction of premaxilla, maxilla, and vomer, about 58 mm posterior to the rostral apex. (For­ merly, it was suggested that the palatine fissure is absent in Ce­ tacea; Kellogg, 1936.) Anteriorly, the groove ends at a shallow depression presumed to be for the vomeronasal (Jacobson's) organ, about 8 mm behind the apex of the rostrum (Figure 5c). Dorsally, the premaxillary sac fossa (premaxillary plate or spiracular plate of some authors), associated sulci, and foram­ ina dominate the premaxilla. The premaxillary sac fossa is the long, elevated, and rather tabular dorsal surface of the premax­ illa bounded by the posterolateral sulcus and the mesorostral groove (Figures 2c, 6). In living odontocetes, this surface car­ ries a lobe of the nasal passages, the premaxillary sac (Mead, 1975). The fossa is narrow and pointed anteriorly. Its depressed lateral edge between 115 mm and 140 mm from the rostral apex perhaps marks the origin for the nasal plug muscle, al- 192 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY premaxilla depression for ?vomeronsal organ position of canine (alveolus not visible) palatine fissure 10 cm major breaks or areas of matrix palatine sulcus major palatine foramen antorbital notch preorbital ridge orbit postorbital ridge orbitotemporal crest pterygoid - temporal fossa zygomatic process of squamosal subtemporal crest path of mandibular nerve alisphenoid falciform process postglenoid process periotic external auditory meatus posterior process of tympanic bulla inferred path of facial nerve paroccipital process articulation for stylohyal groove for ?posterior sinus major palatine foramen jugal lacrimal remnant infraorbital process of maxilla frontal foramina postorbital process of frontal palatal crest vomer internal naris pterygoid sinus fossa pterygoid hamulus horizontal plate of vomer foramen ovale facet for contact with pterygoid ventral condyloid fossa occipital condyle basioccipital crest hypoglossal foramen parietal FIGURE 4.—Simocetus rayi. USNM 256517, ventral view of holotype skull, based upon Figure 2B, showing main features. (Scale bar= 10 cm.) though the bone here is smooth rather than rough. Farther pos­ teriorly, the premaxillary sac fossa widens as its tabular dorsal surface rises above the level of the rest of the rostrum. The fossa reaches a width of about 25 mm level with the indistinct prenarial constriction, then narrows abruptly as the premaxilla rises toward the nasal. Immediately lateral to the anterior part of the premaxillary sac fossa are two large premaxillary foramina, inferred to trans­ mit branches of the internal maxillary artery and associated nerves. The anterior premaxillary foramen opens within the premaxilla (seen on the right, obscured on the left) into an an- teromedial sulcus (sensu Barnes, 1978) on the dorsal surface, about 95 mm behind the apex of the rostrum and just lateral to the mesorostral groove. The posterior premaxillary foramen NUMBER 93 193 FIGURE 5.—Simocetus rayi, USNM 256517, and unnamed odontocetes USNM 205491 (Oligocene, Oregon) and USNM 299482 (Oligocene, Washington): A-C, whitened with ammonium chloride; A, holotype skull of S. rayi, USNM 256517, right basicranium; B, skull of unnamed odontocete USNM 205491 (Oligocene, Oregon), for comparison with Simocetus rayi; C, holotype skull of S. rayi, USNM 256517, detail of ventral surface of rostrum; D,E, incomplete skull of unnamed odontocete USNM 299482 (Oligocene, Oregon); D, dorsal view, showing elongate nasals over snout; E, posterior view of eroded section through olfactory cavity showing turbinals within olfactory cavity. (Scale bar= 10 cm.) (best preserved on the left) opens lateral and slightly ventral to the apex of the premaxillary sac fossa, 135-150 mm from the rostral apex (Figures 2A, 3). Posteriorly, the posterolateral sul­ cus deepens and narrows as it rises dorsomedially toward the nasals; it can be traced about 25 mm behind the level of the ex­ ternal nares. The posteromedian sulcus is possibly represented by an indistinct, wide, shallow groove that originates laterally near the apex of the fossa and trends posteromedially toward the elevated internal edge of the premaxilla. Another fine sulcus, interpreted as the median premaxillary 194 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY nuchal crest ascending process of premaxilla ascending process of maxilla orbitotemporal crest groove for ?facial nerve postorbital process of frontal orbit postglenoid process of squamosal antorbital notch anterior premaxillary foramen anteromedial sulcus premaxilla nasal mesethmoid external naris posterior dorsal infraorbital foramina posterolateral sulcus median premaxillary cleft supraorbital process of frontal lacrimal dorsal infraorbital foramina posterior premaxillary foramen maxilla premaxillary sac fossa mesorostral groove 10 cm | major breaks or areas of matrix FIGURE 6.—Simocetus rayi, USNM 256517, anterior view of holotype skull, based upon Figure 2c, showing main features. (Scale bar= 10 cm.) cleft of Fordyce (1994:149), appears as a fissure in the other­ wise smooth dorsal surface of the premaxillary sac fossa. In S. rayi, the anterior origin of the median premaxillary cleft is ob­ scure, but in USNM 299482 (an undescribed archaic odonto­ cete; Figure 5D), the median premaxillary cleft appears to orig­ inate at the posterior premaxillary foramen. It meanders posteriorly then diverges abruptly toward the internal edge of the premaxillary sac fossa at about the mid-length of the fossa; thence it trends farther posteriorly and slightly externally to contact the posterolateral sulcus at about the level of the ante­ rior end of the nasal. Broken sections through the median pre­ maxillary cleft in USNM 299482 indicate that it extends far ventrally within the premaxilla, and the same is inferred for S. rayi. The function of this sulcus is uncertain. Lateral to the premaxillary sac fossa and the posterolateral sulcus are remnants of a thin, elongate flange of premaxilla 10 cm nasal nuchal crest external occipital crest occipital condyle zygomatic process of squamosal parietal (broken section) margin of foramen ovale foramen magnum pterygoid hamulus paroccipital process basioccipital crest- (dorsally-crushed) jugular notch FIGURE 7.—Simocetus rayi, USNM 256517, posterior view of holotype skull, based upon Figure 2D, showing main features. (Scale bar= 10 cm.) NUMBER 93 195 FIGURE 8.—Simocetus rayi, USNM 256517, holotype skull and right mandible: A, skull, right lateral; B, right mandible, lateral (buccal) view; c, skull, left lateral view; D, right mandible, medial (lingual) view; E, skull, oblique ventrolateral view, left side. (Scale bar= 10 cm.) that, before being eroded, originally overlay the dorsal surface plate; the plate is conspicuous in lateral view (e.g., Figure 9). of the maxilla (Figure 2A,C). Posteriorly, at the level of the na- Dorsally, the posterolateral plate passes into the posteromedial sal, this thin portion of premaxilla becomes the posterolateral process. The left and right posteromedial processes are sym- 196 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY interparietal frontal parietal border of optic infundibulum squamosal dorsal condyloid fossa occipital condyle fossa for neck muscles paroccipital process- post-tympanic process- posteromedial process of premaxilla Jposition of external nares ^- posterolateral plate of premaxilla posterolateral sulcus posterior premaxillary foramen maxilla premaxilla- "facet for jugal pterygoid hamulus falciform process postglenoid process external auditory meatus major breaks or areas of matrix FIGURE 9.—Simocetus rayi, USNM 256517, right lateral view of holotype skull, based upon Figure 8A, showing main features. (Scale bar= 10 cm.) metrical; each is narrow (<2 mm wide), parallel sided, and long, extending far posteriorly between the maxilla and the na­ sal-frontal. The right process reaches about 95 mm behind the nares, well posterior to the naso-frontal suture and the orbit (Figures 2A, 3, 9). Maxilla: Ventrally, the maxilla forms most of the rostrum, including the massive thick dorsoventrally rounded lateral edge, which extends to within about 25 mm of the rostral apex. Anterior teeth lie at the edge of the maxilla, but the posterior teeth lie a little medially (Figures 2B, 4). There are alveoli for a canine (matrix-filled with eroded borders, on the rostral margin of the maxilla and not readily seen in figures) and six cheek teeth, identified as Pl—4 and Ml -2 ; teeth are detailed below. Anteriorly, the alveoli on the rostral margin are too eroded to be sure of orientations and fine structure, but posterior alveoli are separated by diastemata (interdental spaces) long enough to accommodate crowns of lower cheek teeth. An indistinct em­ brasure pit lies between and slightly medial to right P4 and Ml (Figure 2B) . M2 lies about 35 mm anterior to the antorbital notch (Figures 2B, 4). There are no obvious alveolar juga, and the alveolar process is not distinct from the rest of the maxilla. Most of the rostral (oral, ventral) surface (palatine process) of the maxilla is gently convex in transverse profile, but, poste­ rolaterally near the antorbital notch, the surface is concave coronoid crest mental foramen symphysis FIGURE 10.—Simocetus rayi, USNM 256517, right mandible, showing main features: A, lateral (buccal), based upon Figure 8A; B, medial (lingual), based upon Figure 8C. (Scale bar=10 cm.) NUMBER 93 197 • depression for ?vomeronasal organ squamosal supraoccipital parietal alisphenoid foramen ovale carotid foramen remnant inner lamina of pterygoid occipital condyle ventral condyloid fossa basioccipital crest jugular notch paroccipital process posterior process of bulla external auditory meatus postglenoid process FIGURE 11.—Simocetus rayi, USNM 256517, oblique ventrolateral view of holotype skull, based upon Figure 8E, showing main features. (Scale bar=10 cm.) (Figures 2B, 8E). Nearby, the conspicuous and simple maxil­ lary-palatine suture (Figure 2B) traces an S-shaped profile as it runs laterally from the midline. A short and narrow major (or greater) palatine sulcus at the maxillary-palatine suture (Figure 4) indicates a major palatine foramen. Farther anteriorly, an­ other aperture of the major palatine foramen opens into a pa­ latine sulcus medially, about level with P3 and P4. Smaller fo­ ramina that open ventrally on the maxilla probably represent other multiple openings for the palatine foramina. Posteriorly, the maxilla has only a small orbital surface. Here, a prominent transverse ridge of the maxilla in the orbital wall, immediately below the ventral infraorbital foramen (or­ bital opening of infraorbital canal, or maxillary foramen of au­ thors) is a vestigial infraorbital process. Because the ridge is not prolonged posteriorly, there is neither a maxillary tuberos­ ity nor a pterygopalatine fossa. Medially, a stout projection of the maxilla partly separates the ventral infraorbital foramen from the sphenopalatine foramen (Figures 11, 12). The ventral infraorbital and sphenopalatine foramina open within a common infundibulum anterior to the preorbital ridge (Figures 8E, 11-13) and topographically within the orbit. Within the infundibulum, the larger infraorbital portion lies FIGURE 12.—Simocetus rayi, USNM 256517, oblique ventrolateral view of left orbit of holotype skull. Skull whitened with ammonium chloride. (Scale bar= 10 cm.) 198 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY optic foramen frontal foramina ethmoid foramen orbitosphenoid lacrimal antorbital notch jugal ventral infraorbital foramen sphenorbital fissure path of maxillary nerve V2 foramen rotundum • scattered palatine foramina - vomer. base of pterygoid hamulus ventromedial extension of sinus fossa onto hamulus - internal naris basioccipital carotid foramen foramen ovale keel of vomer broken internal lamina of pterygoid pterygoid foramen FIGURE 13.—Simocetus rayi, USNM 256517, oblique ventrolateral view of left orbit of holotype skull, based upon Figure 12, showing main features. (Scale bar=5 cm.) medially in the anterior wall of the orbit, whereas the smaller sphenopalatine portion lies posteroventrally on the lateral wall of the skull below the orbit. The infundibulum is bounded ven­ trally by the maxilla, dorsally and anterolaterally by the frontal, and posteriorly and posteroventrally by the palatine; it is uncer­ tain if the lacrimal contributes to the lateral border. Anteriorly, the large ventral infraorbital foramen gives rise to the infraor­ bital canal from which the dorsal infraorbital foramina (maxil­ lary foramina of authors) and premaxillary foramen open on the dorsal surface of the skull. The small sphenopalatine fora­ men lies posteriorly in the roof of the infundibulum, where it is directed dorsomedially. As seen ventrally, the maxilla at the antorbital notch is in­ vaded by two small lobes of the jugal (zygomatic bone), a shorter lateral lobe and a longer medial lobe. That part of the maxilla bounded by the two lobes of the jugal may represent the zygomatic process of the maxilla, although the zygomatic process in other mammals usually lies laterally on the skull wall. There is no clear evidence in the maxilla here of an ante­ rior lacrimal crest, lacrimal notch, lacrimal sulcus, or lacrimal canal. Dorsally on the rostrum, the maxilla is rounded in transverse profile until about the level of the posterior premaxillary fora­ men, behind which its surface becomes slightly concave. Dor­ sal infraorbital foramina open on the maxilla; on the left, the anteriormost foramen opens near the posterior premaxillary fo­ ramen (150 mm behind the rostral apex), and a prominent clus­ ter of foramina lies farther posteriorly near the antorbital notch. Because of poor surface detail on the right side, it is not clear whether the number and size of foramina is symmetrical from left to right, although the two maxillae clearly are asymmetrical in profile. These large, multiple, dorsal infraorbital foramina are taken to indicate that parts of the nasofacial muscles origi­ nated on the rostrum anterior to the antorbital notch. The lacrimal bisects the maxilla at the antorbital notch, sepa­ rating the rostral from the more-posteromedial facial parts of the maxilla (Figures 2c, 6). At the base of the supraorbital pro­ cess, the maxilla does not expand laterally over the lacrimal; thus, it lacks an antorbital process. NUMBER 93 199 Posterior to the lacrimal, the maxilla forms a long, roughly parallel-sided supraorbital process (ascending process, frontal process) that extends across the frontal and rises medially to­ ward the vertex (Figures 2A, 3). Here, the maxillary surface is slightly concave from side to side, forming an indistinct facial fossa for nasofrontal muscles. Seen dorsally, the supraorbital process has a lateral margin that is gently concave rather than convex. Further, the supraorbital process is not expanded out­ ward over the frontal. Because the left and right supraorbital processes are symmetrical, and there is no evidence of erosion or abnormal bone surface on maxilla and frontal, the lack of lateral expansion is taken as an original condition. Two poste­ rior dorsal infraorbital foramina open posterodorsally in the su­ praorbital process of each maxilla; they lie about level with the mid-orbit as viewed laterally, about 15 mm below the level of the vertex, and about 40 mm lateral to the cranial midline. Each foramen opens posteriorly into a shallow widening groove; the anterior foramen is the larger. These foramina are taken to indi­ cate that the maxilla here formed an origin for a significant vol­ ume of facial muscles. Nearby, the posterolateral corner of the supraorbital process is sharp, whereas the posteromedial corner is rounded. Posteriorly, the supraorbital process does not reach the level of the orbitotemporal crest of the frontal. By analogy with an undescribed archaic odontocete, USNM 299482 (Fig­ ure 5D), the straight and slightly raised medial border of the su­ praorbital process of maxilla probably overlies part of the pos­ teromedial process of the premaxilla. Preservation is too poor to tell if the maxilla has a nasal crest within the bony naris. Vomer: The vomer lines the prominent mesorostral groove where, in dorsal view, it reaches to within about 10 mm of the rostral apex (Figures 2A, 3). The vomer, and thus the mesoros­ tral groove, is widest (about 28 mm) about 70 mm behind the rostral apex. Ventrally, the vomer is exposed as a short thin sliver (up to 7 mm wide) anteriorly on the palate, wedged be­ tween the premaxillae anteriorly and the maxillae posteriorly. A cross section of rostrum seen during preparation indicates that each side of the vomer widens to more than 25 mm within the rostrum. Farther posteriorly, a diamond-shaped medial ex­ posure of vomer lies between the palatines and pterygoids (Figures 2B, 4); compared with most odontocetes, this expo­ sure of vomer is quite large. Posterodorsally, this diamond- shaped exposure rises into a long median keel of vomer, which separates the choanae back to about the level of the foramen ovale (Figures 8E, 11-13). The slightly outward-flared hori­ zontal plate of vomer, which roofs the choanae, extends back about 20 mm farther, about level with the anterior of the basio­ ccipital crests. Presumably, the horizontal plate of vomer orig­ inally contacted the now-missing internal lamina of the ptery­ goid here. Palatine: The broadly exposed palatine forms much of the posterior flat ventral surface (horizontal lamina) of the palate, and it is inferred to contribute to the hard palate as seen in other mammals but in contrast to living odontocetes. The prominent median interpalatine suture is about 71 mm long, and each pa­ latine is more than 84 mm long, so that the pterygoid is sepa­ rated widely from the maxilla. Anteriorly, the palatine extends well forward of the antorbital notch, whereas posteriorly the palatine reaches behind the level of the postorbital process of the frontal. Ventral and lateral faces of the palatine are sepa­ rated by a gently rounded palatal crest that arises at the in­ fundibulum for the infraorbital and sphenopalatine foramina and descends posteroventrally toward the pterygoid. There is no obvious pterygopalatine fossa. Above this palatal crest, the palatine carries a smoothly rounded surface posterior to the in­ fraorbital infundibulum and ventral to the optic infundibulum; this surface is presumed to be the origin of the internal ptery­ goid muscle. Dorsally, there is no sphenopalatine notch, so that the orbital process and spheno-ethmoid lamina are continuous; sutures with frontal and orbitosphenoid are prominent. Postero­ dorsally, a few millimeters of the palatine contacts the parietal. Posteriorly, the suture with the pterygoid is conspicuous, with­ out a lateral lamina of palatine, and the pterygoid sinus does not invade the palatine. Ventral sutures with the pterygoid and vomer are deep. Grooves at the anterolateral corner of the maxillo-palatine suture form the major palatine sulcus and associated (matrix- filled) major palatine foramen, and small foramina scattered on the horizontal lamina represent minor palatine foramina. The palatine carries other scattered small foramina, some localized on the palatal crest. At its anterodorsal limit, and within the orbit, the palatine contributes to a large common infundibulum for the more pos­ teriorly placed sphenopalatine foramen and anteriorly placed ventral infraorbital foramen. The caudal palatine foramen (proximal opening of canal for palatine vessels and nerves) probably also opens here, although it is not visible. Nasal: The small, dorsally convex, and anteriorly deflected nasals form the vertex of the skull at the prominent "snout," where they lie well forward of the orbits. In dorsal view, each nasal is sculptured anteromedially and is longer than it is wide. The gently convex lateral border is markedly longer than the medial border, which is marked by a slightly depressed interna- sal suture. The nasals are slightly asymmetrical in shape and sculpture, but it is uncertain if this is original or postmortem. Anteriorly, each nasal forms a thin downturned roof over the bony nares. The two nasals are separated posteriorly by the blunt narial process of the frontals, and the naso-frontal suture is slightly depressed. Mesethmoid: The mesethmoid occurs in the posterior of the mesorostral groove (Figure 6), where it is some 32 mm high and 10-12 mm wide. Matrix obscures details. Presumably, the mesethmoid (mesorostral or septal) cartilage originally filled the mesorostral groove anterior to the ossified meseth­ moid. A round element seen in cross section in the narial cavity above the large exposure of the mesethmoid could be a dorsal part of the mesethmoid; it is about 14 mm high and 9-10 mm wide. The ectethmoid is not apparent. 200 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY Frontal: On the vertex, the frontals form a long, narrow (-28 mm wide), median exposure posterior to the nasals and medial to the ascending processes of the premaxillae. Anteri­ orly, the frontals are fused to form a broad, bluntly rounded narial process, which partly separates the nasals. Posteriorly, a median interfrontal suture arises from the fronto-parietal suture to separate the frontals for about 30 mm. The frontals at the vertex carry at least 12 small (<1 mm diameter) scattered fo­ ramina, interpreted as remnant supraorbital foramina compara­ ble with those of some Dorudontinae and archaic Mysticeti. From the posterior limit on the vertex, a narrow continuous strip of frontal descends laterally toward the orbit, bounded by the supraorbital process of the maxilla (anterior) and parietal (posterior). The parietal margin has a well-developed foliate fronto-parietal suture, on which the orbitotemporal crest origi­ nates. In a dorsal or anterodorsal view (Figures 2c, 6) of the skull, the two orbitotemporal crests form a conspicuous semi­ circular posterior border of the face; the crests probably indi­ cate the posterior limit of the origin of the nasofacial muscles. Each orbitotemporal crest grades laterally onto the supraorbital region, where it forms a strong ridge at the base of the postor­ bital process. The postorbital process is a wide, thin plate that descends steeply toward the apex of the zygomatic process of the squamosal; it does not extend far posteriorly. Farther anteri­ orly is the thick and robust orbital (supraorbital) part of the frontal, which is overlapped medially by the rather narrow as­ cending or supraorbital process of the maxilla. The preorbital process of the frontal appears blunt in dorsal view and is dors­ oventrally thick, but it is the lacrimal that here forms the ante­ rolateral part of the cranium at the antorbital notch. As seen laterally, the orbit is strongly arched dorsally, con­ trasting with the shallow, long orbit in most odontocetes. Fur­ thermore, its anterior profile is strongly concave posteriorly. The orbit lies above the level of the lateral border of the rostral part of the maxilla (above the base of the rostrum), but a little below the dorsoventral midpoint of the skull. The frontal forms most of the orbit, with some contribution of the lacrimal, jugal, and maxilla to the anterior wall. At least two presumed frontal foramina open in the roof of the left orbit about 30 mm internal to the edge of the orbit, just anterior to the postorbital ridge. Medially, the orbit passes into the deep groove of the optic in­ fundibulum, where the narrow, slit-like ethmoid foramen (see left side; Figures 12, 13) for the ethmoidal vessel and nerve opens immediately anterior to orbitosphenoid-frontal suture. Other details of the optic infundibulum are given under the or­ bitosphenoid. Posteriorly, the orbit is bounded by a strong postorbital ridge (crista orbitalis superior; Figures 2B, 4), which is indistinct lat­ erally on the postorbital process but which becomes more dis­ tinct medially as it grades into the dorsal edge of the optic in­ fundibulum (sensu Fraser and Purves, 1960). This ridge marks the anteroventral limit of origin of the temporalis on the fron­ tal. The preorbital ridge (ventral orbital crest, crista orbitalis in­ ferior), which normally delimits the anteroventral part of the orbit in mammals (Davis, 1964; Evans, 1993), is not developed laterally, but it is thick and low medially where it separates the optic infundibulum from the infraorbital infundibulum. There is no obvious pterygopalatine fossa. Anteroventrally, the orbit is delimited by a strong ridge of ju­ gal and maxilla at the posterior limit of the rostrum, to which the frontal does not contribute. Laterally, this ridge involves only the jugal (the lacrimal lies more dorsally), whereas farther posteromedially, the maxilla forms the ridge just ventral to the infundibulum for the infraorbital and sphenopalatine foramina. This ridge is a vestigial equivalent of the infraorbital process of the maxilla of other mammals. Lacrimal: The lacrimal is unexpectedly prominent, and the facial surface is exposed well to dorsal view and not covered by the maxilla. It is transversely elongate and dorsoventrally thick in the anterior wall of the orbit but does not contribute to the roof of the anterior part of the orbit (posterodorsal limits of the lacrimal within the orbit are not clear). Medially, the lacri­ mal extends beyond the antorbital notch onto the rostrum; the left side is well preserved. There is no evidence of a fossa for the lacrimal sac, fossa for oblique muscle, or lacrimal canal, and an orbital crest cannot be identified. Jugal: The jugal (zygomatic bone) is identified provision­ ally by analogy with the situation in young specimens of extant Ziphiidae, the only group of extant Odontoceti in which limits of the normally fused jugal and lacrimal can be determined eas­ ily. As seen ventrally, the left jugal in S. rayi has two anteriorly directed processes that invade the maxilla; the medial process is the larger. (Limits to the right jugal are obscure.) In the ab­ sence of a clear suture between the jugal and the more dorsal lacrimal, these two bones are presumed to be fused fully. Orig­ inally, a thin process of the jugal presumably underlay the or­ bit, passing posteriorly to contact the apex of the zygomatic process of the squamosal. Parietal and Interparietal: The prominent, short, and transversely rounded intertemporal region is formed by the pa­ rietals and interparietal. The interparietal has a narrow medial exposure, is bounded by ill-defined serrate to foliate sutures with the parietal, and fully separates the parietals; it extends from the supraoccipital to the frontal. There is a low, indistinct median ridge but no sagittal crest. Anteriorly directed parietal foramina open on the interparietal. Immediately lateral to the interparietal, each parietal is smooth with two indistinct subparallel parasagittal temporal crests (temporal lines, supratemporal crests of Novacek, 1986) developed posterolaterally (Figures 2A, 3). Anteriorly, about 17 mm from the midline, each parietal gives rise to a sharp orbito­ temporal crest that passes laterally and slightly anteriorly onto the frontal. Here, at its frontal border, the parietal dorsally but­ tresses the supraorbital process of the frontal (Figure 2A). Far­ ther ventrally, where, on the left, a prominent serrate suture is present (Figures 12, 13), the parietal extends a little anteriorly along the cranial wall. Anteroventrally, the parietal forms the posterior margin of the optic infundibulum, thus contributing to NUMBER 93 201 the postorbital ridge, and it also contacts the palatine and ptery­ goid immediately ventral to the optic infundibulum (Figures 12, 13). Farther posteriorly, the parietal extends ventrally to the infratemporal crest (= subtemporal crest; Fordyce, 1994), but details of contact here with the alisphenoid are uncertain. Also within the temporal fossa, the posterior suture with the squa­ mosal at the squamous border cannot be localized; on the bro­ ken left side, the parieto-squamosal suture is not identifiable on the preserved part of the braincase wall. The suture with the su­ praoccipital cannot be localized at the nuchal crest. Ventrally and posteroventrally, within the temporal fossa, the braincase widens markedly. The parietal appears to be present in the basicranium dorsal and medial to the periotic. This identification is not certain be­ cause the in situ periotic covers the area where the parieto- squamosal suture is expected, and it is possible that the element here may be squamosal only. The parietal is, however, known to be present in the identical position in an archaic odontocete USNM 205491 (Figure 5B), which is comparable in basicranial grade to S. rayi. The parietal also occurs in this position in Waipatia maerewhenua Fordyce (holotype, OU 22095; see Fordyce, 1994, for discussion) and in a wide range of other od­ ontocetes. Identified thus, the parietal forms the lateral border to the posterior lacerate foramen. Although the parietal extends medially toward the basioccipital, it does not contact the latter; thus the posterior lacerate foramen and more anteriorly placed foramen ovale are confluent, with a constricted apostrophe-like profile. The posterior suture of the parietal with the exoccipital can be localized only to within a few millimeters. Anteromedi­ ally, this basicranial part of the parietal forms a wedge that ex­ tends forward, bounded laterally by the alisphenoid, to form part of the foramen ovale. Supraoccipital: The broad supraoccipital slopes forward at about 45° from horizontal. It is roughly hemispherical in dorsal or posterior view (Figures 2A,D, 3, 7), whereas in lateral view the longitudinal profile is convex, more so laterally than medi­ ally. Anterodorsally, the surface of the supraoccipital is raised, but there is no distinct projecting external occipital protuber­ ance. Also anterodorsally, a short low external occipital crest lies just to the right of the midline (Figure 7), but there is no significant asymmetry. Below the external occipital crest is a faint median depression that passes ventrally into a raised re­ gion, without nuchal tubercles, just above the foramen mag­ num. The anterolateral quadrants of the supraoccipital are gen­ tly convex dorsally. The apex of the supraoccipital at the nuchal crest rises dor­ sally about 6 mm above the adjacent interparietal and parietals. Laterally, the nuchal crest (here sometimes termed the lamb­ doidal crest) overhangs the parietal and squamosal by up to 10 mm. The crest descends steeply toward the exoccipital and, at the posterior of the temporal fossa, swings abruptly forward onto the zygomatic process of the squamosal (forming the tem­ poral crest of some authors). Exoccipital: Dorsally, the exoccipital is fused completely with the supraoccipital, so that the limits of these elements are uncertain. The right condyle lacks the medial border, so that its outline is incomplete; as preserved it is reniform and trans­ versely narrow in posterior view. Enough remains to show that, originally, the condyles formed an articular surface with a primitively wide and oblate (rather than subcircular) profile (Figures 2D, 7), with a circular foramen magnum. The dorsal intercondylar notch is broad and shallow, although remnants indicate a more narrow ventral notch. The condyle lacks an ele­ vated pedicle, but the dorsal condyloid fossa is prominent and deep. The ventral condyloid fossa is open and spacious but less clearly delimited. Preservation is too poor to judge the state of the condyloid foramen (in other mammals, for condyloid vein between basilar sinus and sigmoid sinus), but presumably the foramen is small or absent as in other odontocetes. The long axis of the rather tabular and roughly parallel-sided paroccipital process extends ventrolateral^ and posteriorly (Figures 2D, 6, 14B, 16) to almost reach the level of the outer face of the zygomatic process of the squamosal. Ventrally, the apex reaches below the level of the basioccipital crest (al­ though the latter is pushed dorsally by postmortem crushing) and, viewed dorsally, extends posteriorly almost to the level of the posterior of the condyle. In lateral view, the paroccipital process appears mostly thin, but distally, at its contact with the post-tympanic process of the squamosal, it thickens markedly. This thick apex, seen ven­ trally, is nodular, wider than it is long, and roughly oval in pro­ file. The diffuse articulation for the stylohyal lies ventral to a horizontal, anteriorly facing, narrow groove or notch inter­ preted as the fossa for a posterior sinus, which originated from the elliptical foramen of the bulla. Immediately dorsally, be­ tween the exoccipital and posterior process of the tympanic bulla, is an open groove that probably carried the facial nerve externally; it is not clear whether the paroccipital process con­ tacts the posterior process of the bulla in the roof of this groove, or whether 1-2 mm of squamosal lies between these el­ ements. Farther dorsomedially, the anterior face of the paroc­ cipital process is convex, without any excavated fossa, but the open region between the paroccipital process and the pars co­ chlearis of the periotic probably held a lobe of the peribullary sinus. This interpretation, based upon study of extant odonto­ cetes, differs from that of previous authors (e.g., Fraser and Purves, 1960), who have identified this open region in diverse odontocetes as the fossa for the posterior sinus. Farther medi­ ally, the exoccipital bounds all of the jugular notch and de­ scends onto the base of the basioccipital crest. The notch is wide and shallow, and the small hypoglossal foramen (for nerve XII and perhaps the vein of the hypoglossal canal) opens anteromedially. Immediately anterodorsal to the hypoglossal foramen, the exoccipital is grooved, presumably for vessels that exited via the posterior lacerate foramen. Basioccipital: Broken and compressed bone obscures some details of the basioccipital; sutures with the exoccipital 202 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY FIGURE 14.—Simocetus rayi, USNM 256517, holotype elements, and unnamed odontocete USNM 205491: A, stereophotographs of right periotic of holotype skull of S. rayi, USNM 256517, ventral view (skull whitened with ammonium chloride); B, lateral view of right periotic of holotype skull of S. rayi, USNM 256517, with ventral surface upper (skull whitened with ammonium chloride); C, right periotic of unnamed odontocete USNM 205491, to show anteroexternal sulcus and posterior process; D,E, isolated anterior teeth associated with holotype skull of S. rayi, USNM 256517; D, buccal view; E, lingual view. (Scale bar=l cm.) are fused except at a crushed contact at the posterior of the ba­ sioccipital crest. The right basioccipital crest is robust and an­ teroposteriorly short, reaching forward to about the mid-level of the foramen ovale. Its current posteroventral and markedly lateral orientation probably reflects postmortem crushing; orig­ inally the crest extended farther ventrally and less posteriorly or laterally, with the posterior lacerate foramen better exposed to ventral view. Rough bone medially near the thickened apex of the crest probably marks the insertion for the longus capitis. Anteriorly, the basioccipital contacts the vomer medially; later­ ally, the basioccipital crest carries a small, narrow, oval facet interpreted as the suture for the medial lamina of the pterygoid. NUMBER 93 203 An indistinct tubercle lies medially at the contact with the vomer, anterior to a gentle elevation that runs transversely be­ tween the bases of the two basioccipital crests. Posteromedi­ ally, bilateral ridges that flare back toward each condyle delimit the insertions for the rectus capitis ventralis. Dorsolaterally, opposite to the periotic, the basioccipital crest lacks any obvious fossa for a peribullary sinus, although such a sinus probably was present. Dorsally, near the border for the posterior lacerate foramen, the basioccipital carries a narrow elongate shelf and an associated groove; because the region be­ tween the tympanic bulla and basioccipital is equivalent to the petro-occipital canal of other mammals, the groove may be for the ventral (inferior) petrosal sinus. The basioccipital contributes little to the margin of the later­ ally directed posterior lacerate foramen (= cranial hiatus in part of Fraser and Purves, 1960; see Fordyce, 1994), which other­ wise is bounded by the exoccipital (posterior) and, particularly, the parietal (lateral). The foramen is long and narrow (-20 mm x -10 mm), and anteriorly it is confluent with the much larger foramen ovale. Presumably the foramen transmitted cranial nerves IX-XI and any vessels from the sigmoid sinus. The re­ gion ventral to the posterior lacerate foramen includes the tym- pano-occipital fissure and jugular foramen of other mammals, now indistinct because the bulla and periotic are effectively ex­ tra-cranial, displaced ventrolaterally from the braincase. Squamosal: The squamosal appears to form the posterolat­ eral wall of the temporal fossa at and posterior to a prominent bulge in the braincase, although the parieto-squamosal suture cannot be localized here. A long, deep, narrow cleft in the dor­ sal surface of the squamosal floors the temporal fossa and sepa­ rates the braincase from the zygomatic process (Figure 2A, right side). Anteroventrally, the cleft is particularly narrow, confined anteriorly by the convex subtemporal crest and by the medial edge of the zygomatic process, as if to form a channel for a tendinous part of temporalis. The zygomatic process of the squamosal parallels the skull axis and forms the most lateral part of the skull (dorsal view; Figures 2A,B, 3, 4). Its dorsal crest is abruptly rounded trans­ versely, and the medial wall is steeper than the lateral. The zy­ gomatic process is deeper than it is wide, with a roughly comma-shaped cross section, and the thin ventral margin ex­ tends a little laterally. In lateral view (Figures 8A, 9), the dorsal margin of the process is subhorizontal, descending gently to a sharp apex below which is an elongate small facet for the con­ tact with the jugal. Posteriorly, the base of the zygomatic process has a deep, rough fossa for muscles of the neck. Subdivisions are not obvi­ ous; the muscles included some or all of the sternomastoideus, splenius, longissimus capitis, and mastohumeralis (see Schulte and Smith, 1918; Howell, 1927). The fossa extends anterodor­ sally above the external auditory meatus, is pitted deeply in its midpoint, and extends ventrally onto the post-tympanic process of the squamosal (lateral view; Figure 8 A). By analogy with ar- tiodactyls, the post-tympanic part of the fossa is inferred to be an origin for the sternomastoideus. The ventral surface of the squamosal is complex. The lateral edge of the zygomatic process has a marked crest, but limits to the glenoid cavity otherwise are not clear except at the robust postglenoid process. Here the tympano-squamosal recess for the middle sinus (fide Fraser and Purves, 1960) descends onto the postglenoid process, with the distal part of the recess bounded by a distinct ridge on the anterolateral part of the post­ glenoid process (Figure 5A). Although there is no excavated tympano-squamosal recess anteriorly, the structure of the squa­ mosal between the zygomatic process and steep-sided falci­ form process is compatible with a well-developed middle sinus (sensu Fraser and Purves, 1960) originating at the lower tym­ panic aperture between the periotic and tympanic bulla. Farther anteriorly (Figure 5A), the squamosal forms a broad shelf on which a thick rounded subtemporal crest ("projecting...tempo­ ral rim" of squamosal, of Kellogg, 1936:101) separates the temporal fossa from the basicranium. As in Waipatia maere- whenua, the squamosal lacks the anterior transverse ridge and the vestigial (nonpatent) postglenoid (retroarticular, retrogle- noid) foramen seen in Basilosauridae; see also the discussion of postglenoid foramen and foramen spinosum below. Anteriorly, between the pterygoid sinus and the groove for the mandibular nerve, the alisphenoid-squamosal suture is foli­ ate and complex, with multiple fine interdigitation. The falciform process is thin, with a long base; it projects ventrally, skewing slightly anteromedially, and is bifurcated distally. On the skull in modern odontocetes in which the tym­ panic bulla is articulated (e.g., growth series of Tursiops trun­ catus (Montagu) in USNM), the anteroventral or distal part of the falciform process lies close to the descending anterior edge of the outer lip of the bulla. This distal part is missing in S. rayi, presumably through postmortem loss. A second, more an­ terodorsal, division of the falciform process often is identifi­ able in odontocetes, immediately ventral to the path of the mandibular nerve. This division may contact the bony lateral lamina of the pterygoid sinus fossa. There no evidence that any extensive sheet of outer lamina of pterygoid contacted the fal­ ciform process in S. rayi. Posteriorly, the falciform process lies closely against the anterolateral face of the anterior process of the periotic, but farther posteriorly, the squamosal rises rapidly to expose much of the anterior process of the periotic to lateral view (Figures 14B, 16). A large foramen between the squamo­ sal and the fovea epitubaria of the periotic, immediately ante­ rior to the lateral tuberosity, marks the ventral opening of the anteroexternal sulcus, which in turn is presumably for the mid­ dle meningeal artery (see Fordyce, 1994). The in situ periotic obscures the squamosal within the periotic fossa. The external auditory meatus is more deep than wide distally, with the anterior and posterior walls roughly parallel for the distal half of its length, and is bounded by the steep posterior face of the postglenoid process. Medially, the meatus is more open and is separated from the tympano-squamosal recess by a 204 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY falciform process anterior bullar facet anterior process foramen at dorsal part of anterolateral sulcus lateral tuberosity mallear fossa inferred path of middle sinus postglenoid process anterior meatal crest spiny process of squamosal external auditory meatus posterior meatal crest facial sulcus posterior process of tympanic bulla post-tympanic process of squamosal 1 cm alisphenoid anterior keel anteroventral angle parietal fovea epitubaria fossa for origin of tensor tympani anterointernal angle of pars cochlearis external opening of facial canal stapes pars cochlearis fenestra rotunda fossa for stapedial muscle cochlear crest paroccipital process nferred path of facial nerve jugular notch FIGURE 15.—Simocetus rayi, USNM 256517, ventral view of right periotic of holotype skull, based upon Figure 14A, showing main features. (Scale bar=l cm.) sharp anterior meatal crest. The latter extends from the postgle­ noid process to the apex of the spiny process (sensu Muizon, 1987), in the hiatus epitympanicus of the periotic, immediately posterior to the lateral tuberosity (Figures 14A,B, 15, 16). Here the spiny process of the squamosal carries two grooves, of which the anterior marks the path of the middle sinus and the posterior is the meatus. The suture between the squamosal and tympanic bulla here is obscure, but the posterior meatal crest forms at least the dorsal few millimeters of the wall of the me­ atus, with the rest formed by the posterior process of the bulla. The latter covers the squamosal that is presumed to have a post-tympanic process comparable to that of Waipatia maere- whenua. The amastoid state of the skull (posterior process of the periotic not visible on the skull wall) is consistent with the interpretation of a substantial post-tympanic process. Periotic: The right periotic (Figures 14A,B, 15, 16) is pre­ served in situ and is applied closely to the squamosal (lateral, anterior), alisphenoid (anterior), parietal (anteromedial and me­ dial), exoccipital (posterior), and tympanic bulla (posterolateral and ventral). There is no clear evidence of original intimate su­ tural contact with these elements, other than at and near the posterior process of the bulla. Details of the lateral and dorsal surfaces of the periotic cannot be seen. The anterior process, body, and pars cochlearis are all skewed anteromedially relative to the skull axis. The anterior process is about as long as the body, is more narrow than deep, and has a convex lateral face and flatter medial face. As viewed medially, the anterior process is dorsoventrally deepest near its apex. Also, the anterior keel is curved as seen in medial view, with the anteroventral angle apparently more rounded than the anterodorsal angle, which, together with the medial face, is prolonged dorsomedially. The anterodorsal angle approaches a groove (at the parietal-alisphenoid suture) interpreted to lead toward the foramen spinosum. There are several subhorizontal fine grooves (anterointernal sulci of Fordyce, 1994) ventrally on the medial face of the anterior process, and one of these may be for the lesser petrosal nerve. Another groove traverses the medial face toward the anterodorsal angle. Fine details on this face of the periotic (e.g., presence or absence of a vertical ca­ nal) are obscured by resistant adhering matrix. Posteriorly, the medial face of the anterior process meets the perpendicular face of the pars cochlearis at an open groove; a fissure, which in odontocetes commonly accompanies an inflated pars coch­ learis, is absent. Fordyce (1994) and others, following Kellogg (1936), have interpreted the grooved contact of the anterior process with the pars cochlearis as representing the origin for the tensor tympani; however, it seems likely that the origin for the tensor tympani lies farther ventrally, within the shallow and more subhorizontal groove between the fovea epitubaria and the ventral surface of the pars cochlearis. A long (-9 mm), shallow, grooved anterior bullar facet lies on the ventral surface of the anterior process. The facet is bounded posteriorly by a wide shallow fovea epitubaria for the anterior pedicle and ac­ cessory ossicle of the bulla. The fovea is slightly deeper later- NUMBER 93 205 - horizontal plate of vomer^- basioccipital groove for mandibular nerve basioccipital crest posterior lacerate foramen anterior bullar facet fossa for origin of tensor tympani mallear fossa hypoglossal foramen jugular notch articulation for stylohya) paroccipital process ?fossa for posterior sinus pars cochlearis foramen ovale falciform process anterior process of periotic tympanosquamosaJ recess for middle sinus foramen at dorsal part of anterolateral sulcus posterior meataJ crest of squamosal external auditory meatus anterior meatal crest postglenoid process inferred path of facial nerve stapes in fenestra ovalis external opening of facial canal spiny process of squamosal lateral tuberosity inferred path of middle sinus FIGURE 16.—Simocetus rayi, USNM 256517, lateral view of right periotic of holotype with ventral surface upper, based upon Figure 14B, showing main features. (Scale bar=l cm.) ally, where it is succeeded by the prominent dorsal end of the anterolateral (or anteroexternal) sulcus. The indistinct ventral part of this sulcus is curved, as seen also in some platanistoids and eurhinodelphinids. Comparisons with other archaic Ceta­ cea (e.g., Zygorhiza kochii (Reichenbach), Waipatia maere- whenua, and archaic odontocete USNM 205491 of Figures 5B, 14c) suggest that this groove, bounded laterally by the squamo­ sal, trends toward the anterodorsal angle on the anterior process of the periotic and marks the path of the middle meningeal ar­ tery. The body is delimited anteriorly by the lateral tuberosity and by the transverse ridge that anteriorly bounds the mallear fossa. The latter ridge is indistinctly tuberculate both medially and laterally, and its relationship with the now-lost anterior pedicle of the bulla is uncertain. Posterodorsally, the large hemispheri­ cal lateral tuberosity is undercut deeply. An indistinct facet on the ventrolateral face of the tuberosity probably marks the point where the sigmoid process of the bulla originally lay close to the periotic. The irregularly subspherical mallear fossa is shallow and large, deepest dorsomedially, and faces ob­ liquely posteroventrally. Medially, the fossa merges smoothly onto the pars cochlearis without an obvious groove. There is no distinct fossa incudis, although whether this is an original ab­ sence or a postmortem loss is uncertain. The spiny process of the squamosal and, posteriorly, posterior process of bulla, ob­ scure the hiatus epitympanicus. The fenestra ovalis, with in situ stapes, faces posterolaterally and ventrally; its anterior margin is just visible in a lateral view from the skull margin into the external auditory meatus (Figures 14B, 16). The facial canal opens lateral and just anterior to the fenestra ovalis. Because of broken bone, details are uncertain about the facial sulcus (groove for facial nerve), the deep fossa for the stapedial mus­ cle, and the condition of the tympanohyal. The stapes is unre- vealing. The pars cochlearis is long and narrow, is not inflated, and has a smooth, somewhat tabular ventral surface. Its narrow an­ terior face passes via a rounded anterointernal angle onto the long, steep, irregular medial face. There is a sharp posterointer­ nal angle. The gently concave posterior face is prolonged vent- rolaterally into a laterally compressed, blunt, posterior cochlear crest (new term) that closely approaches the posterior process of the bulla. Laterally, the pars cochlearis rises abruptly toward the fenestra ovalis but does not obscure the latter from ventral view. Posteriorly, the indistinctly reniform fenestra rotunda is only just visible to ventral view. A slight nodule at the dorsal lip of the fenestra rotunda merges into a ridge directed toward the presumed position of the aperture for the cochlear aque­ duct. 206 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY The posterior process of the periotic is directed laterally, as seen from the jugular notch, but is not exposed on the skull wall; hence, the skull is amastoid. There is no evidence on the skull wall of a mastoid foramen. Tympanic Bulla: Most of the bulla is missing. The long, narrow, posterior process of the bulla is directed posterolater­ ally, forming the ventral part of the posterior wall of the exter­ nal auditory meatus (Figures 14A, 15). The proximal two-thirds of the process has a smooth crest that separates the rounded an­ terior surface from the tabular posterior surface. There is no ob­ vious groove for the facial nerve on the bulla, and there is no stylomastoid foramen laterally; rather, the facial nerve proba­ bly lay in the groove between the posterior process of the bulla and the paroccipital process. Laterally the process thins to a distal apex. Sutures between the dorsal surface of the posterior process and adjacent elements, which can be localized to within a few millimeters, indicate contacts similar to those in Waipatia maerewhenua and other archaic odontocetes (e.g., Squalodontidae, Eurhinodelphinidae), namely, with the poste­ rior meatal crest anteriorly, with the posterior process of the pe­ riotic posteriorly, and with the post-tympanic process of the squamosal laterally. Pterygoid and Fossa for Pterygoid Sinus: The pterygoid lines part of the fossa for the pterygoid sinus, and it forms a prominent hamulus. Unsurprisingly, delicate parts of the ptery­ goid are absent; the bone is inferred to have had lateral, ventral, and medial laminae (sensu Fraser and Purves, 1960), which mostly were lost postmortem. The long left hamulus, skewed to the right after death, forms the ventral extremity of the skull just below the level of the palate and basicranium. Ventrally, the transversely rounded anterior surface of the hamulus passes back into a broad-based conical distal portion, broken apically. Although this distal part is dorsoventrally compressed, it is ro­ bust and not invaded by the pterygoid sinus. The hamulus ex­ tends posteriorly at least to a level with the falciform process. Originally, the medially apposed hamuli formed a prominent posterior spine on the palate. More-lateral and more-dorsal features of pterygoid are inter­ preted next in terms of the pterygoid sinus fossa, which here is a large, dorsoventrally deep, compressed cavity lateral and, in part, ventral to the choana (Figures 2B , 4, 8E, 11-13). Rem­ nants at the hamulus indicate that the medial lamina of the pterygoid arose from the anterior half of the hamulus to flank the internal naris at the nasopharyngeal surface. A facet on the anterior face of the basioccipital crest (Figures 4, 5A) indicates that the medial lamina of the pterygoid extended posteriorly to contact the basioccipital, as in other odontocetes. Judging from the remnant of preserved profile at the hamulus, the eustachian notch opened at a level slightly posterior to the midpoint of the pterygoid sinus fossa. Part of the fossa for the pterygoid sinus invades the base of the hamulus, so that in life the sinus under­ lay the external half of the narial passage. A remnant of ventral lamina arises from the hamulus; in life, the ventral lamina probably extended outward to merge dorsally into the lateral lamina of the pterygoid. Remnants of the lateral lamina occur at the suture with the palatine (at the anterior of the pterygoid sinus fossa) and dorsally along the outer margin of the sinus fossa. A broken surface of lateral lamina indicates a posterior extent at least to the point where the subtemporal crest bulges laterally away from the adjacent pterygoid sinus fossa, and it is possible that the lamina reached farther back, to within 30 mm of the falciform process. The indifferently preserved thin dorsal lamina of the pterygoid forms the anterior and anteromedial roof of the oval elongate sinus fossa, with the alisphenoid forming the posterolateral surface. Anteriorly, the fossa is ex­ cavated dorsally above the level of the foramen ovale and sub­ temporal crest, but it is shallower posteriorly. A small foramen (diameter - 2 mm) placed medially near the apex of the fossa may be the pterygoid foramen, which opens dorsally into the foramen rotundum (Figure 13). The dorsal roof of the fossa, in the region of the pterygoid-alisphenoid suture, possesses many small foramina. Posteriorly, the fossa is bounded by a low ridge that marks the path of the mandibular nerve (V3) from the fora­ men ovale. There is no evidence that the pterygoid sinus ex­ tended beyond the skull base and into the orbit, or that it had medial or posteromedial lobes. Alisphenoid: Most sutures of the alisphenoid are fused or obliterated, and their limits are determinable only to within a few millimeters. The foramen ovale identifies the posterior border of the alisphenoid. The alisphenoid also forms the pos­ terior and posterolateral parts of the roof of the pterygoid sinus fossa (Figures 5A, 12, 13). The large subcircular foramen ovale is not occluded posteriorly by the parietal and basioccipital, so it is confluent posteriorly with the cranial hiatus. The postero­ medial limit of the alisphenoid is indicated by the carotid fora­ men in the adjacent basioccipital. Relationships with the squa­ mosal are uncertain because they are obscured by complex sutures; the alisphenoid apparently forms only a small medial part of the groove for the mandibular nerve. A splint of alisphe­ noid extends posteriorly between the squamosal and parietal to reach the anterodorsal angle of the anterior process of the peri­ otic (Figure 5A). The deep groove here between the parietal and alisphenoid is real, not a postmortem artifact; its position and orientation, and comparisons with other odontocetes (Waipatia maerewhenua and the unnamed odontocete USNM 205491 of Figure 5B), suggest that it is a fissure for the middle meningeal artery and that its anterior limit represents the fora­ men spinosum. Basisphenoid: The carotid foramen, which opens level with but ventromedial to the foramen ovale, indicates the pos­ terior extent of the basisphenoid (Figures 12, 13). Limits to the basisphenoid otherwise are obscured by the vomer ventrally, and by the inner lamina of the pterygoid laterally. The carotid foramen is small and elongate. This foramen is thought not likely to transmit an internal carotid artery in adult living mys- ticetes and odontocetes (Fraser and Purves, 1960; Vogl and Fisher, 1981), although an artery may be present in juveniles (Melnikov, 1997). NUMBER 93 207 , 1 cm FIGURE 17.—Simocetus rayi, USNM 256517, teeth, whitened with ammonium chloride: A,B, isolated crown of a cheek tooth; A, buccal; B, lingual; C,D, mandibular teeth; c, buccal; D, lingual. (Scale bar=l cm.) Orbitosphenoid: Details are seen best in the left orbit (Fig­ ures 11-13) in which, however, proximal parts of foramina could not be exposed fully. Kellogg's (1936, fig. 31c) illustra­ tion of Zygorhiza kochii helps interpret this region. The orbito­ sphenoid surrounds much of the elongate deep cleft of the optic infundibulum but cannot be identified positively in the anterior wall. The optic foramen opens anteriorly in the roof of the in­ fundibulum, which is level with but posteromedial to the eth­ moid foramen. The path for the optic nerve is perpendicular to the condylobasal axis. The orbital fissure (=portion of optic fo­ ramen for VI of Kellogg, 1936, fig. 31c) is not seen, but it is inferred to open farther posteriorly, deep within the infundibu­ lum. A small foramen presumably for the ophthalmic artery lies at the junction of the orbitosphenoid, frontal, and parietal, immediately dorsal to the optic infundibulum. Also on the or­ bitosphenoid is the path of the maxillary nerve (V2) that arises from the foramen rotundum (hidden) and runs anteriorly in a shallow groove that is lateral and ventral to the deeper part of the infundibulum. A low horizontal ridge bounds the groove dorsally, and the ventral edge of the groove is close to the su­ ture with the palatine. Mandible: Notable features of the incomplete right mandi­ ble (Figures 8B,D, 10A,B, 17C,D include its gracile, laterally compressed and ventrally deflected anterior portion, and ven­ trally and laterally inflated panbone (lateral wall of the mandib­ ular foramen). There are alveoli for 11 teeth, presumably i l -3 , c, pl-4, and m l - 3 . Missing are the apex of the coronoid pro­ cess, maxillary notch, condyle, condylar crest, angular process, and angle. Anteriorly, the tooth-bearing body has a sharp apex that passes back into the subparallel dorsal and ventral surfaces. Depth increases markedly about the level of m l - m 3 , as the panbone becomes laterally and ventrally inflated. The dorsal profile behind the posteriormost cheek tooth is barely elevated above the level of tooth insertion, but not enough of the dorsal profile remains to judge the shape of the now-missing apex of the coronoid process. In dorsal view, the mandible becomes in­ flated posterior to about m3. Overall, the mandible is suboval in cross section, although the crests are developed anteroven­ trally below about i l -c; posterodorsally, a sharp, narrow coro­ noid crest is developed behind m3. The apical 85 mm of the mandible, from pl forward, is deflected ventrally to comple­ ment the inferred original profile of the now-distorted rostrum. In dorsal view, the mandible is roughly straight except anteri­ orly where it is slightly recurved medially toward the symphy­ sis. A left mandible, if reconstructed to comparable profiles and articulated with the right, would result in an acutely pointed lower jaw with a narrow mandibular space. Proportions of the rostrum, however, point to a broader mandibular space. It thus seems likely that the orientations of the mandible and symphysis were distorted postmortem. The mandibular symphysis is short, indistinct, and not partic­ ularly prominent in dorsal or internal views. It extends from the apex of the mandible posteriorly to about level with the alveo­ lus of i3. Surface detail is too poor to tell patterns of symphy­ seal ridges and grooves, if any. There is no evidence that the left and right mandibles were fused. Four single alveoli with prominent diastemata occupy the an­ terior 67 mm of the mandible. These, and the alveolus for p l , lie toward the lateral (buccal) face of the mandible, whereas the more-posterior alveoli are on the dorsal surface. The forward- pointing tiny alveolus for il lies ventral to the apex. Much larger alveoli for i2 and i3 are suboval and directed anterolater- ally, so that these teeth probably were procumbent. A more rounded alveolus for the canine indicates a more vertically ori­ ented, single-rooted tooth. Here and farther posteriorly, there are no alveolar juga. A fine longitudinal ridge links the medial (lingual) margins of the alveoli. Two-rooted alveoli lie farther posteriorly; the seventh and ninth through 11th teeth (p3, ml-3) are in situ (Figure 17c,D). The shortest diastemata are between pl and p3; posteriorly thereafter, the diastemata lengthen enough to have accommo- 208 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY dated the crowns of opposing maxillary teeth. Poorly preserved shallow pits in the diastemata may be embrasure pits for the apices of upper cheek teeth. Five small elongate mental foram­ ina open on the buccal face (Figure 10A) between i2,3, level with the anterior of p2, level with the anterior of p4, level with the anterior of m2, and between m2,3. Only the crushed anterior margin of the mandibular foramen is preserved. As viewed laterally, the pan-bone is markedly in­ flated ventrally, relative to the rather straight dorsal profile be­ tween pl and the broken base of the coronoid process. As seen posteriorly, the medial (lingual) face of the mandible is gently concave. The mandibular canal is at least 70 mm deep and 20 mm wide, and its walls are less than 1 mm thick. Teeth: The teeth (Figures 14D,E, 17A-D) are heterodont and apparently not polydont. The tooth complement appears to be 10/3, Cl /1 , P4/4, M2/3, based upon the count of teeth, roots in place, and the alveoli, with the proviso that anterior alveoli are eroded and the matrix is filled. There is no evidence of decidu­ ous teeth. It is uncertain whether premolars and molars can be differentiated on the basis of crown structure; in this specimen they are identified by position. Premaxillary teeth are not rec­ ognizable and, furthermore, each premaxilla lacks alveoli. Teeth preserved in the skull (Figure 2B) are the right P4 and Ml, left P2, and roots of left P3 and Ml; none of these upper teeth is preserved well enough to warrant detailed illustrations. Mandibular teeth are listed above. Isolated teeth are three ante­ rior teeth (i and/or c; Figure 14D,E), one cheek-tooth crown (perhaps left ml ; Figure 17A,B), and a fragment of one cheek­ tooth crown. The isolated, slender, single-rooted teeth with simple crowns are incisors and possibly a canine. All are rather small in abso­ lute size and also in size relative to the mandible. The longest, with a relatively small crown, may represent the procumbent lower right i2, for which the alveolus is large and directed for­ ward. Of the two smaller teeth, the straighter is perhaps the lower left i3; the more recurved could represent the lower left or upper right C. In each single-rooted tooth, the crown is acutely pointed with a small apical angle (sensu Rothausen, 1968), is recurved slightly to markedly lingually, and is antero­ posteriorly keeled. The basal cross section is oval and trans­ versely compressed, and the buccal face is more convex than the lingual face; transverse compression is pronounced api­ cally. The enamel is smooth or has microscopic vertical striae buccally, whereas the lingual face has a few low, vertical, sub- parallel ridges. In each tooth, the root has a smooth surface, is thicker below the crown, and is roughly circular in cross sec­ tion. Features of the cheek teeth are shown in Figure 17A-D. On the skull, only the right P4 appears to be in undistorted original position, facing ventrally and slightly laterally (buccally) (Fig­ ure 8A). The cheek teeth in place are emergent, with the crown well clear of the alveolus. Generally, the triangular crowns are laterally compressed, are recurved lingually in the upper teeth, and have a triangular main (apical) denticle and two or more accessory (anterior and posterior) denticles. In other archaic odontocetes, the apical denticle becomes smaller posteriorly in the tooth row relative to the accessory denticle(s); in S. rayi the apical denticle in the lower p3 is larger than in the ml, but more teeth are needed to determine the real trend. Accessory denticles are small, compressed, free-standing, and keeled. They are arranged anteroposteriorly along the keels of each tooth. In the lower teeth, an occlusal view shows that this ar­ rangement is roughly linear, but for the upper teeth, the anterior and posterior keels, and their associated denticles, trend bucca­ lly as they descend from the main denticle. Some accessory denticles are preserved well enough to show apical wear. There are large facets on the left P2 (anterolingual face), right Ml (anterobuccal edge), ml (posterior keel), m2 (posterior keel), and on the isolated crown, resulting from tooth-to-tooth wear that has removed denticles. The right m3 lacks large anterior denticles and is clearly asymmetrical in lateral view. Anteri­ orly, the keel on m3 carries a flat face bounded by two distinct ridges, similar to the anterior vertical groove bounded by ridges in the lower molars of basilosaurids (see Kellogg, 1936:124- 125). Ornament is present on all cheek teeth, less marked buccally than lingually. The buccal ornament (Figure 17c) comprises coarse, low, broad-based ridges that arise in the midline toward the crown base and converge apically toward the main denticle. The ridges are shorter and coarser basally. The buccal orna­ ment is most pronounced on the posterior mandibular teeth. The lingual ornament (Figure 1 7D) is more raised, is longer, has sharper crests, and is distributed farther apically and an­ teroposteriorly along the crown than for the buccal ornament. A faint papillate cingulum, in the shape of an inverted open V, is present lingually on lower m2,3. There is no cingulum on the more-anterior lower cheek teeth or the in situ upper cheek teeth, although ornament is more pronounced toward the crown base as if forming an incipient cingulum. In all cheek teeth, a vertical median sulcus extends onto the crown base buccally and lingually from between the roots, so that a cross section near the crown base has a compressed figure-8 shape. The enamelocementum boundary varies in profile (lateral view) from gently convex apically to an inverted V shape, with the apex of the V marking the median sulcus on the crown base. Cheek-tooth roots are subparallel and fused for all of their exposed length out of the alveoli. In the slightly displaced up­ per left P2, the exposed roots curve posteriorly. The right P4 has what may be a lingual third root, fused to the face of the posterior root and skewed posteriorly into the alveolus. Some roots have a prominent basal swelling near the crown base. Roots of the right P4 and Ml also have conspicuous lines, pre­ sumably growth lines, on the cementum parallel to the surface of insertion in the alveolus. NUMBER 93 209 Discussion of Morphology, Homology, and Function The skull of Simocetus rayi helps us understand the early evolution of cranial functional complexes in odontocetes. Pre­ viously, patterns of odontocete cranial evolution have been in­ ferred mainly from Neogene and living odontocetes (Miller, 1923; Kellogg, 1928). The few archaic odontocetes described, such as Agorophius pygmaeus (Muller, 1849) and Archaeodel- phis patrius Allen, 1921, have given tantalizing glimpses of primitive morphologies, which, however, need to be put into modern context. For Simocetus rayi, the functional complexes discussed below are the face, olfactory complex, orbit, feeding apparatus, and basicranium. Face: Simocetus rayi may be interpreted in light of the fa­ cial structure in living odontocetes. In odontocetes, the unique posteriorly expanded supraorbital or ascending process of the maxilla (Winge, 1921; Miller, 1923) forms an origin for hyper- trophied maxillo-naso-labialis (nasofacial) muscles (Lawrence and Schevill, 1956; Schenkkan, 1973; Mead, 1975; Heyning, 1989; Curry, 1992). Nasofacial muscles and the associated na­ sal diverticula and melon are thought to produce and transmit high frequency sounds used in echolocation (Mead, 1975; Wood and Evans, 1980; Heyning, 1989; Cranford et al., 1996). On the basicranium, the pterygoid sinuses may help receive and process high-frequency sounds (Norris, 1968). Mysticetes lack an odontocete-like supraorbital process of the maxilla; they also lack hypertrophied nasofacial muscles, nasal divertic­ ula, and a large melon (Heyning and Mead, 1990). Further, mysticetes are not known to produce and use high-frequency sounds in the manner of odontocetes (Heyning and Mead, 1990). The face of Simocetus rayi shows some broad sutural and to­ pographic patterns similar to those of extant odontocetes. In­ deed, such features are critical in identifying S. rayi as an odon­ tocete. Furthermore, the skull structure points to S. rayi as having essentially the same facial soft tissues as are seen in ex­ tant odontocetes. Below is a generalized summary about facial structures in living odontocetes that might also apply to S. rayi. This summary is based upon Lawrence and Schevill (1956), Moris (1969), Schenkkan (1973), Schenkkan and Purves (1973), Mead (1975), Heyning (1989), Curry (1992), and Cran­ ford etal. (1996). In odontocetes, the external nose is a dorsal single blowhole, generally median and well posterior to the apex of the rostrum. Complex nasal diverticula are developed in the soft tissues of the face between the blowhole and the skull. Generally, for the nasal diverticula, it is only the premaxillary sac that has a dis­ tinct bony origin (on the premaxillary sac fossa). Proximally, nasal plugs occlude the nasal passages at the external nares. Nasofacial muscles originating from the premaxilla, and par­ ticularly the maxilla, open and close the nostril (blowhole) and manipulate the nasal diverticula. Nasofacial muscles have dis­ tinct origins on the premaxilla (nasal plug muscle), on rostral and facial parts of the maxilla, and on some of the frontal. Pos­ teriorly, the hypertrophied nasofacial muscles commonly oc­ cupy a distinct depressed facial fossa, with little, if any, bony differentiation of muscle origins in the facial fossa. Anterior limits of the nasofacial muscles on the rostrum are diffuse. A melon is present anteriorly, although without a consistent dis­ crete bony fossa, so that it is difficult to predict size and shape of the melon from skull form alone. The premaxillary foramina transmit branches of the internal maxillary artery and the maxillary nerves to parts of the nasofa­ cial muscles anteriorly and posteriorly on the rostrum (see, e.g., Schenkkan, 1973, fig. 5, on Mesoplodon bidens (Sowerby)). More posteriorly placed dorsal infraorbital foramina supply branches of the internal maxillary artery and maxillary nerves to the region near the antorbital notch and the facial fossa. There is little bony evidence of venous drainage from the face (Mead, 1975). The odontocete face is innervated by sensory infraorbital branches of the maxillary division (V2) of the trigeminal nerve, which issue from the dorsal infraorbital foramina, and the facial nerve, which passes dorsally to the face via the antorbital notch (Huber, 1930, 1934; Mead, 1975). Within each bony naris is a diagonal membrane, a soft tissue structure that may have a role in sound production (Mead, 1975). The membrane lies in the posterolateral corner of each bony naris and is inserted at the posterior of the nasal septum distal to small foramina in the mesethmoid that may be a vesti­ gial olfactory foramina or a foramina for the nasal nerve. The orbit is displaced ventrolaterally to become functionally independent of the rest of the face. Mysticetes and archaeo­ cetes, which lack an enlarged facial fossa, have a similar orbital form, so that such changes probably are not caused solely by development of the nasofacial muscles. All the above features are inferred for S. rayi. There is no reason to think that bone and soft tissues had a fundamentally different structure from that of living odontocetes. Facial to­ pography, detailed suture patterns, and positions of foramina differ from those of living species, however, and require com­ ment. For example, S. rayi has a prominent depression laterally on the maxilla immediately anterior to the antorbital notch; this rostral area perhaps held a substantial volume of nasofacial muscles that inserted around the nose. Extant odontocetes, in contrast, have nasofacial muscles arising mainly on the cra­ nium, with a limited rostral component (Mead, 1975; Heyning, 1989; Cranford et al., 1996). Simocetus rayi also differs in fa­ cial topography in having a long rounded "snout," presumably for a well-developed olfactory complex (discussed below). The internal and external borders of the premaxillary sac fossa are obvious in S. rayi, as in extant odontocetes. The lack of a distinct roughened origin for the nasal plug muscle does not preclude this muscle's presence; the bony origin is indis­ tinct, for example, in the living Platanista spp. and mysticetes that have nasal plugs but lack a discrete roughened fossa on the premaxilla. Simocetus rayi has only an indistinct prenarial con­ striction, with the premaxillae barely expanded over the meso­ rostral groove anterior to the narial opening; a cartilaginous median septum was probably developed here and farther poste- 210 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY riorly, to form a median border for the nasal plugs. The pre­ maxillary sacs would have lain anterolateral rather than ante­ rior to the bony nares, with posteromedial rather than posterior openings into the narial cavity. Extant adult odontocetes gener­ ally have one, not two, premaxillary foramina, in contrast to S. rayi, but double foramina do occur sporadically. For example, an adult of Lipotes vexillifer Miller has a second premaxillary foramen in the anteromedian sulcus (AMNH 57333; see also Brownell and Herald, 1972, fig. 1), and neonatal delphinids may show two closely associated premaxillary foramina that become confluent in adults. Two well-developed premaxillary foramina occur in the premaxilla of some extinct odontocetes (e.g., Patriocetus ehrlichi (Van Beneden) of Rothausen, 1968, fig. 2a; Squalodon bariensis (Jourdan) of Muizon, 1991, fig. 7a). In these fossils, the posterior premaxillary foramen lies within the posterolateral sulcus, as in S. rayi. A posteromedian sulcus, which is well developed at the pre­ maxillary sac fossa of many other odontocetes (e.g., Lophoce- tus repenningi Barnes (1978:4); Mesoplodon bidens (Schenk­ kan 1973:5)), is not obvious in S. rayi, but another sulcus, the median premaxillary cleft, is notable. A comparable feature oc­ curs in the fossils Waipatia maerewhenua Fordyce, 1994 (holo­ type OU 22095) and Lomacetus ginsburgi Muizon, 1986 (holo­ type MNHN PPI 104; fissure originally interpreted as a taxonomically important median suture between the premaxilla and maxilla—see Muizon, 1988c:33: "fissure longitudinale"). On the left premaxilla of the unnamed odontocete USNM 299482, the median premaxillary cleft meets the eroded rem­ nant of the posterolateral sulcus at a premaxillary foramen. Similar clefts arise from the premaxillary foramen in some ex­ tant odontocetes (Mesoplodon densirostris (Blainville), USNM 486173; Mesoplodon hectori (Gray), USNM 504260; Phoc- oena phocoena (Linnaeus), USNM 550843 and USNM 550844). Simocetus rayi differs dramatically from extant odontocetes in the structure of its bony nares. The nasal passages of S. rayi are directed obliquely, with the bony nares roofed by long and dorsoventrally thin nasals and frontals, so that the external bony nares are more similar in position and orientation to those of extant mysticetes than to extant odontocetes. The nasal re­ gion in S. rayi could not be prepared, but some other structures may be inferred. In the unnamed archaic odontocete USNM 299482, which is similar in external topography and bone con­ tacts to S. rayi, well-developed turbinal bones (Figure 5D) are exposed in the broken cross section at the posterior of a long nasal cavity (Figure 5E). Vestigial turbinals also occur in mys­ ticetes (Edinger, 1955) and are developed well in archaeocetes (Stromer, 1908); among odontocetes, Squalodontidae may pos­ sess small cribriform plates, turbinals, and prominent olfactory peduncles within the cranium (e.g., Dart, 1923, fig. 19; Kellogg, 1928:198-202; Flynn, 1948). Turbinals are inferred, therefore, in S. rayi. Also in unnamed archaic odontocete USNM 299482, as in mysticetes and, presumably, S. rayi, the external bony nares are rather removed from the anteroposteri­ orly elongate olfactory cavity and turbinals. In contrast, most Neogene and living odontocetes have near-vertical narial pas­ sages, and anteroposteriorly short and nodular nasals that rarely roof the nares, and they lack an capacious olfactory cavity with the turbinals. Small foramina that could be vestigial olfactory foramina are nearly ubiquitous in the mesethmoid of extant od­ ontocetes; alternatively, some of these foramina could mark the path of the nasal nerve and associated vessels. Simocetus rayi thus probably had soft tissues of the face that functioned around a snout more similar in shape to that of modern mys­ ticetes than to modern odontocetes. It is not clear how the prominent snout in S. rayi would have constrained the structure and operation of the nasal plugs, the diagonal membrane, and the nasal diverticula. Retracted nasals, a reduced olfactory complex, and development of a roughly vertical narial passage have been regarded as functionally linked to the telescoped odontocete skull (Miller, 1923; Norris, 1968). Structures in S. rayi, however, indicate that, during od­ ontocete history, the nasofacial muscles migrated posteriorly first, and the olfactory complex was reduced later. FACIAL ASYMMETRY.—Bilateral asymmetry at the flared rostral margins of the maxillae and antorbital notches in S. rayi is comparable to that in some other archaic odontocetes (Waipatia maerewhenua, Microcetus sharkovi Dubrovo, Squa- loziphius emlongi Muizon) discussed by Fordyce (1994:165). Because this maxillary and rostral asymmetry occurs in a range of archaic odontocetes, it is presumed to be directional, with functional implications, and not merely fluctuating or random asymmetry that might be ascribed to ontogenetic "noise." Bony rostral asymmetry probably reflects asymmetry in overlying muscles, although study of living species does not help to un­ derstand which of the rostral muscles are involved (Mead, 1975; Heyning, 1989; J.G. Mead, pers. comm., 2001). Nasal asymmetry is seen in S. rayi; the right nasal has a more convex lateral border and blunter posterolateral profile, but whether this is fluctuating or directional asymmetry is uncertain. Asymmetrical nasofacial muscles and/or nasal diverticula are ubiquitous in living odontocetes (Schenkkan, 1973; Mead, 1975; Heyning, 1989), even those with apparently symmetrical skulls (e.g., Pontoporia blainvillei (Gervais and d'Orbigny); cf. Kellogg, 1928, fig. 11, and Schenkkan, 1973, fig. 13). It seems likely that the largely symmetrical S. rayi probably also had asymmetrical facial soft tissues. GRADES OF EVOLUTION OF THE FACE.—The grade of facial structure for Simocetus rayi is intermediate between that of the most archaic odontocete described, Archaeodelphis patrius Allen, 1921, and the rather more modern Waipatia maere­ whenua Fordyce, 1994. A comparison between these species helps to understand changes in the grade of evolution of naso­ facial muscles. Two other widely cited archaic taxa are not con­ sidered in this section but are discussed under "Phylogenetic Relationships," below. One, Agorophius pygmaeus (Muller) NUMBER 93 211 (see True, 1907), is known certainly only from a single, now- lost, skull for which details are unknown; it is similar in grade to S. rayi. Another, the enigmatic Xenorophus sloani Kellogg (1923b), is rather similar to Archaeodelphis patrius. In A. patrius, the supraorbital or ascending process of the maxilla extends back little, the orbit is elevated with a promi­ nent infraorbital process of the maxilla, development of a facial fossa on the cranium is minimal, and the dorsal infraorbital fo­ ramen lies rather anteriorly about level with the antorbital notch. Such features indicate that the nasofacial muscles origi­ nated far anteriorly on the cranium and on the base of the ros­ trum. Because the supraorbital process of the maxilla extends posteriorly well beyond the facial fossa and dorsal infraorbital foramina, "telescoping" of the maxilla is perhaps more than just a simple response to posterior migration of the nasofacial muscles. Anteriorly, the premaxilla is lost, so that details are not known for the premaxillary sac fossa and associated sulci. In S. rayi, the supraorbital process extends posteriorly to­ ward the orbitotemporal crest but not far laterally over the su­ praorbital process of the frontal; this condition is presumed to be intermediate between that seen in A. patrius and most other odontocetes. The orbit is elevated rather less than in A. patrius, the infraorbital process of the maxilla is minimal, and a facial fossa lies above the orbit. Multiple dorsal infraorbital foramina open both on the base of the rostrum near the antorbital notch and on the supraorbital process of the maxilla. Judging from the topography of the supraorbital region, the facial portion of the nasofacial muscles was better developed than in A. patrius. The prominent lateral crest on the supraorbital process of the frontal probably marks the limits of origins of the nasofacial muscles; interestingly, a substantial part of the supraorbital ori­ gin of the nasofrontal muscles lies on the frontal, rather than mainly on the maxilla. The premaxillary sac fossa and associ­ ated sulci lie well anterior to the antorbital notch. Waipatia maerewhenua shows a more advanced grade, com­ parable to that of many extant groups: the facial fossa is large; the roof of the shallow orbit is about level with the lateral bor­ der of the rostral part of the maxilla, so that the origins of the rostral muscle 5 and facial muscle 5 are roughly on the same plane; and the maxilla does not contribute to the orbit. Well-de­ veloped premaxillary foramina and sulci are associated with a broad "spiracular plate" for the premaxillary sac fossa. Fordyce (1994) concluded that W. maerewhenua, with facial structure fundamentally the same as in extant Odontoceti, was probably capable of echolocation. FEEDING APPARATUS.—Teeth in S. rayi contrast with the dentitions of archaeocetes in that the teeth are absolutely and relatively small, the cheek teeth are separated by marked di­ astemata, and the posteriormost cheek teeth are inserted far an­ terior to the antorbital notch. Such features also occur in the ar­ chaic mysticete genus Aetiocetus (see Barnes et al., 1995). The tooth complement of S. rayi is unusual for an odontocete in that upper incisors are absent and the dentition is not polydont. The lack of alveoli in the premaxillae suggests that 11-3 ei­ ther were lost or were tiny and embedded only in the gums. In contrast, upper incisors are well developed in other archaic od­ ontocetes, such as squalodontids and kentriodontids, and in basilosaurid archaeocetes, all of which have three premaxillary teeth. Incisors also occur in archaic mysticetes. In extant odon­ tocetes, functional incisors are retained if functional maxillary teeth also are present. Upper incisors are absent in some living species of odontocete (e.g., Grampus griseus (Cuvier) and many Ziphiidae) that lack a functional upper dentition. Among fossil taxa, only eurhinodelphinids reportedly have toothless premaxillae. A special explanation, below, seems warranted for incisor loss in Simocetus. Polydonty, or increase in tooth number above the usual mammalian complement, previously has been interpreted as a synapomorphy for Odontoceti (e.g., Fordyce, 1983a; Barnes, 1990:21, item 10), so that the lack of polydonty in S. rayi, an archaic odontocete, is unexpected. All extant Odontoceti, and fossil Odontoceti for which complete dentitions are known, ei­ ther are polydont or have dentitions plausibly reduced from a polydont condition. For example, Xenorophus sloani Kellogg (1923b), perhaps the most archaic odontocete for which the dentition is documented, has 10 maxillary teeth. Among other Cetacea, archaic mysticetes and embryos of some living mys­ ticetes are polydont, whereas archaeocetes and a few archaic mysticetes are not polydont (Kellogg, 1936; Barnes et al., 1995). For Simocetus, tooth complement could be plesiomor­ phic with archaeocetes, or it could reflect a secondary reversal to a nonpolydont state. The feeding apparatus of S. rayi is quite unlike that described for other odontocetes, thereby allowing a novel interpretation of function. Alveolar form for i l -c suggests that these lower teeth were procumbent, with il vestigial. Anteriorly, the dorsal surface of the conjoined mandibles at the symphysis is narrow and flat, without teeth that protrude above the jawline, so that the apically downturned mandibles probably occluded against the similarly downturned, flat, edentulous anterior of the pal­ ate. Analogs are not seen in other Cetacea, but among Sirenia, Domning (1978) observed that the degree of deflection of the sirenian rostrum appears to correlate directly with the degree to which bottom feeding is used. Sirenia generally lack functional anterior teeth and use horny pads instead to crop and crush veg­ etation. The apices of the rostrum and mandible are strength­ ened, the mandible by thickening of the dorsal edges of the horizontal bodies and lateral edges of the masticatory surface (partly developed in S. rayi), and the snout by buttressing of the dorsal outline (not in S. rayi). Therefore, Simocetus rayi may have been a bottom feeder. In view of the relatively delicate cheek teeth and nonbuttressed rostral apex, a durophagous diet (e.g., molluscs) seems unlikely; soft-bodied benthic inverte­ brates were probably taken. Further evidence in support of bottom feeding is provided by the presence of indistinct large, shallow, paired pits at the ros- 212 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY tral apex, presumed to be for the vomeronasal (Jacobson's) or­ gan. A groove from each pit leads toward the presumed pa­ latine fissure. The vomeronasal organ is a branch of the olfactory complex that is functional in some other mammals (e.g., Evans, 1993) and has been reported in extant mysticetes (Quay and Mitchell, 1971). A vomeronasal organ could have functioned in S. rayi as a chemoreceptor in bottom feeding. The cheek teeth are delicate, with small denticles, and are widely spaced. Occlusal wear is on the anterior (mesial) and posterior (distal) faces, rather than on the buccal or lingual. Carnassial-like shearing seems unlikely; rather, teeth probably functioned in simple grasping or perhaps in a sieve-feeding system like that of the extant crab-eater seal, Lobodon carcino- phagus (Hombron and Jacquinot) (see King, 1961). The com­ bination of a broad palate, short mandibular symphysis, and long mandibular space indicates a voluminous mouth that could have functioned to hold water during suction feeding or filter feeding. Simocetus rayi perhaps fed on small epifaunal or shallow infaunal invertebrates detected with the aid of the vomeronasal organ. Perhaps it was a mud-grubber and used teeth to filter food from a substrate-water slurry. The long, robust, conical unexcavated proximal portions of the pterygoid hamuli possibly functioned as secondary poste­ rior extensions of the hard palate. Similar long hamuli occur in a few other extinct odontocetes (e.g., unnamed problematic ar­ chaic odontocete USNM 243979, the eurhinodelphinids Eurhi- nodelphis bossi Kellogg, 1925, Argyrocetus joaquinensis Kellogg, 1932, and the presumed eurhinodelphinid Squaloziph- ius emlongi Muizon, 1991), but the functional significance is not clear. Accounts of the nasopharyngeal muscles in extant odontocetes (Fraser and Purves, 1960) indicate that enlarged hamuli could play a major role in the function of the palato­ pharyngeal muscles. FEEDING MUSCLES.—Skull proportions indicate that the temporalis was the largest masticatory muscle, which suggests a simple hinge closure of the mandible as in archaic eutherians, where the temporal is dominant (Turnbull, 1970). The propor­ tions of cranium to rostrum, and to temporal fossa, indicate a lever action of the mandible that was powerful and slow; in contrast, Waipatia maerewhenua had proportionally smaller temporal fossa, longer forceps-like jaws, and an inferred faster but weaker snap (Fordyce, 1994). Other features related to the temporalis are puzzling; examples are the long tabular postor­ bital process (implicated in the action of the temporalis; Pen-in, 1975); the straight, long inner face of the zygomatic process; and the deep cleft on the squamosal in the posterior floor of the temporal fossa. The masseter and zygomatico-mandibularis muscles were probably small, given the reduced infraorbital process of the maxilla and the small zygomatic arch of the ju­ gal. The palatine, perhaps aided by the lateral lamina of the pterygoid, formed a large origin for the pterygoid muscles, but there is no evidence of hypertrophied pterygoideus or signifi­ cant lateral/medial movements of the mandibles in feeding. The paroccipital process is large, potentially providing a large origin for the digastric muscle. Alternatively, large size could relate to articulation of the stylohyal. ORBIT.—Key features of the orbit are the arched profile ex­ tending well above the level of the lateral border of the rostral part of the maxilla, a deep optic infundibulum, an indistinct preorbital ridge, a large infraorbital infundibulum, and a lim­ ited contribution of the maxilla to the anterior wall. In archaeo­ cetes, the orbit is arched more strongly, with a more distinct preorbital ridge (so that the sphenopalatine foramen is better separated from the orbit); in contrast, the orbit in Neogene and living odontocetes lies about level with the alveoli and usually is flattened from above by the facial fossa. In archaeocetes, the optic infundibulum is shallower, with contributing foramina placed more posteromedially in a much narrower interorbital region than in Simocetus; possibly the depth of the optic in­ fundibulum in S. rayi relates to relative interorbital width, al­ though the functional significance of change in interorbital width is uncertain. There is no reason to think that the volumi­ nous optic infundibulum held a rete. Simocetus rayi differs from the dorudontine archaeocete Zygorhiza kochii (see Kellogg, 1936, fig. 31c) in that the foramen rotundum for the maxillary branch of the trigeminal nerve, V2 (=sphenorbital fissure of Kellogg), opens within the optic infundibulum, to­ gether with the optic foramen and orbital fissure. Change in the position of the foramen rotundum relative to other orbital fo­ ramina may reflect shortening of the intertemporal region. Mysticetes, which have a shortened intertemporal region, also have a foramen rotundum that opens within the optic in­ fundibulum. The relative importance of the orbital and associ­ ated foramina is hard to judge from their size; foramen size in extant Cetacea does not always correlate with size of the nerve or vessel that issues from it, and some foramina may be en­ larged through the development of retia (Breathnach, 1960). In terms of maxillary contribution to the orbit, S. rayi is in­ termediate between archaeocetes and modern odontocetes. In archaeocetes, the maxilla forms the anterior wall of the strongly arched orbit, whereas there is minimal contribution in modern odontocetes. In living species, the orbit and ventral in­ fraorbital foramen are in about the same horizontal plane as the lateral border of the rostral part of the maxilla. The maxilla ef­ fectively forms the anterodorsal roof of the orbit. A preorbital ridge (inferior orbital crest) may be present. These changes in the orbital region perhaps reflect the posterior movement of fa­ cial muscles and development of an enlarged facial fossa, rather than major changes in eye function. PTERYGOID SINUS.—Norris (1968) suggested that air sinuses in the basicranium may isolate the auditory region from self- produced sound and may help to channel external sound to the periotic. Therefore, the development of sinuses may give a guide to hearing capabilities. In S. rayi, the fossae for the ptery­ goid sinuses are more derived than those of basilosaurid ar­ chaeocetes; they are longer relative to cranial length, extend NUMBER 93 213 relatively farther forward, are relatively more excavated dor­ sally, and extend ventromedially into the pterygoid hamuli. Relative to modern odontocetes, however, the fossae in S. rayi are more primitive in many respects. They do not extend as far anteriorly on the cranium and appear to be relatively smaller than, for example, the superficially similar but large sinus fos­ sae of Ziphiidae and Physeteroidea. Further, the sinuses in S. rayi are restricted to the basicranium, in contrast to many other odontocete groups in which the sinuses invade the orbit (Fraser and Purves, 1960) through dorsal expansion, often with loss of a bony wall on the pterygoid sinus fossae. In S. rayi, the alisphenoid forms the posterodorsal and poste­ rolateral parts of the pterygoid sinus fossa, as in basilosaurid archaeocetes and the archaic mysticete Mammalodon colliveri Pritchard. None of these cetaceans is preserved well enough to see clearly how alisphenoid relates to the pterygoid at the lat­ eral wall of the fossa. Of the more-posterior parts of the pterygoid sinus complex, little can be said about the middle sinus, other than to empha­ size that bone topography is consistent with the presence of that sinus. Proportions of the peribullary sinus (e.g., volume between bulla and basioccipital) cannot be judged because the tympanic bulla is missing. The basioccipital crest is relatively deeper and narrower than in archaeocetes, possibly contribut­ ing to an enlarged fossa for the peribullary sinus. The crest is not excavated to form a thin plate, however, as in some species of Delphinidae, for example. Concepts of the posterior sinus among odontocetes and mys­ ticetes are confused. Many odontocetes carry a fossa, com­ monly termed the fossa for the posterior sinus, on the anterior face of the paroccipital process dorsal to the apex of this pro­ cess. The cavity may be deeply concave, e.g., as in Phocoena phocoena and Pontoporia blainvillei, or may be shallow but well delimited, as in Tursiops truncatus. In many odontocetes, there is a less distinct excavation anteriorly at the apex of the paroccipital process, right at the articulation of the stylohyal and clearly distal to the position of the often more distinct larger fossa. In heads of extant delphinids, injections of the pterygoid sinus complex with silicone rubber (Stenella longi- rostris (Gray), USNM 396173; Lagenodelphis hosei Fraser, USNM 396079) revealed that this less distinct excavation car­ ries a small lobe of sinus that originates from the eustachian cavity via the elliptical foramen of the tympanic bulla. This lobe must be the posterior sinus sensu stricto (Fraser and Purves, 1960:9). Accordingly, the subtle fossa that lies more ventrally (more distally), near the apex of the paroccipital pro­ cess, is the posterior sinus fossa sensu stricto, whereas the cav­ ity that lies more dorsally in the paroccipital process is proba­ bly for a lobe of the peribullary sinus. Overall, structures seen in S. rayi give no new insight into the function of the pterygoid sinus system, but they indicate that the basicranial sinuses were comparable in structure and hence function to those of extant odontocetes. EAR, HEARING, AND BASICRANIAL CIRCULATION.—Com­ ment on these aspects is limited, because the periotic is indif­ ferently preserved and obscures the adjacent squamosal, and because most of the tympanic bulla is missing. The bony groove for the external auditory meatus is primitively larger than the narrow vestigial cleft of most extant odontocetes. Squamosal-bulla-periotic contacts are similar to those of Waipatia maerewhenua, Notocetus vanbenedeni Moreno, and the extant Platanista gangetica (Roxburgh) (see Muizon, 1987; Fordyce, 1994), judging from the amastoid skull wall and from contact relationships of the posterior process of the bulla with the squamosal at both the external auditory meatus and post- tympanic process. It is notable that, in all extant and fossil od­ ontocetes studied during this project, the posterior meatal crest of the squamosal articulates with the anterior face of the poste­ rior process of the tympanic bulla. The articulation is present even in Delphinidae (although very reduced); delphinids com­ monly are regarded as having the tympanoperiotic disarticu­ lated from the adjacent skull. Some inferences may be made about cranial circulation in S. rayi, but with important provisos: (1) Details of arterial circula­ tion are known reliably for only a few extant species of ceta­ ceans (see below). (2) General features of venous circulation were described by Fraser and Purves (1960; see also their sum­ mary of Boenninghaus, 1904), but there are no detailed modern accounts of paths of individual vessels from the cranial cavity to the basicranium. (3) There seem to be no accounts of the rel­ ative contributions of nerves, arteries, and veins for any one of the major cranial foramina in odontocetes, so that it is difficult to infer vascular or neural anatomy for fossils from the form of cranial foramina alone. Comments below focus on the brain­ case; the face, orbit, and rostrum are not considered. In extant Cetacea, the main arterial supply to the brain ap­ pears to be from a thoracico-spinal rete via spinal meningeal arteries that enter the foramen magnum. The internal carotid ar­ tery is small (Fraser and Purves, 1960). In fetal Physeter cat- odon Linnaeus, the internal carotid enters the cranium to partic­ ipate in the carotid rete mirabile (Melnikov, 1997), but in adult Physeter catodon, Tursiops truncatus, and Monodon monoc- eros Linnaeus the vessel is occluded (McFarland et al., 1979; Vogl and Fisher, 1981; Melnikov, 1997). Nonetheless, a patent carotid foramen is a persistent feature in Cetacea. The role of arteries other than the spinal meningeal and internal carotid is not clear; Fraser and Purves (1960:26) noted that, in one in­ jected unnamed odontocete, the middle meningeal artery, a branch of the internal maxillary artery, was involved in intra­ cranial supply. Cranial drainage in Odontoceti is understood less well. Fraser and Purves (1960) identified the pterygoid vein, supple­ mented by the internal maxillary and the internal jugular veins, as important in draining the basicranium. Apparently, none of these veins is associated with foramina that might otherwise be used to judge venous size and function. Furthermore, some 214 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY vessels that are intracranial in other mammals are extracranial in extant odontocetes (Fraser and Purves, 1960, fig. 13), through movement of the tympanoperiotic away from the braincase. Now-extracranial features in extant species, and pre­ sumably in S. rayi, include the ventral and dorsal petrosal venous sinuses and the cavernous sinus. There is no petro-oc- cipital canal for the ventral petrosal sinus, and the squamosal and/or parietal occludes the posterior lacerate foramen dorsal to the periotic (e.g., Fraser and Purves, 1960, pis. 13, 19, 26, 28). Of other venous foramina, the mastoid foramen, for the oc­ cipital emissary vein, is obscured or lost through change in contribution of the periotic to the braincase wall. The structure of the ethmoid foramen, orbital fissure, and foramen ovale in S. rayi are unrevealing about venous drainage. In non-cetacean mammals, the most notable foramen that has a venous function alone is probably the postglenoid foramen (also known as the retroarticular, retroglenoid, temporal, or spurious jugular foramen). This lies at the ventral opening of the temporal canal, for passage of the temporal sinus (or dorsal cerebral vein) from the transverse sinus (Sisson and Grossman, 1953; Whitmore, 1953; Padget, 1957, pl. 6; Evans, 1993). In non-cetacean mammals, including Artiodactyla (the presumed sister group to Cetacea), the temporal canal lies wholly within the squamosal. A comparable foramen is absent in Odontoceti and Mysticeti, suggesting that the temporal sinus is lost. It could be claimed that the pathway that Fordyce (1994) identi­ fied in W. maerewhenua (and, here, in S. rayi) as that for the middle meningeal artery actually represents the path of the temporal sinus and, thus, the temporal canal and postglenoid foramen, but this seems unlikely. First, in archaeocetes, a vesti­ gial (non-patent) postglenoid foramen lies at the base of the postglenoid process on the squamosal, posterolateral to the well-developed foramen between the periotic and squamosal that is inferred to mark the path for the middle meningeal ar­ tery. Second, the temporal canal in non-cetacean mammals lies within the squamosal, whereas the pathway for the presumed middle meningeal artery in W. maerewhenua is a groove on the ventromedial surface of the squamosal, dorsal and lateral to the periotic and associated with a small subcircular fossa (Fordyce, 1994). Some features of S. rayi are similar to those of W. maere­ whenua. A prominent foramen opens between the squamosal and the lateral wall of the periotic at the fovea epitubaria. More internally (cranially), a foramen, presumably for the middle meningeal artery, lies at the posterior of the parietal-alisphe- noid suture near the apex of the anterior process of the periotic; this foramen is presumed to be the homolog of the subcircular fossa as seen in Squalodelphinidae and, as a small cavity, in W. maerewhenua. Muizon (1994:137) discussed the broader sig­ nificance of the subcircular fossa in Platanistoidea, identifying it as present in the squamosal of some species of Squalodon­ tidae, but that homology is questionable. In one well-preserved specimen referred to Squalodon calvertensis Kellogg (USNM 23537), a large foramen is present in the ventral surface of the squamosal dorsal to the periotic, but it is directed dorsally into the squamosal; it is not associated with sutures with adjacent bones, and it lacks an associated groove leading toward the fo­ ramen ovale. Such a situation contrasts with the combination of foramen (foramen spinosum) and fissure (path for presumed middle meningeal artery, associated with the parietal-alisphe- noid-squamosal) that runs toward the foramen ovale from above the periotic in Waipatia (Waipatiidae) and from the sub- circular fossa above the periotic in Notocetus (Squalodelphin­ idae) and Zarhachis (Platanistidae). In Platanista, the foramen opens in the squamosal lateral to the periotic, which is visible in ventral view. The feature in S. calvertensis thus is probably not related to the subcircular fossa; it could be a nutrient fora­ men for the squamosal, or it may indicate venous drainage of the posterior of the temporal fossa. Because of its position dor­ sal to the periotic, it is unlikely that this is the temporal canal. Comparable foramina occur sporadically in other odontocetes, but their taxonomic patterns and function are uncertain. Other basicranial foramina on odontocete skulls also are problematic. Kellogg (1925) noted small foramina in the basic­ ranium of Zarhachis flagellator (Cope), and possibly homolo­ gous foramina occur in W. maerewhenua (see Fordyce, 1994, fig. 8, foramina 1 and 2). Ridewood (1922) named the squamo­ sal cleft in some mysticetes, where it is a deep fissure that ex­ tends ventrally from the temporal fossa to a point on the squa­ mosal opposite the periotic. Comparable features in some extant odontocetes (Monodontidae) are associated with the squamosal/parietal suture in the temporal fossa and may indi­ cate venous drainage; such a pathway is not seen in S. rayi. Phylogenetic Relationships CLADISTIC ANALYSIS.—Simocetus rayi shows derived struc­ tures representative of the Odontoceti, and it is not an archaeo- cete or a mysticete. Further, the overall primitive condition of many features in S. rayi suggests a basal position in the clade Odontoceti. Traditionally, basal odontocetes are placed in the family Agorophiidae, a group reviewed below. Archaic form in an organism, however, does not rule out relationships with a more-crownward taxon, and with this in mind extensive point- by-point comparisons of S. rayi were made with representa­ tives of all other major odontocete clades. No single features or, notably, structural complexes, were identified that placed S. rayi convincingly close to one or more of the Physeteroidea, Ziphiidae, Delphinida, Eurhinodelphinidae, or Platanistoidea. A few ambiguous similarities between S. rayi and some of the latter taxa (e.g., toothless premaxilla, elongate conical ptery­ goid hamulus) are of uncertain value in placing Simocetus (see below). To explore relationships further, S. rayi was included in com­ puter-assisted cladistic analyses of odontocetes, which used the taxa, characters, and matrix of Fordyce (1994) as a starting point and added the few derived features (such as toothless pre­ maxilla, elongate conical pterygoid hamulus, and persistent in- NUMBER 93 215 terparietal) that S. rayi shares with some more-crownward od­ ontocetes. (Although comparisons are needed with other archaic "agorophiid" odontocetes, none of the latter provide enough information to warrant their inclusion in a cladistic analysis; see below.) In all analyses, S. rayi plotted consistently toward the base of the Odontoceti. On the cladogram of Fordyce (1994, fig. 15), it plotted immediately crownward from Archaeodelphispatrius but stemward from representative Physeteroidea, Ziphiidae, Delphinida, Eurhinodelphinidae, and Platanistoidea. Figure 18 summarizes the latter relationships. Other than positioning S. rayi toward the base of the odonto­ cetes, the cladistic analyses added no new information about patterns among more-crownward odontocetes. For this reason, the characters, procedures, and results offered by Fordyce (1994) are not presented again herein, and the cladistic analysis is not taken any further. As mentioned above, archaic odontocetes comparable with Simocetus previously have been put in the widely debated fam­ ily Agorophiidae. On this point, several issues immediately re­ quire attention. Should the Agorophiidae be used as a clade based upon Agorophius pygmaeus, as Fordyce (1981) sug­ gested? Or should Agorophiidae be used more broadly (e.g., Whitmore and Sanders, 1977) as a grade, forming a paraphyl­ etic and perhaps polyphyletic receptacle for enigmatic archaic odontocetes? The latter seems undesirable, although S. rayi is still compared below with other described "agorophiid" odon­ tocetes. On broader relationships, some authors have suggested that the family Agorophiidae belongs with the Squalodontidae in a superfamily Squalodontoidea and, further, that the Squalodon­ tidae may be a near-basal group of odontocetes whence some living groups evolved (e.g., Abel, 1913:221; Simpson, 1945; Slijper, 1979, fig. 36). Rothausen (1968) recognized the desir­ ability of a cladistic approach to the Squalodontidae, and Muizon (1987, 1988a, 1988b, 1991, 1994) later convincingly argued that squalodontids are related closely to Platanista and other Platanistoidea. The conclusion herein follows Muizon (and also Fordyce, 1994), and Squalodontoidea is regarded as a synonym of Platanistoidea. There is no reason to think that S. rayi is related closely to Squalodontidae. Relationships be­ tween A. pygmaeus and Squalodontidae have yet to be re­ solved. Finally, despite the basal position of Simocetus rayi, the spe­ cies does have a few highly specialized features of the feeding apparatus. These features preclude S. rayi from being ancestral to any other described species and indicate that it was an early side branch in odontocete history. For such reasons the species is placed in a new family. All these points are elaborated below. Simocetus As AN ODONTOCETE.—Simocetus rayi shows a range of features unique to odontocetes, including a posteriorly telescoped supraorbital (ascending) process of the maxilla that is broadened laterally behind the level of the antorbital notch, a posteriorly placed dorsal infraorbital (maxillary) foramen, and the presence of a premaxillary sac fossa, premaxillary sulci, Zygorhiza Archaeodelphis Simocetus Physeter Kogia Ziphiidae Eurhinodelphidae Kentriodon Pontoporia Delphinidae Prosqualodon Squalodontidae Waipatia Platanistidae Squalodelphidae FIGURE 18.—Cladogram showing inferred relationship of Simocetus rayi to other major groups of Odontoceti. ( t= extinct taxa.) and premaxillary foramen (or foramina). In the orbit, the in­ fraorbital process of the maxilla is greatly reduced, rather than conspicuous (archaeocetes) or large and plate-like (mysticetes). At the posterior of the face, the orbitotemporal crest is dis­ placed posterodorsally relative to the postorbital ridge, so that the posterior of the face partly roofs the temporal fossa. In con­ trast, the orbitotemporal crest in archaeocetes (and in the odon­ tocete family Physeteridae) lies immediately dorsal to the pos­ torbital ridge, whereas in mysticetes the orbitotemporal crest is lost or migrates anterodorsally. The distinct antorbital notch forms a vertical groove that faces anteriorly to anterolaterally; this grooved notch is probably a concomitant of the facial fossa developed above the orbit (see comparisons with the archaic odontocetes Archaeodelphis and Xenorophus, below). An ossi­ fied mesethmoid lies between the bony nares. In the basicra­ nium, the parietal lies dorsomedial to the periotic to occlude the cranial hiatus, and the squamosal is presumed to lie above the periotic. Finally, the region of the mandibular foramen is relatively large and ventrally inflated. SIMILARITIES WITH MYSTICETI.—Odontocetes and mystice­ tes do share features not seen in archaeocetes, pointing to their sister-group relationship. Some of these features are conspicu­ ous in S. rayi. For example, a cluster of dorsal infraorbital fo­ ramina opens at the base of the rostrum in S. rayi and in the ar­ chaic mysticete Mammalodon colliveri. The mesorostral groove in S. rayi and mysticetes is well developed anteriorly on the rostrum, but there is no anterior groove in archaeocetes; rather, the premaxillae contact each other here at a planar me­ dian suture. Simocetus rayi, other odontocetes, and Mysticeti are amastoid, with the posterior (mastoid) process of the peri- 216 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY otic not exposed laterally on the skull wall. This may be linked with the reduced or lost contact between the periotic and exoc­ cipital, which, in turn, perhaps reflects the development of the peribullary sinus. Teeth in odontocetes and toothed mysticetes lie anterior to the antorbital notch, but in all described odonto­ cetes they are farther forward than in archaic mysticetes of the genus Aetiocetus (see Barnes et al., 1995). The short mandibu­ lar symphysis in S. rayi is reminiscent of that seen in mystice­ tes, but the structure in S. rayi is more extensive anteroposteri­ orly. As in mysticetes, there is an indistinct longitudinal groove toward the ventral surface of the symphysis, but whether the groove is homoplasious or synapomorphic is uncertain. The large mandibular foramen and its associated thin lateral wall ("panbone") are similar between odontocetes and archaic mys­ ticetes, apart from the ventral inflation seen in odontocetes. Comparable structures occur in basilosaurid archaeocetes; more study of homology and function is needed. CURRENT CONCEPTS AND PROBLEMS OF ODONTOCETE PHY­ LOGENY.—Recent cladistic analyses by Muizon (1987, 1988a, 1988b, 1991), Heyning (1989, 1997), Barnes (1990), and For­ dyce (1994) help to elucidate family-level patterns within the Odontoceti and help to interpret the relationships of S. rayi. Sperm whales, Physeteroidea (Physeteridae and Kogiidae), are odontocetes (cf. Milinkovitch, 1995) that form a sister group to Ziphiidae. Physeteroidea and Ziphiidae together represent the Physeterida (see Muizon, 1991; Fordyce, 1994). Alternatively, sperm whales may constitute a sister group to all other odonto­ cetes (Heyning, 1989, 1997; Barnes, 1990). The Platanistoidea (sensu Muizon, 1987, 1991; Fordyce 1994) encompasses Pla- tanistidae, Squalodelphinidae, Waipatiidae, Squalodontidae, and perhaps Dalpiazinidae. Delphinida (sensu Muizon, 1988b) includes the Delphinoidea (Delphinidae, Phocoenidae, Mon- odontidae, Albireonidae, and Kentriodontidae) along with some of the "river dolphins" (Iniidae, Pontoporiidae, and the uncertainly distinct Lipotidae). Less clear are the relationships of Eurhinodelphinidae and Eoplatanistidae, which Muizon (1991) included in Eurhinodelphinoidea, a sister group to Del­ phinida. Despite recent efforts, the cladistic relationships between the major groups Physeteroidea, Ziphiidae, Delphinida, Eurhino­ delphinoidea, and Platanistoidea seem weakly resolved. Char­ acters cited by Muizon and Barnes in support of their cla­ dograms are debatable in terms of distribution among taxa (homoplasy was not always identified), applicability (a state was not always seen in all members of a clade), and polarity (the primitive versus derived states were not always clear, and transition series were not always clear-cut). Fordyce (1994) at­ tempted to circumvent such problems through a computer anal­ ysis of an explicit matrix of characters, but even then (Fordyce, 1994, fig. 15) some of the key nodes among the Odontoceti were supported by rather few characters including reversals and features of debatable polarity (e.g., ossified lateral lamina of pterygoid sinus fossa). Heyning (1989, without data matrix; 1997, with data presented) considered only extant odontocetes, yet fossils in general are known to influence cladograms pro­ foundly (Donoghue et al., 1989). Is there hope to get beyond this situation by using traditional anatomy, as opposed to mo­ lecular phylogeny? The described species of odontocetes, fossil and recent, have been studied in such detail and for so long that it is difficult to imagine new insights into higher relationships arising from more studies of described material. Rather, new discoveries of well-preserved fossil skulls will more likely pro­ vide the key. This is particularly the case for Oligocene fossils; material in the largely unstudied Emlong collection offers great promise. FAMILY AGOROPHIIDAE.—The Family Agorophiidae has long been considered to encompass the phylogenetically and geologically oldest odontocetes. Supposed diagnostic features of the family (e.g., Kellogg, 1923b, and other references re­ viewed by Fordyce, 1981) include the presence of heterodont teeth and a primitively telescoped skull with a large intertem­ poral constriction. Such features occur in Simocetus, which might, therefore, be viewed as an agorophiid. These supposed diagnostic features of agorophiids, however, are plesiomor- phies of no value in assessing immediate relationships. To judge the relationship between Simocetus and Agorophiidae, the latter family first should be rediagnosed on the basis of de­ rived characters. Any rediagnosis must be based initially upon Agorophius pygmaeus (Muller, 1849), which is the type species and only species in the genus Agorophius Cope, 1895, and in turn is the type genus of the family Agorophiidae Abel, 1913. Agorophius pygmaeus is known with certainty only from the holotype skull, from the Cooper Marl (Chattian, late Oli­ gocene) of South Carolina (Whitmore and Sanders, 1977). No other described fossils undoubtedly belong to this species or in Agorophius, although Albert E. Sanders recently collected an apparently conspecific skull. One species possibly close to A. pygmaeus is "Squalodon (Microzeuglodon?)" wingei Ravn, 1926, the hypodigm of which consists of teeth, a bulla, and un­ described skull fragments from the upper Oligocene of Den­ mark. (Rothausen (1970) indicated that this species does not belong to Squalodon, and used the new generic name "Oli- gosqualodon," a nomen nudum, for the species.) Like A. pyg­ maeus, "Squalodon (Microzeuglodoniy wingei has a cheek tooth with a high crown, indistinctly elevated denticles, and limited ornament. Relationships between these species could be elucidated if other topotypic material is found. The holotype skull of A. pygmaeus is lost, and only a cheek tooth remains (specimen MCZ 8761; Fordyce, 1981), but some details of the lost skull appear in a lithograph (True, 1907, pl. 6). Clearly, Agorophius pygmaeus is an archaic odontocete. The lithograph shows overall profiles and some sutures and fo­ ramina; the skull has a prominent supraorbital process of the maxilla, a marked intertemporal constriction, a robust zygo­ matic process, and a moderately large, denticulate, smooth, high-crowned cheek tooth. The basicranium, tympanoperiotic, most teeth, and the mandible appear to be missing, however, and few or no details can be seen for the supraorbital process of NUMBER 93 217 the maxilla, dorsal infraorbital foramen, premaxillary sac fossa and sulci, anterior of rostrum, antorbital notch, lacrimal, nasals, orbit, and palate. Overall, the lithograph seems good enough to judge whether any new topotypic fossils might be conspecific with the lost holotype skull, and it allows limited comparisons with S. rayi and other archaic odontocetes. It appears, however, not to provide enough information to allow the Agorophiidae to be rediagnosed in a cladistic sense. Should we then maintain a grade family Agorophiidae for archaic odontocetes such as Xenorophus sloani, Archaeodelphis patrius, and Simocetus rayP. Perhaps not; although grade families offer easy classifi­ cation of incomplete specimens, and limit the numbers of mo­ notypic high-level basal clades, they also obscure relationships and thus oversimplify phylogenetic history. For now, the broad traditional use of Agorophiidae is not followed, and the family is used herein only for A. pygmaeus. Beyond Agorophius pygmaeus, odontocetes previously re­ ferred to the Agorophiidae include Archaeodelphis patrius Allen (1921), Atropatenocetus posteocenicus Aslanova (1977), Microzeuglodon caucasicum (Lydekker, 1893), Mirocetus ria- binini Mchedlidze (1970), and Xenorophus sloani Kellogg (1923b). Kellogg (1923a), Simpson (1945), Whitmore and Sanders (1977), Barnes (1978), and Fordyce (1981), among others, have commented on the family placement of these spe­ cies. Comparisons with all, below, are based largely upon pub­ lished accounts. COMPARISONS OF Simocetus AND Agorophius.—Conspicu­ ous primitive (plesiomorphic) features on the skull of S. rayi provide superficial similarity with A. pygmaeus, and, indeed, Muizon (1991:303) identified USNM 256517 (the holotype of S. rayi) as belonging in Agorophius. Shared primitive features include a facial fossa that is only moderately developed, a large temporal fossa open to dorsal view rather than roofed by adja­ cent bones, an intertemporal constriction with the parietals ex­ posed dorsally, prominent nuchal crests, and heterodont teeth. (S. rayi does show other plesiomorphies, such as the broadly exposed palatine and a pterygoid sinus fossa that is restricted to the basicranium, but comparable features cannot be seen in A. pygmaeus.) More importantly, S. rayi is more derived than A. pygmaeus in some features. Its rostrum is quite dorsoventrally com­ pressed and anteriorly deflected, with a transversely convex ventral surface, whereas that of A. pygmaeus appears to be straight. The braincase is more inflated at the base of the zygo­ matic process, and there is a deep cleft between the zygomatic process and the braincase. The outline of the supraoccipital is more hemispherical (dorsal view), the postglenoid process is thicker (lateral view), and the exoccipital extends farther later­ ally. Notably, Agorophius pygmaeus is more derived than S. rayi in its more laterally expanded supraorbital process of the maxilla and greater number of maxillary teeth (eight or more). Comparisons of teeth also reveal differences, although it is not easy to judge their taxonomic significance. Mandibular cheek teeth in S. rayi differ from the one upper middle cheek tooth of A. pygmaeus in the (presumably) derived states of smaller absolute size, relatively lower crowns, relatively smaller main denticle and more free-standing accessory denti­ cles, and more transversely compressed crowns. In S. rayi, also, tooth crowns are more asymmetrical (lateral view) and or­ namented, with the main denticles displaced relatively anteri­ orly, but whether these conditions are primitive or derived is uncertain. In summary, the plesiomorphies seen in these two species do not demonstrate close relationships. Furthermore, each species shows specialized features that seem to rule out close relation­ ships. For this reason, S. rayi is excluded from the Agorophi­ idae. Differences with other archaic odontocetes, elaborated below, reinforce the suggestion that S. rayi belongs in its own family Simocetidae. COMPARISONS OF Simocetus AND Archaeodelphis.—The enigmatic, monotypic small species Archaeodelphis patrius Allen, 1921, is perhaps the most archaic odontocete described. Whitmore and Sanders (1977:305) assigned it to the Cetacea incertae sedis, whereas Fordyce (1994, fig. 15) placed it at the base of the Odontoceti. Its age is possibly Oligocene (Whit­ more and Sanders, 1977). Brief examination of the holotype and only specimen of^L patrius (MCZ 15749; skull lacking rostrum and bullae) re­ vealed notable differences from S. rayi. Simocetus rayi is more derived in having facial structures more posteriorly displaced (facial fossa, supraorbital process of the maxilla, an ascending process of the premaxilla, and dorsal maxillary foramina), a posteriorly bifurcated premaxilla, a more elevated lateral mar­ gin of the rostrum, the relatively shorter nasals, a relatively longer and more delicate postorbital process of the frontal, a ro­ bust and transversely thickened zygomatic process of the squa­ mosal, larger pterygoid sinus fossae, long, medially apposed pterygoid hamuli into which the pterygoid sinuses extend, a more prominent basioccipital crest, a more anteroposteriorly thickened postglenoid process, and a more posterodorsally dis­ played orbitotemporal crest on the frontal. Despite its overall archaic form, Archaeodelphis patrius seems quite specialized in its relatively large lacrimal, deep medial cleft between the palatines posteriorly, and thick, plate­ like extensions of the medial lamina of the pterygoid that meet medially to roof the choanae. As in most other odontocetes (cf. S. rayi), the foramen ovale is not confluent with the posterior lacerate foramen. The structure of the medial lamina of the pterygoid is so unusual that, despite its archaic structure,^. patrius is probably not ancestral to any other known odonto­ cete, although perhaps the enlarged lacrimal indicates affinity with Xenorophus sloani. There is no evidence to support a close relationship with S. rayi. COMPARISONS OF Simocetus AND Xenorophus.—The late Oligocene species Xenorophus sloani Kellogg (1923b) is known from the holotype (USNM 11049), an incomplete skull that lacks the tip of the rostrum, the region around the premax­ illary sac fossa, and the braincase posterior to the orbits. Other 218 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY specimens referable to Xenorophus were mentioned and fig­ ured by Whitmore and Sanders (1977), but they have not been described formally. Comparisons below are based upon study of the holotype skull and upon published illustrations (Miller, 1923, pl. 5: fig. 6; Whitmore and Sanders, 1977, fig. la). There are major differences between S. rayi andX. sloani. As with A. patrius, S. rayi is more derived in its posteriorly dis­ placed facial structures (facial fossa, supraorbital process of the maxilla, ascending process of the premaxilla, and posterodor­ sally oriented dorsal maxillary foramina), posteriorly bifur­ cated premaxilla, and more elevated lateral margin of the ros­ trum (resulting in a less abruptly elevated orbit). The antorbital notch is more prominent (dorsoventrally more shallow, antero­ posteriorly deeper), the premaxillary sac fossa is broader and more tabular (although the premaxilla in X. sloani is broken, a narrow long fossa is indicated), the premaxillary sulci are more prominent, and the anterior of the mesorostral groove is more open. The cheek teeth are relatively smaller, more transversely compressed, and more emergent from the alveoli, and they have large diastemata. The palate is ventrally convex rather than flat in transverse profile and the anterior of the rostrum is dorsoventrally thin and deflected ventrally (what remains in X. sloani does not indicate deflection). The rostrum is relatively shorter, wider, and less abruptly attenuated, and ventrally, the posterior of the palate is relatively wider and less abruptly at­ tenuated. The postorbital process of the frontal is relatively longer and more delicate, and the intertemporal area is shorter and wider. Finally, the supraoccipital is produced more anteri­ orly. Despite its generally archaic form, Xenorophus sloani shows some intriguing derived features. A very large lacrimal domi­ nates the preorbital and supraorbital parts of the cranium, and the supraorbital process of the maxilla is large and produced far posteriorly. Also notable are the prominent prenarial constric­ tion, nares that open well behind the level of the antorbital notch, multiple cheek teeth with a triangular low crown and multiple small denticles, and a relatively short exposure of the frontals on the vertex. Lacrimal size alone seems to rule out af­ finities with any more-crownward odontocete, and there is no evidence to support a close relationship with S. rayi. COMPARISONS OF Simocetus AND Atropatenocetus.—Atro- patenocetus posteocenicus Aslanova, 1977, was described as a new genus and species of Agorophiidae. The holotype and only described specimen is a quite incomplete skull with fragmen­ tary mandibles presumably from the upper Oligocene of Ap­ sheron Peninsula, Azerbaijan. The supraorbital process of the maxilla appears to be displaced posterodorsally over the fron­ tal. The cheek teeth are relatively larger and lower crowned than S. rayi, and they possess papillate ornament on the cin­ gula. Otherwise, more meaningful comparisons are difficult. There is no hint of features that would place this species close to Simocetus rayi or, for that matter, Agorophius pygmaeus. COMPARISONS OF Simocetus AND Microzeuglodon.—Mi- crozeuglodon caucasicum (Lydekker, 1893) is known only from the holotype: a posterior fragment of the left mandible with four cheek teeth, an unfigured second fragment of the jaw with five broken teeth, a left humerus, and an incomplete cau­ dal vertebra, from an uncertain horizon in the Oligocene of Az­ erbaijan (Kellogg, 1923a; Mchedlidze, 1964). The species has been discussed widely without consensus as to its affinities. The large size of the mandible and teeth, relatively high crowns, prominent lingual cingula and level of insertion of the teeth, and relatively narrow diastemata suggest that M. cauca­ sicum is not related closely to S. rayi. COMPARISONS OF Simocetus AND Mirocetus.—Mirocetus riabinini Mchedlidze, 1970, is known only from the holotype skull and associated partial postcranial skeleton from the upper Oligocene of Caucasus. (Initially, Riabinin (1938) identified the holotype as Microzeuglodon aff. caucasicus.) The skull is distorted, albeit with basic topography preserved (Riabinin, 1938), but sutural details are unclear (K. Rothausen, pers. comm., 1980). Of note, the skull appears to possess posteriorly displaced facial fossae, which suggests that M. riabinini is an odontocete, rather than a species of Aetiocetidae (cf. Mchedlidze, 1976). Some features in M. riabinini suggest at least that it is not conspecific with S. rayi; for example, the skull and teeth are notably larger, the rostrum is more narrow (plesiomorphy), the supraorbital process extends farther later­ ally, the postorbital process is robust with a widened apex, a sagittal crest appears to be present (plesiomorphy), the zygo­ matic process is relatively shorter and possesses a more re­ curved ventral surface (plesiomorphy), the braincase is not markedly inflated anteroexternally, the supraoccipital apex is sharp, and the teeth are not inserted at the lateral edge of the rostrum. Mirocetus riabinini is similar to S. rayi in that the cheek teeth appear to be emergent, with parallel roots united by an isthmus, and diastemata are prominent. No features clearly indicate close relationship between M. riabinini and S. rayi or, for that matter, A. pygmaeus. COMPARISONS WITH OTHER HETERODONT ODONTOCETES.— Other named species of heterodont odontocetes might be com­ pared with S. rayi. Among these are many so-called Squal­ odontidae based upon isolated teeth; examples from the middle Tertiary of the North American Atlantic Coastal Plain include "Phoca'1'' debilis, "Phoca''' modesta, Colophonodon holmesii, and "Squalodon'' protervus. Arguably, such names based upon isolated teeth are nomina dubia, although it is always possible that such teeth will be found in place in topotypic skulls, which, in turn, might elucidate relationships. For now, these taxa cannot be compared usefully with S. rayi. RELATIONSHIPS WITH OTHER GROUPS O F ODONTOCETES.— A few other comparisons warrant attention. Although S. rayi is dramatically different from Eurhinodelphinidae in overall skull form, it is similar to some eurhinodelphinids in having a tooth­ less premaxilla, an elongate conical pterygoid hamulus, a thick postglenoid process, and a recurved anteroexternal sulcus on the periotic. In all species of Eurhinodelphis (e.g., E. bossi Kellogg, 1925), the toothless premaxillae are elongate, forming NUMBER 93 219 a tapered distal part to the rostrum similar to that of longirostral teleosts, and in marked contrast to the short flattened distal toothless premaxillae of S. rayi. Simocetus rayi has elongate conical pterygoid hamuli, similar to those of Argyrocetus joaquinensis (figured in Kellogg, 1932). Elongate subcylindri- cal hamuli also occur in an unnamed archaic odontocete USNM 243979 (the so-called non-squalodontid odontocete of Whitmore and Sanders, 1977, fig. 2a, from ?Pysht Formation, Washington) that is not demonstrably a eurhinodelphinid. Ha- mular structure is known too poorly among basal odontocetes to allow reliable use in higher taxonomy. The postglenoid pro­ cess in S. rayi is anteroposteriorly thick, as seen also in species of Eurhinodelphis (e.g., Kellogg, 1925). A thick postglenoid process, however, also occurs in the enigmatic Squaloziphius emlongi Muizon, 1991 (a possible eurhinodelphinid), and in some ziphiids; the functional significance of such thickening is uncertain. On the periotic, the ventral part of the anteroexternal sulcus is recurved forward, as seen in some eurhinodelphinids (Fordyce, 1983b), but the sulcus is shallow and indistinct in contrast to the deep sulcus of eurhinodelphinids. In this respect, the anteroexternal sulcus is more similar to that of the platanis- toid groups Waipatia and Squalodelphinidae. For now, eurhino­ delphinid relationships cannot be ruled out, but they are not supported strongly. Simocetus rayi is similar to some extant Delphinidae (such as Orcaella brevirostris (Owen)) in having an identifiable or in­ completely sutured interparietal in the adult. Sutures between the interparietal and adjacent elements sometimes persist in adults of other species of delphinid. This condition may be in­ terpreted as paedomorphic. An interparietal is distinct also in late fetal stages of some mysticetes (Ridewood, 1922). Many delphinids, as with S. rayi, also have a short unfused mandibu­ lar symphysis. Some delphinids (e.g., Grampus griseus) also have a palate with scattered multiple small palatine foramina, although this condition also is seen in Squalodontidae. Such features might indicate a relationship with Delphinidae or other Delphinoidea, but more convincing similarities (involving, e.g., the pterygoid sinus complex, premaxillae, and periotic) are lacking. As with the platanistoids Waipatia maerewhenua and Squa­ lodelphinidae, S. rayi has a pronounced cleft—associated with or representing the foramen spinosum, and thus presumably for the middle meningeal artery—arising near the foramen ovale and trending posterolaterally toward the periotic along or near the parieto-squamosal suture. Even in the rather archaic W. maerewhenua, however, the foramen spinosum lies dorsal to and is obscured by the periotic, and the foramen in Squalodel­ phinidae is transformed into a large subcircular fossa, whereas in S. rayi a foramen opens medial to the periotic and is clearly visible in ventral view. The condition in S. rayi could be inter­ preted as trending toward that of some platanistoids. Similar forms of the anteroexternal sulcus of the periotic were noted above. Beyond these, there is little particular support for pla- tanistoid relationships. Summary Morphological comparisons do not clearly indicate relation­ ships. This was emphasized during cladistic analyses, when, despite the above similarities, S. rayi did not cluster with any particular more-crownward clade. A more-crownward position might have been forced by character weighting, or by invoking an ordered transition series for some supposed key characters, or by invoking accelerated character transformation, but such techniques seem inappropriate given the poor understanding of the anatomy of archaic odontocetes. Superimposed on the archaic skull of this peculiar ancient taxon are some features that are, unexpectedly, highly special­ ized: the premaxilla is toothless and dorsoventrally flattened, the rostrum is relatively short and broad, and the anterior of the rostrum and mandible are markedly downturned. Such at­ tributes indicate specialized feeding behavior. Other unusual features are plausibly but less clearly autapomorphies; these in­ clude the nonpolydont tooth complement (primitive, or second­ arily reduced from polydont state?), broad diastemata between posterior cheek teeth, narrow rather than laterally expanded su­ praorbital process of the maxilla (again, primitive, or second­ arily reduced from a broader state?), and the mediolaterally deep optic infundibulum. Previously published literature might give the impression that archaic odontocetes—"agorophiids"— are primitive overall, but Simocetus demonstrates that archaic taxa may indeed represent specialized side branches in odonto­ cete history that are not close to modern clades. Simocetus rayi nevertheless shows the distinctive facial structure seen in most other odontocetes, emphasizing the early development—pre­ sumably in the most archaic of the odontocetes—of the nasofa­ cial soft tissues and probably of echolocation abilities. Literature Cited Abel, 0 . 1913. Die Vorfahren der Bartenwale. Denkschriften der Akademie der Wis­ senschaften, Wien, Mathematisch-Naturwissenschaftliche Klasse, 90:155-224, 12 plates. Allen, G.M. 1921. A New Fossil Cetacean (Archaeodelphis patrius gen. et sp. nov.). Bulletin of the Museum of Comparative Zoology at Harvard Col­ lege. 65:1-14. Aslanova, S.M. 1977. [A New Genus of Cetaceans (Atropatenocetus posteocenicus gen. et sp. nov.) from the Oligocene of Azerbaijan.] Doklady Akademia Nauk Azerbaidzhanskoi SSR. 33:60-64. [In Russian.] 220 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY Barnes, L.G. 1978. A Review of Lophocetus and Liolithax and Their Relationships to the Delphinoid Family Kentriodontidae (Cetacea: Odontoceti). Sci­ ence Bulletin of the Natural History Museum of Los Angeles County, 28:1-35. 1990. The Fossil Record and Evolutionary Relationships of the Genus Tursiops. In S. Leatherwood and R.R. Reeves, editors, The Bottle­ nose Dolphin, pages 3-26. San Diego: Academic Press. Barnes, L.G., M. Kimura, H. Furusawa, and H. Sawamura 1995. Classification and Distribution of Oligocene Aetiocetidae (Cetacea; Mysticeti) from Western North America and Japan. The Island Arc, 3(4):392-431. Boenninghaus, G. 1904. Das Ohr des Zahnwales. Zoologische Jahrbucher, Abteilung fiir Anatomie und Ontogenie der Tiere, 19:189-360. Breathnach, A.S. 1960. The Cetacean Nervous System. Biological Review, 35:187-230. Brownell, R.L., and E.S. Herald 1972. Lipotes vexillifer. Mammalian Species, 10: 4 pages. Cope, E.D. 1895. Fourth Contribution to the Marine Fauna of the Miocene Period of the United States. Proceedings of the American Philosophical Soci­ ety, 34:135-155. Cranford, T.W., M. Amundin, and K.S. Norris 1996. Functional Morphology and Homology in the Odontocete Nasal Complex: Implications for Sound Generation. Journal of Morphol­ ogy, 228:223-285. Curry, B.E. 1992. Facial Anatomy and Potential Function of Facial Structures for Sound Production in the Harbor Porpoise (Phocoena phocoena) and Dall's Porpoise (Phocoenoides dalli). Canadian Journal of Zoology, 70:2103-2114. Dart, R.A. 1923. The Brain of the Zeuglodontidae (Cetacea). Proceedings of the Zoo­ logical Society of London, 1923:615-648. Davis, D.D. 1964. The Giant Panda: A Morphological Study of Evolutionary Mecha­ nisms. Fieldiana, Zoology Memoirs. 3:1-339. Domning, D.P. 1978. Sirenian Evolution in the North Pacific. University of California Publications in Geological Sciences, 118:1-176. Domning, D.P., C E . Ray, and M.C. McKenna 1986. Two New Oligocene Desmostylians and a Discussion of Tethythe­ rian Systematics. Smithsonian Contributions to Paleobiology, 59: 1-56. Donoghue, M.J., J.A. Doyle, J. Gauthier, A.G. Kluge, and T. Rowe 1989. The Importance of Fossils in Phylogeny Reconstruction. Annual Re­ views of Ecology and Systematics, 20:431—460. Edinger, T. 1955. Hearing and Smell in Cetacean History. Monatschrift fur Psychia­ tric und Neurologic 129:37-58. Evans, H.E., editor 1993. Miller's Anatomy of the Dog. Third edition, 1113 pages. Philadel­ phia: W.B. Saunders. Flynn, T.T. 1948. Description of Prosqualodon davidi Flynn, a Fossil Cetacean from Tasmania. Transactions of the Zoological Society of London, 25: 153-197. Fordyce, R.E, 1981. Systematics of the Odontocete Agorophius pygmaeus and the Fam­ ily Agorophiidae (Mammalia: Cetacea). Journal of Paleontology, 55:1028-1045. 1983a. Dental Anomaly in a Fossil Squalodont Dolphin from New Zealand, and the Evolution of Polydonty in Whales. New Zealand Journal of Zoology, 9:419-426. 1983b. Rhabdosteid Dolphins (Mammalia: Cetacea) from the Middle Mi­ ocene, Lake Frome Area, South Australia. Alcheringa, 7:27-40. 1994. Waipatia maerewhenua. New Genus and New Species (Waipatiidae, New Family), an Archaic Late Oligocene Dolphin (Cetacea: Odon­ toceti: Platanistoidea) from New Zealand. Proceedings of the San Diego Museum of Natural History, 29:147-176. Fraser, F.C, and P.E. Purves 1960. Hearing in Cetaceans: Evolution of the Accessory Air Sacs and the Structure of the Outer and Middle Ear in Recent Cetaceans. Bulletin of the British Museum (Natural History), Zoology, 7:1-140 pages, frontispiece, plates 1-53. Heyning, J.E. 1989. Comparative Facial Anatomy of Beaked Whales (Ziphiidae) and a Systematic Revision among the Families of Extant Odontoceti. Contributions in Science, Natural History Museum of Los Angeles County, 405:1-64. 1997. Sperm Whale Phylogeny Revisited: Analysis of the Morphological Evidence. Marine Mammal Science, 13(4):596—613. Heyning, J.E., and J.G. Mead 1990. Evolution of the Nasal Anatomy of Cetaceans. In J. Thomas and R. Kastelein, editors, Sensory Abilities of Cetaceans, pages 67-69. New York: Plenum Press. Howell, A.B. 1927. Contribution to the Anatomy of the Chinese Finless Porpoise Neo- meris phocaenoides. Proceedings of the United States National Mu­ seum, 70(13): 43 pages [publication 2662]. Huber, E. 1930. Evolution of Facial Musculature and Cutaneous Field of Trigemi­ nus. Quarterly Review of Biology, 5:133-187. 1934. Anatomical Notes on Pinnipedia and Cetacea. Carnegie Institution of Washington Publication, 447:105-136. Kasuya, T. 1973. Systematic Consideration of Recent Toothed Whales Based on the Morphology of Tympano-Periotic Bone. Scientific Reports of the Whales Research Institute. Tokyo, 25:1-103. Kellogg, R. 1923a. Description of Two Squalodonts Recently Discovered in the Calvert Cliffs, Maryland; and Notes on the Shark-Toothed Dolphins. Pro­ ceedings of the United States National Museum, 62(6): 1-69. 1923b. Description of an Apparently New Toothed Cetacean from South Carolina. Smithsonian Miscellaneous Collections, 76(7): 7 pages, 2 plates. 1925. On the Occurrence of Remains of Fossil Porpoises of the Genus Eu- rhinodelphis in North America. Proceedings of the United Slates National Museum, 66(26): 1—40. 1926. Supplementary Observations on the Skull of the Fossil Porpoise Zarhachis flagellator. Proceedings of the United States National Museum, 67(28): 18 pages, 5 plates. 1928. History of Whales: Their Adaptation to Life in the Water. Quarterly Review of Biology, 3:29-76, 174-208. 1932. A Miocene Long-Beaked Porpoise from California. Smithsonian Miscellaneous Collections, 87(2): 11 pages, 4 plates. 1936. A Review of the Archaeoceti. Carnegie Institution of Washington Publication. 82: 366 pages, 37 plates. King, J.E. 1961. The Feeding Mechanism and Jaws of the Crabeater Seal (Lobodon carcinophagus). Mammalia, 25:462—466. Lawrence, B., and W.E. Schevill 1956. The Functional Anatomy of the Delphinid Nose. Bulletin of the Mu­ seum of Comparative Zoology at Harvard College, 114:103-151. Lydekker, R. 1893. On Zeuglodont and Other Cetacean Remains from the Tertiary of the Caucasus. Proceedings of the Zoological Society of London, 1892:558-564. NUMBER 93 221 McFarland, W.L., M.S. Jacobs, and P.J. Morgane 1979. Blood Supply to the Brain of the Dolphin, Tursiops truncatus, with Comparative Observations on Special Aspects of the Cerebrovascu­ lar Supply of Other Vertebrates. Neuroscience and Biobehavioural Reviews, supplement 1, 3:1-93. Mchedlidze, G.A. 1964. [Fossil Cetaceans of Caucasus.] 144 pages. Tbilisi, Georgia: Izda- tel'stvo Metsniereba. [In Russian.] 1970. [Some Features of the Historical Development of the Cetacea, Part L] 112 pages. Tbilisi, Georgia : Metsniereba. [In Russian.] 1976. [The Basic Features of the Paleobiological History of the Cetacea.] 136 pages. Tbilisi, Georgia: Metsniereba. [In Russian.] Mead, J.G. 1975. Anatomy of the External Nasal Passages and Facial Complex in the Delphinidae (Mammalia: Cetacea). Smithsonian Contributions to Zoology, 207:1-72. Melnikov V.V. 1997. The Arterial System of the Sperm Whale (Physeter macrocephalus). Journal of Morphology, 234( 1 ):37-50. Milinkovitch, M.C. 1995. Molecular Phylogeny of Cetaceans Prompts Revision of Morpho­ logical Transformations. Trends in Ecology and Evolution, 10(8): 328-334. Miller, G.S. 1923. The Telescoping of the Cetacean Skull. Smithsonian Miscellaneous Collections, 76(5): 70 pages, 8 plates. Moris, F. 1969. Etude anatomique de la region cephalique du marsouin Phocaena phocaena L. (Cetace: Odontocete). Mammalia, 33:666-726. Muller, J. 1849. Uber die fossilen Reste der Zeuglodonlen von Nordamerica, mit Rucksicht auf die europdischen Reste aus dieser familie. 38 pages. Berlin: Reimer. Muizon, C. de 1986. Un nouveau Phocoenidae (Odontoceti, Cetacea) du Miocene superieur de la Formation Pisco (Perou). Comptes Rendus de I'Academie des Sciences, series 2, 303:1509-1512. 1987. The Affinities of Notocetus vanbenedeni, an Early Miocene Pla- tanistoid (Cetacea, Mammalia) from Patagonia, Southern Argentina. American Museum Novitates. 2904: 27 pages. 1988a. Le polyphyletisme des Acrodelphidae, Odontocetes longirostres du Miocene europeen. Bulletin du Museum Nationale d'Histoire Na­ turelle (Paris), series 4, section C, 10(l):31-88. 1988b. Les relations phylogenetiques des Delphinida (Cetacea, Mammalia). Annates de Paleontologie, 74(4): 159-227. 1988c. Les vertebres fossiles de la Formation Pisco (Perou), Troisieme par- tie: Les odontocetes (Cetacea, Mammalia) du Miocene. Memoires de I 'Institut Francais d 'Etudes A ndines. 7 8:1 -244. 1991. A New Ziphiidae (Cetacea) from the Early Miocene of Washington State (USA) and Phylogenetic Analysis of the Major Groups of Odontocetes. Bulletin du Museum Nationale d'Histoire Naturelle, Paris, series 4, section C, 12(3^I):279-326. 1994. Are the Squalodonts Related to the Platanistoids? Proceedings of the San Diego Museum of Natural History, 29:135-146. Norris, K.S. 1968. The Evolution of Acoustic Mechanisms in Odontocete Cetaceans. In E.T. Drake, editor, Evolution and Environment, pages 297-324. New Haven: Yale University Press. Novacek, M.J. 1986. The Skull of Leptictid Insectivorans and the Higher-Level Classifi­ cation of Eutherian Mammals. Bulletin of the American Museum of Natural History, 183(1):1-112. Padget, D.H. 1957. The Development of the Cranial Venous System in Man, from the Viewpoint of Comparative Anatomy. Contributions to Embryology, Carnegie Institution of Washington, 36:79-140. Pen-in, W.F 1975. Variation of Spotted and Spinner Porpoise (Genus Stenella) in the Eastern Tropical Pacific and Hawaii. Bulletin of the Scripps Institu­ tion of Oceanography, 21:1 -206. Quay, W.B., and E.D. Mitchell 1971. Structure and Sensory Apparatus of Oral Remnants of the Nasopa­ latine Canals in the Fin Whale (Balaenoptera physalus L.). Journal of Morphology, 34:271-280. Ravn, J.P.J. 1926. On a Cetacean, Squalodon (Microzeuglodon!) wingei nov. sp., from the Oligocene of Jutland. Meddelelser fra Dansk Geologisk Foren- ing, 7:45-54. Ray, C E . 1976. Fossil Marine Mammals of Oregon. Systematic Zoology, 25:420- 436. [Date on title page is 1976; actually published in 1977.] 1980. [Obituary] Douglas Ralph Emlong 1942-1980. News Bulletin of the Society of Vertebrate Paleontology, 120:45—46. Riabinin, A.N. 1938. Microzeuglodon aff. caucasicum Lyd. iz verkne-Maikopskikh otloz­ henii Kabristana. Problemy Paleontologii, 4:137-185. Ridewood, W.G. 1922. Observations on the Skull in Foetal Specimens of Whales in the Genera Megaptera and Balaenoptera. Philosophical Transactions of the Royal Society of London, series B, 211:209-272. Rothausen, K. 1968. Die systematische Stellung der europaischen Squalodontidae (Odon­ toceti: Mamm.). Paldontologische Zeitschrift, 42:83-104. 1971. Marine Reptilia and Mammalia and the Problem of the Oligocene- Miocene Boundary. Giornale di Geologia, series 2, 35:181-189. Schenkkan, E.J. 1973. On the Comparative Anatomy and Function of the Nasal Tract in Odontocetes (Mammalia, Cetacea). Bijdragen Tot de Dierkunde, 43:127-159. Schenkkan, E.J., and P.E. Purves 1973. The Comparative Anatomy of the Nasal Tract and the Function of the Spermaceti Organ in the Physeteridae (Mammalia, Odontoceti). Bijdragen Tot de Dierkunde. 43( 1 ):93-l 12. Schulte, H. von, and M. de F. Smith 1918. The External Characters, Skeletal Muscles, and Peripheral Nerves of Kogia breviceps (Blainville). Bulletin of the American Museum of Natural History. 38:7-72. Simpson, G.G. 1945. The Principles of Classification, and a Classification of Mammals. Bulletin of the American Museum of Natural History, 85:1-350. Sisson, S., and J.D. Grossman 1953. The Anatomy of the Domestic Animals. Fourth edition, 972 pages. Philadelphia: W.B. Saunders. Slijper, E.J. 1979. Whales. Second edition, 511 pages. London: Hutchinson. [Trans­ lated from the original, Walvissen, by A.J. Pomerans] Snavely, P.D., N.S. McLeod, W.W. Rau, W.D. Addicott, and J.W. Pearl 1975. Alsea Formation: An Oligocene Marine Sedimentary Sequence in the Oregon Coast Range. Bulletin of the U.S. Geological Survey, 1395F:1-21. Snavely, P.D., N.S. MacLeod, H.C. Wagner, and W.W. Rau 1976. Geology of the Yaquina and Toledo Quadrangles, Oregon. U.S. Geological Survey, Miscellaneous Investigations Series, 1:62,500, map 1-867:1-21. Stromer, E. 1908. Die Archaeoceti des agyptischen Eocans. Beitrage zur Geologie und Paldontologie von Osterreich-Ungarn, 21:106-178. True, F.W. 1907. Remarks on the Type of the Fossil Cetacean Agorophius pygmaeus (Muller). Smithsonian Institution Publication, 1694:1-8. 222 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY Turnbull, W.D. 1970. Mammalian Masticatory Apparatus. Fieldiana, Geology 18:147- 356. Vogl, A.W., and H.D. Fisher 1981. The Internal Carotid Artery Does Not Directly Supply the Brain in the Monodontidae (Order Cetacea). Journal of Morphology, 170: 207-214. Whitmore, F.C. 1953. Cranial Morphology of Some Oligocene Artiodactyla. U.S. Geolog­ ical Survey Professional Paper, 243-H: 117-160. Whitmore, F.C, and A.E. Sanders 1977. Review of the Oligocene Cetacea. Systematic Zoology, 25:304-320. Winge, H. 1921. A Review of the Interrelationships of the Cetacea. Smithsonian Mis­ cellaneous Collections, 72(8): 1-97. Wood, F.G., and W.E. Evans 1980. Adaptiveness and Ecology of Echolocation in Toothed Whales. In R.G. Busnel and J.F. Fish, editors, Animal Sonar Systems, pages 381^425. New York: Plenum Press. Odobenocetops peruvianus, the Walrus-Convergent Delphinoid (Mammalia: Cetacea) from the Early Pliocene of Peru Christian de Muizon, Daryl P. Domning, and Darlene R. Ketten ABSTRACT Odobenocetops peruvianus Muizon, 1993 (early Pliocene, southern Peru), is a bizarre cetacean that is convergent in its skull, general aspect, and presumably feeding habits with the modern walrus Odobenus rosmarus (Linnaeus). Its cranial specializations are unique among cetaceans and include loss of the elongated ros­ trum, development of large premaxillary processes housing asym­ metrical tusks, forward migration of the bony nares, reversal of the typical cetacean telescoping of the skull, dorsal binocular vision, large vaulted palate, and an inferred upper lip. The structure of the basicranium (possession of palatine expansions of the pterygoid sinus and presence of a large cranial hiatus) and face (possession of a medial portion of the maxillae at the anterior border of the nares) indicates that it belongs to the odontocete infraorder Del­ phinida and to the superfamily Delphinoidea. Within this group Odobenocetops is related to the Monodontidae because of the lat­ eral lamina of its palatine flooring the optic groove, the anteropos­ terior elongation of the temporal fossa, and the thickness of the alisphenoid and squamosal in the region of the foramen ovale. We hypothesize that Odobenocetops, like the walrus, fed upon shal­ low-water benthic invertebrates and probably used its tongue and upper lip jointly in extracting the soft parts of bivalves or other invertebrates by suction. The highly modified morphology of the rostrum indicates that there was no melon as in all other odonto­ cetes, and therefore that Odobenocetops was probably unable to echolocate; binocular vision could have compensated for this inability. The most probable function of the tusks themselves was social, as in the living walrus, but we suggest that the historically primary function of both the premaxillary processes of Odobenoc- Christian de Muizon, UMR 8569 CNRS, Laboratoire de Paleontolo­ gie, Museum National d'Histoire Naturelle, 75005 Paris, France. Email: muizon@cimrsl .mnhn.fr. Daryl P. Domning, Research Associ­ ate, Department of Paleobiology, National Museum of Natural His­ tory, Smithsonian Institution, Washington, D.C. 20560-0121, and Laboratory of Evolutionary Biology, Department of Anatomy, College of Medicine, Howard University, Washington, D.C. 20059, United States. Email: ddomning@fac.howard.edu. Darlene R. Ketten, Massa­ chusetts Eye and Ear Infirmary, Harvard Medical School, Boston, Massachusetts 02114, United States. Email: dketten@whoi.edu. etops and the tusks of Odobenus was as orientation guides in feed­ ing. This reopens the question of whether the tusks of walruses play a role in feeding, as it seems that these also may be useful as orientation guides for the mouth and vibrissal array. Introduction Abundant remains of fossil odontocete cetaceans have been found in the rocks of the Pisco Formation near the southern coast of Peru. Although the occurrence of cetacean bones in this area has been known for more than one hundred years (Lis- son, 1890), the first odontocete described from this formation was Incacetus brogii Colbert, 1944. Subsequently, Hoffstetter (1968) was the first to show the importance of the faunal as­ semblage (fish, reptiles, birds, and mammals) of the locality of Sacaco, in the southern outcrops of the Pisco Formation, 540 km south of Lima. Further studies by Muizon (1981, 1983a, 1983b, 1983c, 1984, 1986, 1988), Pilleri (1989, 1990), and Cheneval (1992) have described part of the vertebrate fauna of the Pisco Formation, but abundant material still remains un­ studied (work in progress includes that of C. de Muizon and G. McDonald on mammals, and J. Cheneval on birds). As estab­ lished by Marocco and Muizon (1988), the Sacaco vertebrate fauna was deposited under shallow waters in a littoral and beach environment. The preservation of the fossils is excep­ tional, and associated skeletons are not rare; both characteris­ tics indicate calm waters. In 1990, the skull of an unexpected walrus-like cetacean (Muizon, 1993a) was found in the locality called "Sud-Sacaco" by Muizon (1981). In the Sacaco area, Muizon (1981, 1984, 1988a), Muizon and DeVries (1985), and Muizon and Bellon (1986) recognized five vertebrate horizons, which range from the late Miocene to the late Pliocene (approximately 9 Ma to 3 Ma). The specimen came from the Sud-Sacaco (SAS) horizon, earliest Pliocene, which has yielded an abundant vertebrate fauna (Muizon, 1981, 1984). Cetaceans are represented there 223 224 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY by balaenopterids, cetotheriids, pontoporiids (Pliopontos lit- toralis), ziphiids (Ninoziphius platirostris), and phocoenids (Piscolithax longirostris). The walrus-like cetacean, named Odobenocetops peruvianus, was briefly described by Muizon (1993a, 1993b), who concluded that its feeding adaptations were convergent with those of the walrus. Herein we describe that specimen more thoroughly and present a more detailed study of its features as a foundation for observations on the pa­ leobiology of this unique cetacean. Its phylogenetic relation­ ships with delphinoid odontocetes are confirmed. INSTITUTIONAL ABBREVIATIONS.—The following abbrevia­ tions for institutions have been used: MNHN Museum National d'Histoire Naturelle, Paris NMNH National Museum of Natural History, Smithsonian Institution, Washington, D.C. SMNK Staatliches Museum fur Naturkunde, Karlsruhe, Germany USNM Collections of the NMNH, which include those of the former United States National Museum ACKNOWLEDGMENTS.—This research was supported by funds from the Centre National de la Recherche Scientifique (CNRS), Paris, in 1991; by two short-term visitor awards from the Smithsonian Institution, Washington D.C, in 1992 and 1993; and by funds from the Institut Francais d'Etudes Andines (Lima, Peru) in 1994 (to C. de Muizon). CT scanning and peri­ otic analyses were supported by Office of Naval Research con­ tracts N00014-2-J-4000 and N00014-93-1-0940 (to D.R. Ketten) and by the Three-Dimensional Head and Neck Imaging Service, Department of Otolaryngology, Massachusetts Eye and Ear Infirmary. Beluga periotics were provided by S. Ridg­ way, Naval Research and Development (NRD, San Diego. We thank C. Martin, who found the specimen, and P.L. Larson, who collected it and donated it to the Smithsonian Institution. P. Kroehler prepared the specimen. Illustrations, including the life reconstruction (Figure 20), are by Mary Parrish (NMNH); photographs are by Vic Krantz and Chip Clark (both NMNH), and D. Serrette (MNHN). Irina Koretsky helped greatly with the illustrations. Valuable insights through discussion were pro­ vided by L.G. Barnes, T.A. Demere, R.E. Fordyce, J.G. Mead, C E . Ray, J.S. Reidenberg, and A. Werth. Ray and Werth also provided detailed formal reviews of the manuscript. We are pleased to dedicate this paper to Clayton E. Ray, who invited us to study the specimen under his care and who greatly helped and encouraged us in our work. Systematic Paleontology Superfamily DELPHINOIDEA Family ODOBENOCETOPSIDAE Muizon (1993a) TYPE GENUS.—Odobenocetops Muizon (1993a). DIAGNOSIS.—As for the single known species, O. peruvi­ anus Muizon (1993a). Genus Odobenocetops Muizon (1993a) TYPE SPECIES.—Odobenocetops peruvianus Muizon (1993). DIAGNOSIS.—As for the species. Odobenocetops peruvianus Muizon (1993a) FIGURES 1-18, TABLE 1 TYPE SPECIMEN.—USNM 488252 (originally numbered USNM 460306), an incomplete skull lacking much of the left side and all of the auditory bones. REFERRED SPECIMENS.—MNHN SAS 1613, a left periotic and partial tympanic; and MNHN SAS 1614, a left periotic. One other, recently discovered specimen (MNHN SAO 202) is relevant to this study but is not described herein in detail; it will be described in a separate report (Muizon and Domning, 2002). It is a badly weathered partial skeleton with a partial skull, including the left tympanic, periotic and asymmetrical tusk; the distal half of a humerus; a complete radius; several dorsal and caudal vertebrae; and rib fragments. It was found in the SAO horizon of the Pisco Formation (ca. 3.5 Ma), which is younger than the SAS horizon (ca. 5 Ma), and is provisionally referred herein to Odobenocetops sp. (see "Addendum," be­ low; Muizon et al., 1999). ETYMOLOGY.—From odon, tooth, and baino, walk (Greek); and from cetus (Latin, from Greek ketos), whale, and ops (Greek, masculine), like: "the cetacean that seems to walk on its teeth," to refer to its feeding position (see below) and also to its similarity to the walrus (Odobenus). Species: peruvianus, from Peru. DIAGNOSIS.—Delphinoid cetacean characterized by loss of elongate cetacean rostrum and concomitant probable absence of melon; development of large, ventrally directed premaxil­ lary alveolar processes housing straight tusks; right tusk longer than 55 cm, cylindrical, with oval section, with long, open pulp cavity 23 cm deep; left tusk probably unerupted and probably no longer than 20 cm, with short conical pulp cavity 1.5 cm deep; important muscular insertion and numerous neurovascu­ lar foramina on anterior side of premaxillae and on anterior ex­ tremity of palate, suggesting presence of strong upper lip and possible vibrissae; regression in telescoping of skull by for­ ward migration of nares and anterior withdrawal of maxillae and frontals; nasal on vertex of skull, contacting occipital and right maxilla, lying upon frontals but separated from meseth­ moid, which followed bony nares in their forward migration; large temporal fossa open dorsally; large dorsal exposure of pa­ rietals and concomitant development of temporalis muscle ori­ gin; large orbits facing dorsally and not laterally as in other od­ ontocetes; large pre- and postorbital processes expanded anterolaterally; maxillae articulating behind nares; large, deep, and vaulted palate; maxillae excluded from palate but forming part of lateral wall of skull; no maxillary teeth; no jugal; vomer very large and lanceolate; lateral lamina of palatine flooring the anteroposteriorly oriented optic gutter; pterygoid with flattened NUMBER 93 225 subhorizontal hamular process and large rounded lateral crest for origin of pterygoid muscle; ventral surface of zygomatic process of squamosal forming large and deep anteroposteriorly elongated, trough-like glenoid fossa; large cranial hiatus; al­ isphenoid and squamosal strongly thickened on anterolateral region of cranial hiatus. Periotic with relatively small pars co­ chlearis and long and thickened anterior process with small bullar facet; ventral process with relatively flattened ventral rim and large ventral tuberosity; very large dorsal aperture of aqueductus vestibuli; periotic tympanic with clearly sigmoid- shaped involucrum. GEOLOGICAL FORMATION AND AGE.—Pisco Formation, SAS (Sud-Sacaco) level as defined by Muizon (1981) and Muizon and De Vries (1985). As stated by these authors, the age of the Sud-Sacaco beds is early Pliocene, probably basal Pliocene, approximately 5 Ma. TYPE LOCALITY.—Sud-Sacaco, southern coast of Peru, on the west side of the Panamerican Highway at km 542. DESCRIPTION GENERAL FEATURES.—The general morphology of the skull of Odobenocetops peruvianus is far from that of a typical ceta­ cean. The hyperspecialization of this species (within the ceta­ ceans) has resulted in the modification of the characteristic telescoping of the cetacean skull. The skull is bilobate in dorsal view, with a large braincase posteriorly and an anterior part made up of the narial, orbital, and rostral regions. These are separated by a well-marked constriction situated behind the su­ praorbital processes of the frontals. Such a condition is not ob­ served in the other Neogene odontocetes, where the posterior expansions of the frontals and maxillae cover the braincase and the temporal fossa. In some primitive odontocetes (Xenoro­ phus, Archaeodelphis, and Agorophius) and in archaeocetes, a similar condition can be observed because of the slight tele­ scoping of the skull; in Odobenocetops, the condition described above is the result of a character reversal that considerably re­ duces the telescoping. Related to this reversal is the anterior position of the nares and of the orbits. The nares are very large compared with those in the Holocene Delphinoidea. They are anteroposteriorly elongate, and their length is more than double their width. The left naris is slightly larger than the right. The dorsal openings of the nares are horizontal (contrary to other Delphinoidea, where they are anterodorsally oriented), and they are not partially overhung in their anterior part by the maxillae and the premaxillae. In Odobenocetops the anterior walls of the narial passages face posterodorsally (posteroven­ trally in other delphinoids) and their contact with the dorsal surface of the premaxillae is a gently rounded surface, whereas it is a sharp crest in the other delphinoids. In lateral view, the dorsal profile of the skull is markedly concave; the nuchal crest and the rounded anterior part of the snout clearly overhang the much lower narial fossae and the posterior portion of the max­ illae. Another obvious characteristic of Odobenocetops is the reduction of the rostrum and the development of large premax­ illary alveolar processes housing the tusks that give the skull a strong, although superficial, resemblance to that of the modern walrus. As preserved, the skull is large, with an anteroposterior length of approximately 46 cm (the posterior part of the only condyle preserved is missing) and an estimated bizygomatic width of 18x2=36 cm (Figure 1). Premaxilla: The premaxillae are certainly the most amaz­ ing bones of the skull of O. peruvianus. Compared with other odontocetes, the elongate rostrum has disappeared, and the pre­ maxillae show long ventral processes just anterior to the preor­ bital notch that form an angle of approximately 60° with the horizontal plane of the skull (Figures 2, 3). This condition de­ fines a small dorsal horizontal portion of the premaxillae, situ­ ated between the anterior edge of the naris and the apex of the snout, and a large subvertical portion, the alveolar process, lo­ cated almost entirely on the ventral part of the skull (i.e., below the supraorbital process of the frontal). The ventral extremity of the alveolar process is posterior to its dorsal extremity and below the postorbital process of the frontal. It is hollowed by large alveoli housing the tusks (see below). The right premax­ illa is almost complete, but the left is broken and lacks most of the alveolar process. On the dorsal side of the skull (Figures 1, 2), the premaxillae form the anterolateral edges of the nares; in that region, the bones have a conspicuous posterolateral process that runs along the anterior half of the lateral border of the nares on the left side, and along nine-tenths of its length on the right side. The anterior border of each naris is formed medially by a small medial portion of the maxilla and laterally by the premaxilla. Anterior to the nares the premaxillae become narrower, being narrowest just behind the fairly large premaxillary foramina. In that region they are thick and each bears a dorsolateral rounded keel that runs anteriorly from the posterolateral process, passes lateral to the premaxillary foramen (forming its lateral border), and converges with the other keel in the anteriormost region of the snout. These keels mark the limits of an elongate triangle just ante­ rior to the bony nares; in this area, in other odontocetes, are lo­ cated the premaxillary sacs (posterior to the premaxillary fo­ ramina) and the origins of the nasal plug muscles (medial and anterior to the premaxillary foramina). The premaxillary sacs of the other Delphinoidea are generally situated upon the part of the premaxilla posterior to the premaxillary foramina and anterior and lateral to the nares; in Odobenocetops there was very little space for such air sacs and, if present, they must have been extremely reduced. The origin of the nasal plug muscle on the premaxillae of other delphinoids is generally a rugose loz­ enge-shaped or triangular area located medial and mostly ante­ rior to the premaxillary foramina; in Odobenocetops, this part of the premaxilla is small, suggesting that the nasal plug mus­ cle, if present, would have been fairly weak. Demarcating the lateral edges of the triangular surface, and anterior to the premaxillary foramina, the rounded keels men- 226 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY NA FIGURE 1.—Odobenocetops peruvianus, holotype (USNM 488252): skull in dorsal view (a), ventral view (b), lateral view (c), and anterior view (d); right premaxillary tusk (e) in lateral view (bottom) and proximal view (top); left premaxillary tusk (f) in lateral view (bottom) and proximal view (top). tioned above are very thickened and most probably represent eral to the origin of the nasal plug muscle receive fibers of the strong muscular attachments. These strong attachments are not medial portion of the rostral muscle (Lawrence and Schevill, in the position of the nasal plug muscle of other delphinoids, 1956; Mead, 1975). In Odobenocetops, they probably represent however, and consequently are not related to it. In fact, in the part of the rostral muscle modified for movements of a very other delphinoids the two rounded keels of the premaxillae lat- strong upper lip (see below and "Discussion"). NUMBER 93 227 FIGURE 2.—Odobenocetops peruvianus, holotype (USNM 488252): dorsal view of the skull. Anterior to the premaxillary foramina, the premaxillae are strongly expanded transversely; at the level of the preorbital process they are approximately twice as wide as at the level of the premaxillary foramina. On the dorsal portion, the right pre­ maxilla is slightly wider and larger than the left, resulting in asymmetry of the skull. The two premaxillae are separated by a prominent premaxillary groove, which is partially filled in its posterior part by the mesethmoid. The anterior region of the snout is formed by the premaxillae only; it is very wide, mas­ sive, and rounded. In anterior view, the apex of the snout is a high and nearly vertical wall, fissured vertically by the premax­ illary groove (Figures 4, 5). From that massive apex, two large alveolar processes are directed ventrolaterally and diverge at an angle of approximately 60°. In anterior view, the rounded apex of the snout shows an irregular and spongy surface that seems to indicate strong muscular attachments (Figure 4). The ante­ rior sides of the premaxillae show a very spongy structure and several large and small foramina (also seen on the anterior part of the palate), which indicate extensive vascularization and in­ nervation. The anterior edge of the alveolar process of the premaxilla presents a strong crest along its whole length (Figure 4). This crest is a long and irregular surface of bone, concave medially, 1-2 cm wide, which might have carried a long and narrow horny pad or, more likely, strong connective tissue for attach­ ment of the upper lip. Several large vascular foramina are ob­ servable on the dorsomedial portion of this attachment area. Its morphology is very similar to what is observed on the dorsal side of the walrus mandible, between the first mandibular tooth and the apex of the symphysis. Fay (1982:167) mentioned that, in the walrus, "the incisive area is covered by an extraordinar­ ily tough firm gingiva, unlike that on any other part of the mouth but closely resembling the skin on the palms and soles of the flippers." Furthermore, Howell (1927:21) mentioned the presence of a very hard horny tissue surrounding the lips of Neophocaena phocaenoides, and Kleinenberg et al. (1969:80) noted in the lips of Delphinapterus leucas a "many-layered keratinized epithelium" indicating that such structures are not uncommon in other delphinoids. Given the important modifica­ tion of the superior edge of its mouth, however, it is likely that the keratinization was much more pronounced in Odobenoce­ tops than in the other delphinoids. The condition of the anterior face of the premaxilla suggests the presence of a well-developed upper lip, and given the vas­ cularization of the bone, it is possible that it possessed strong vibrissae, as observed in the walrus. Hair is known to occur on the snouts of various fetal and newborn odontocetes (Yablokov and Klevezal, 1962; Tomilin, 1967; Brownell, 1989) and on the apex of the rostrum of the living Amazon dolphin Inia geoff- rensis (Best and da Silva, 1989). The rough and vascularized area mentioned above is re­ stricted to the upper one-third of the alveolar process, except for the anterior crest, and may mark the limit of muscle origin and vibrissae. In anterior view, the upper part of the mouth opening had a wide and elevated parabolic shape (Figure 5). In lateral view, the alveolar process of the right premaxilla is in sutural contact along the proximal one-half of its length with the maxilla, whereas its distal half is free. It is slightly concave posteriorly and widened at its apex; the posterior border of its free portion is rounded in cross section, differing therefore from the anterior crest, which received muscular and/or tendinous attachment of the upper lip. The lateral surface of the alveolar process is clearly convex and gently rounded, whereas its medial side is slightly convex at the apex and distinctly concave in its proxi­ mal portion, where it forms most of the anterior bony palate. The section of the alveolar process on its proximal two-thirds is teardrop-shaped, having a wide, convex, and rounded posterior edge and a thin, narrow, angular anterior edge. On the posterior side of the alveolar process, a small ridge runs from the poster- 228 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY FIGURE 3.—Odobenocetops peruvianus, holotype (USNM 488252): lateral view of the skull. omedial point of junction of the alveolar process with the pal­ ate (exactly at the anteroventral angle of the maxilla) to the posterolateral extremity of the process. On the palate, the premaxilla is in contact with the vomer medially and with the palatine posteriorly. The maxilla does not participate in formation of the bony palate. Along its me­ dial suture with the vomer, the premaxilla shows a deep, an­ teroposteriorly elongated fossa with a very large foramen (7-8 mm in diameter) posteriorly; this foramen, which opens in the palatine, is the anterior opening of a canal originating in the in­ fraorbital canal and extending into the premaxillary fossa via a well-marked gutter. This greatly enlarged foramen corresponds to the palatine foramen that is commonly observed, but very re­ duced and sometimes nearly closed, in the other Delphinoidea (Figure 6). On the lateral edge of the premaxillary fossa and lateral to it, several other smaller (but still 2-4 mm in diameter) foramina perforate the premaxilla. There is little doubt that these particularly large foramina contained blood vessels for the abundant vascularization of the upper lip. The fossa is lim­ ited medially by the vomer, which forms its medial wall. The significance of this fossa is not apparent, but in many other mammals, it is the location of a chemical sense organ (the vomeronasal organ). Chemoreception organs in living dol­ phins, however, are located on the tongue (Fried et al., 1990; Kuznetzov, 1990). Nothing definite can be said about the possi­ bility of homology of the premaxillary fossae with the incisive foramina of terrestrial mammals (these foramina are absent in cetaceans), but the various pieces of evidence given below that Odobenocetops is a cetacean would favor the hypothesis that FIGURE 4.—Odobenocetops peruvianus, holotype (USNM 488252): anterior view of the skull approximately perpendicular to a horizontal swimming posi­ tion. NUMBER 93 229 FIGURE 5.—Odobenocetops peruvianus, holotype (USNM 488252): anterior view of the skull approximately parallel to the plane made by the alveolar pro­ cesses (i.e., to the plane parallel to the bottom in a feeding position). these structures are not homologous. The premaxilla extends onto the palate in a triangular poste­ rior expansion that is in contact with the palatine; the contact between the two bones is flat and nearly horizontal, and the su­ ture on the palate forms a large V-shape, opening anteriorly. The alveolus of the right premaxilla is straight, transversely flattened (i.e., oval in section), and 30 cm deep; the left alveo­ lus is not completely preserved, but it is clear that it had a much smaller diameter than the right. Tusks: The right tusk (Figure 7) as preserved is not com­ plete; however, a 39 cm long portion of it indicates that it was a straight cylindrical tooth with a transversely flattened section (at 10 cm from the base of the tusk, the two diameters are 38 mm and 30.5 mm). Its cross section is constant along the length of the whole fragment available, although it is possible that the section of the (badly crushed) anterior extremity of the fossil is slightly smaller than the intra-alveolar portion. The tusk is made of dentine only and does not show any enamel; its sur­ face is regular and only shows fine and regular longitudinal striation. A 23 cm long pulp cavity indicates that the tooth had FIGURE 6.—Odobenocetops peruvianus. holotype (USNM 488252): ventral view of the skull. continuous growth; at its base the tooth itself is only a thin layer of dentine surrounding the pulp cavity, whose diameter decreases toward the apex. A tentative reconstruction of the tooth shows that it could not have been shorter than 45 cm and possibly was no longer than 55 cm, 15-25 cm of the tusk thus being external to the alveolus. The left tusk (Figure 8) is known from a small, 8 cm long fragment. It has a much smaller diameter than the right tusk (approximately 20% smaller); its section also is flattened trans­ versely, and at its base the section is 30 mm long and 24 mm wide. Like the right tusk, it has no enamel and its surface shows very fine longitudinal striations. Unlike the right, its sec­ tion is not constant and at its apex, as preserved, the diameters are 27 mm and 25 mm; in the middle part of the fragment the tooth is slightly inflated, and its diameters are 32 mm and 27 mm. Consequently, the tooth shows a slight but distinct taper- 230 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY 2 cm <4 FIGURE 7 (left).—Odobenocetops peruvianus, holo­ type (USNM 488252), right tusk: a, lateral view; b, proximolateral view, showing the deep pulp cavity. FIGURE 8 (right).—Odobenocetops peruvianus, holotype (USNM 488252), left rusk: a, lateral view; b, proximolateral view, showing the small pulp cav­ ity. ing toward its apex. By projecting apically the edges of the fragment, it is possible to estimate the original length of the tusk to be between 15 cm and 20 cm. A wide-open but very short pulp cavity is present at the base of the tooth; it has a small conical opening, 15 mm deep, 30 mm long, and 24 mm wide at its base. This condition indicates that the tooth was still growing, although obviously much more slowly than the right tusk. Evidence of continued growth does not prove that the tooth was erupted, as indicated by the presence of a deeper pulp cav­ ity on an unerupted incisor of a 17-year-old female Dugong dugon (Marsh, 1980). The condition of the unerupted teeth of the narwhal, however, favors the hypothesis of an erupted tooth (Hay, 1980). In the newborn narwhal, the unerupted teeth (the right in the male and both in the female) have a long pulp cav­ ity that occupies almost the entire length of the tooth; the cavity closes after deposition of approximately eight to nine dentine layers, which generally corresponds to an age younger than 2 cm sexual maturity. With the closure of the cavity, the root of the tooth develops a knot made of dentine and irregular cementum. Consequently, the narwhal's unerupted tusk stops growing be­ fore the animal has attained its full size and sexual maturity. In the skull of Odobenocetops, the pulp cavity is small but clearly NUMBER 93 231 open, the knot at the root of the tooth is lacking, and the indi­ vidual described herein is a relatively old adult, considering the state of fusion of its cranial bones. If the tusk was erupted, therefore, the premaxillae of Odobenocetops could have been strongly asymmetrical and the alveolar process of the left pre­ maxilla should not have been longer than 18 cm, considering that a small portion of the tusk had to be external to the pre­ maxilla; the alveolar process of the left premaxilla would there­ fore have been, at a minimum, almost 10 cm shorter than the right. Such asymmetry, which strongly modifies the external aspect of the head, is unknown in mammals, however; conse­ quently we conclude that the left tusk of Odobenocetops was unerupted and that, as in the female dugong, it was still grow­ ing very slowly in the closed alveolus. The alveolar process of the left premaxilla could, therefore, have had a length similar to that of the right side (or only slightly smaller) and the -20 cm long left tusk could still have grown for a long time in its 30 cm long alveolus. Furthermore, it is noteworthy that the right tusk of Odobeno­ cetops differs from all other tusks found among mammals in being straight and cylindrical. Maxilla: The maxillae of Odobenocetops peruvianus are considerably reduced in comparison with those of other delphi­ noids; they do not cover the braincase posteriorly or the fron­ tals laterally, and they do not participate in the construction of the bony palate. They form the posterolateral borders of the narial openings, and they are in contact behind the nares; be­ tween that region and the occipital crest they are reduced to a narrow strip of bone in the median area of the skull. The right maxilla is two to three times as wide as the left. They lie upon the frontals, and at their posterior ends they contact the nasals. On the specimen described herein, they are only preserved for two-thirds of the distance between the nares and the nasals, but their sutures with the frontals are clearly visible. Lateral to the nares, the maxilla forms a very narrow flange, no more than 2 cm wide, that is strongly withdrawn medially from the lateral edge of the frontal. Its lateral border is sigmoid and more or less follows the lateral profile of the frontal, being wider above the pre- and postorbital processes and narrower above the deep orbital notch. At the level of the preorbital process of the fron­ tal, the maxilla shows a small lateral spine, which corresponds to the preorbital process of the maxilla in the other delphinoids, and which overhangs the jugolacrimal and is lateral to the an­ torbital notch. In Odobenocetops the jugolacrimal has disap­ peared. The maxillary rim, which forms the posterolateral bor­ der of the nares, continues posteriorly (4 cm behind the nares on the right side and 3 on the left side) and overhangs the flat lateral portion of the maxillae that overlaps the base of the su­ praorbital process of the frontal. Three maxillary foramina are found on the right maxilla: (1) a small foramen at the level of the posterior edge of the nares, located between the maxilla and the frontal and opening poste­ riorly; (2) the anterodorsal opening of the infraorbital canal at the contact of the premaxilla and maxilla, which is preceded by a deep groove in the maxilla, communicating with the premax­ illary foramen; and (3) a third small foramen anteromedial to the antorbital notch and distinctly pinched in the "folding" of the rostrum. On the left maxilla, the dorsal opening of the in­ fraorbital canal is divided into four foramina, and through a broken portion of the lateral wall of the alveolus it is possible to see that one or two of these foramina gave passage to blood vessels for irrigation of the tusk pulp cavity. The portion of the maxilla ventral to the supraorbital process of the frontal is a small triangle of bone bordered anteriorly by the premaxilla, posteriorly by the palatine, and dorsally by the frontal. A large posterior opening of the infraorbital canal is present in this ventral portion of the maxilla, just below the su­ ture with the frontal. As mentioned above, the maxillae of Odo­ benocetops have been excluded from the palate and no maxil­ lary teeth are present. The ventral portions of the maxillae are tightly fused to the premaxillae, thus indicating that the indi­ vidual was not a young adult. Mesethmoid: The mesethmoid is a large, relatively thick blade of bone separating the nares. In dorsal view its posterior extremity presents a Y-shaped relief that overhangs a lower portion, posterior to the nares and partially overlapped by the maxillae. This condition is unique among odontocetes and is related to important modifications of the skull, such as the long contact of the maxillae behind the nares, among others. The nares are large and anteroposteriorly elongated (almost three times longer than wide); the left naris is wider than the right and is located a little more anteriorly. The right naris is ~28 mm wide and -62 mm long, and the left naris is -31 mm wide and -62 mm long. On the dorsal face of the nasal passage, at the same level as the anteriormost point of the brain but slightly more lateral, is a group of small foramina that are directed posteromedially. Those foramina are most probably related to the small olfac­ tory lobes of the brain (see below). Frontal: The frontals have the typical cetacean trait of en­ larged supraorbital processes, composed of the preorbital pro­ cess and the postorbital process separated by a deep orbital notch. The preorbital process is an elongate, flat, and horizontal wing; its anterior extremity is square, and its lateral angle is distinctly twisted dorsally. It has a marked anterior orientation and runs along the posterolateral edge of the base of the alveo­ lar process of the premaxilla. The antorbital notch is long (40 mm) and narrow (8 mm) and does not widen anteriorly as it does in most delphinoids. The postorbital process is large and much stouter than the preorbital process. It is a large triangular plate whose apex faces posterolaterally and not laterally as in other delphinoids. Like the preorbital process, it is markedly stretched anterolateral^. It is improbable that the masseter muscle, which arises partly from the postorbital process in the other odontocetes (Howell, 1927), originated on that process in Odobenocetops (see below for discussion). These two pro­ cesses are separated by a deep and wide orbital notch that faces dorsally and not laterally as in other delphinoids, therefore in- 232 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY dicating very good dorsal binocular vision. The eyes were very large, probably the size of a tennis ball, and abutted the poster­ olateral side of the base of the alveolar process of the premax­ illa. In dorsal view, the posterior edge of the supraorbital process is almost straight and oblique, and it forms the anterior border of the temporal fossa. The supraorbital process does not cover the temporal fossa as it does (at least partially) in all Neogene and Holocene odontocetes. In those forms its posterior edge is anteroposteriorly oriented and not oblique as in Odobenoce­ tops. As the supraorbital process is not (or only slightly) cov­ ered by the maxilla, it is widely exposed dorsally on the antero­ lateral region of the braincase. As a consequence of the reduction of the telescoping of the skull, the frontal of Odo­ benocetops does not cover the parietal posteriorly and does not participate in the formation of the roof of the temporal fossa as it does in all the other delphinoids. Consequently, this bone also is widely exposed on the posterolateral region of the brain­ case. The anterolateral portion of the suture with the parietal is transverse, but it shows two big indentations, which are clearly visible on the right side where it is completely preserved. Me­ dially, each frontal is exposed, like the maxilla, in a long me­ dian strip that apparently joins the occipital posteriorly (sutures in the vertex are not easy to determine as the bones are badly encrusted with iron oxide). In that region, therefore, the classi­ cal relationships of the telescoped odontocete skull are pre­ served, as the maxillae overlap the frontals, which overlap the parietals. On the right frontal, the narrow posterior portion is elevated medially and presents a strong keel that runs as far as the occip­ ital; the strip formed by the right maxilla lies on the medial side of this frontal keel and reaches the nasals posteriorly. In fact, lateral to the maxilla-frontal suture there is a deep, elongate fossa located medial to the bottom of the temporal fossa and running from the posterior edge of the supraorbital process to approximately the middle of the braincase. This fossa is absent on the left side of the skull and could be interpreted as a patho­ logical deformation. The keel mentioned above also could be a pathological structure related to the first one; in the middle re­ gion of the braincase, the right frontal appears to have been pinched transversely and thus elevated. It is possible that the animal suffered a minor skull trauma when young, the traces of which are still observable on the skull. Because of that defor­ mation, the posterior median strip of the frontal clearly appears to be, in dorsal view, narrower than the left frontal at that level. In fact, however, if one "unfolds" the right frontal, both fron­ tals have approximately the same width. Another interpretation would be that the asymmetry of these structures is related to the asymmetry that characterizes the del­ phinoids. If this were the case, however, the asymmetry pattern observed in Odobenocetops would be unique among cetaceans. In delphinoids, the asymmetry is always manifested by larger size of the right maxilla and premaxilla and a displacement of the vertex to the left side of the skull. A possible interpretation of that morphology would therefore be that the right elongated fossa and the crest serve as muscular attachments for part of the maxillonasolabialis. In odontocetes the pars nasalis of this muscle is the blowhole muscle, which arises from the maxilla and part of the frontal; the withdrawal of these bones from the roof of the braincase would have forced the fibers to concen­ trate in that fossa and on that crest. It would, however, be sur­ prising if this phenomenon occurred only on the right side. Un­ til new material is discovered, the first interpretation seems more reasonable. Consequently, it is likely that the maxillona­ solabialis was relatively reduced in Odobenocetops. This mus­ cle is related to the movements of the blowhole and of the nasal air sac complex, two structures related to ultrasonic sound pro­ duction, so it is probable that Odobenocetops had a relatively reduced air sac system (already observed in the reduction of the premaxillary sacs) and consequently reduced ultrasonic sound production consistent with loss of echolocation. The posterior extremities of the frontals articulate with the nasals. Ventrolateral^, the frontals contact the alisphenoid and the palatine posteriorly. In ventral view, the supraorbital process of the frontal is in a subhorizontal plane that meets the lateral wall of the palate at a right angle; along this angle, the frontals artic­ ulate successively with the orbitosphenoid (optic gutter), with the palatine, and anteriorly with the maxilla. Nasal: The nasals are small plates of bone that have been pushed posteriorly; they articulate with the frontals anteriorly and with the occipital posteriorly. The nasals apparently lie upon the parietals, whereas in other odontocetes they always lie upon the frontals. The lateral half of the right nasal is preserved in situ, but its medial half, as well as the left nasal, have been lost during fossilization. The sutures, however, are clearly visi­ ble in numerous small anteroposteriorly directed grooves that mark the skull in that region. The nasals were more wide than long, and apparently the left nasal was larger than the right. Parietal: The parietal forms most of the dorsal face of the braincase as a smooth bony plate that is covered only by the frontal medially. It contacts the squamosal by a very strong in­ terdigitated suture, which differs from the squamous suture generally found in mammals. As seen below, this condition is probably related to the feeding adaptations of Odobenocetops. Ventrally, the parietal shows a small expansion in the anterior part of the glenoid cavity. Palatine: The palatine forms an important part of the pal­ ate; it is located posterior to the premaxilla, medial to the ptery­ goid, and lateral to the vomer. As mentioned above, a very large palatine foramen opens anteriorly in a deep elongate fossa on the anteromedial border of the premaxilla. The palate is very large and deeply arched transversely, as well as longitu­ dinally. Very wide anteriorly, at the level of the premaxillae, it narrows posteriorly at the level of the palatines and widens again at its posterior extremity at the level of the pterygoids. The anterior portion of the palatines is located in the narrowest part of the palate, and consequently the anterior part of the pa­ latine (on the palate) is much narrower than the posterior. On NUMBER 93 233 the lateral side of the skull, below the supraorbital process, the palatine forms a subvertical wall with a deeply concave ventral border. The suture with the maxilla is markedly concave pos­ teroventrally. Above the lateral edge of the palate and just ante­ rior to the suture with the pterygoid, the palatine presents an elongate fossa that also extends onto the pterygoid. That fossa is deeper on the palatine and gradually shallows and disappears on the pterygoid posteriorly. It follows the lateral edge of the palate and consequently is oriented anterodorsally-posteroven- trally. Boenninghaus (1904), referring to Phocoena, and Fraser and Purves (1960), referring to Delphinus, mentioned that the origin of the internal pterygoid muscle was on the lateral edge of the maxilla, palatine, pterygoid, and (in the case of Phoc­ oena) basioccipital. The elongated fossa observed on the lateral edge of the palatine, therefore, probably corresponds to a strong attachment of the internal pterygoid muscle, which also attaches on the pterygoid hamulus and on the medial lamina of the pterygoid (see below). Although the palatine is partially broken in its dorsal portion, it clearly had an important contact with the frontal. Ventral to the optic gutter of the orbitosphenoid, the dorsal edge of the pa­ latine shows a small curvature (concave dorsally), which we re­ gard herein as the medial part of the lateral lamina that was covering the optic gutter and contacting the frontal dorsally. Posterodorsally, the palatine also contacts the anterior edge of the alisphenoid. Pterygoid: The pterygoid is excavated by a relatively small pterygoid sinus and, therefore, is divided into lateral and me­ dial laminae. On the palate, its suture with the palatine is an an­ teroposterior line observable at the posterolateral corner of the palate. The suture between the two bones on the palate is actu­ ally a horizontal plane, as the large backward expansion of the palatine covers the pterygoid posteriorly. Because of this mor­ phology of the palatine, the pterygoids are laterally displaced and widely separated, and the palatine must have formed the major part of the posterior edge of the bony palate. Widely sep­ arated pterygoids are known in the Holocene Monodontidae and Phocoenidae and in some living Delphinidae. The pterygoid hamulus is only partially excavated by the pterygoid sinus and possesses a large ventrolateral crest, proba­ bly for the insertion of the pars lateralis of the pterygoid mus­ cle. The outline of the crest is strongly convex in ventral view, in contrast to the markedly concave lateral border of the pa­ latine anteriorly. The anteriormost extremity of the palatine- pterygoid suture is on the lateral edge of the palate at the in­ flexion point of the curve. Although the apex of the pterygoid hamulus is broken on the specimen, it is probable that only a small part is missing. The medial lamina of the pterygoid, posterior to the hamulus, is strongly thickened relative to other odontocetes. Its lateral edge shows a sort of semicylindrical crest, which delimits a semicir­ cular cupule. Probably an extremely strong pterygoid muscle originated partly on this structure. On its medial side and ven­ tral to this muscle attachment is an anteroposteriorly directed notch excavated in a very thickened and dense bony wall. This structure is in the location of the passage of the eustachian tube observed in the other odontocetes, although it is never as con­ spicuous in the latter. The pterygoid forms the dorsal wall of the choanae and partially overlaps the vomer medially, contrary to the condition in other odontocetes, where the vomer gener­ ally overlaps the medial border of the medial lamina of the pterygoid. The lateral lamina of the pterygoid contacts the pa­ latine anteriorly and the alisphenoid posteriorly. It is large and smooth and participates in the formation of a continuous bony bridge between the palate and the frontal, a condition that strongly recalls that observed in the Monodontidae. Vomer: The vomer is very large, and its participation in the palate is much more extensive than in any other odontocete. On the palate the vomer has the characteristic lanceolate shape ob­ served in those odontocetes in which the vomer participates in the formation of the palate. It is long and occupies approxi­ mately two-thirds of the midline of the palate. Its maximum width is at the level of the anteriormost point of the palatine- premaxillary suture; its anterior half contacts the premaxilla, and its posterior half contacts the palatine. The posterior part of the bone, in the basicranial basin, is relatively narrow but much thicker than in the other delphinoids. Orbitosphenoid: The orbitosphenoid has a long and wide optic canal, whose anterior opening is situated below the poste­ rior edge of the supraorbital process and above the inflexion point of the sigmoid lateral border of the palate. It has an al­ most anteroposterior orientation and faces laterally, whereas in the other odontocetes it faces ventrally and forms an angle of approximately 45° with the anteroposterior axis of the skull. A large optic foramen opens in the optic canal just below the nar­ rowest part of the temporal fossa; lateral to it, the large sphe- norbital fissure (anterior lacerate foramen) is walled laterally by the lateral laminae of the palatine and pterygoid. The mor­ phology of this region of the skull is related to the strong ante­ rior displacement of the nares and the orbits, which have dragged the optic canal anteriorly (above the palatine-ptery- goid suture); in most delphinids and phocoenids, the optic ca­ nal lies posterior to the lateral lamina of the pterygoid. It is noteworthy, however, that a condition intermediate between that of Odobenocetops and that of most of the other Delphi­ noidea is observed in Delphinapterus, in which the orbits are located fairly far anteriorly. Alisphenoid: The alisphenoid is a very thick bone located just anterior to the large squamosal gutter. It contacts the fron­ tal dorsally, the squamosal and parietal posteriorly, and the pterygoid ventrally. On its ventral edge can be observed the dorsal border of an enormous foramen ovale. This foramen was not closed and was confluent with the cranial hiatus of the au­ ditory region, contrary to what is observed in the other delphi­ noids. Apart from this feature, the major characteristic of the alisphenoid is its thickness, which, among the delphinoids, also is found in the Monodontidae (Muizon, 1988b). 234 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY Basioccipital: The basioccipital is almost totally broken away, and only the anterior portion of the right side is pre­ served. On its ventrolateral side, a small and shallow gutter very probably represents the passage for the internal carotid. The carotid foramen was, therefore, confluent with the foramen ovale and with the cranial hiatus. On the dorsal side of the ba­ sioccipital, the carotid gutter reaches the lateral edge of the sella turcica, as in the other delphinoids. The carotid gutter has a strong anteroposterior orientation, however, and the internal carotid entered the skull at a point well posterior to the sella turcica. In the other delphinoids the carotid foramen is located approximately at the same level as the sella turcica. Squamosal: The squamosal is a relatively small bone whose entire morphology is conditioned by a wide, deep, and anteroposteriorly elongate gutter, which is formed mainly by the highly modified zygomatic process. The gutter is open an­ teriorly and posteriorly and is approximately cylindrical. It cor­ responds in other delphinoids to the depression observed be­ tween the lateral wall of the zygomatic process and the crest that joins the falcate process to the posteromedial origin of the fossa for the middle sinus. The lateral edge of the squamosal gutter of Odobenocetops is very thick, and its anterior extrem­ ity represents the greatly shortened zygomatic process. It is likely that the glenoid cavity did not occupy the whole volume of the gutter. A comparison with the position of the glenoid fossa in a beluga or narwhal shows that in Odobenocetops it probably occupied the posterolateral quarter of the squamosal gutter. The articulation of the condyle was therefore on the pos­ terior half of the medial side of the lateral wall of the gutter. The posterior region of the gutter bears a posterodorsally extro­ verted saddle-like crest, which strongly suggests the presence, anterior to it, of a very mobile articulation. The prominence of this saddle-shaped crest also indicates the mature development of the articulation, in agreement with the extensive fusion of the bones of the skull, and indicates a relatively old animal. It is therefore likely that the condyle articulated only with the poste­ rolateral extremity of the squamosal gutter. The squamosal gutter (and consequently the glenoid fossa) is located in a much higher position than in the other delphinoids. It is well above the lateral wall of the cranial hiatus, and its bot­ tom is approximately at the same level as the posterior edge of the postorbital process of the frontal. In the other delphinoids, the glenoid cavity is at the same level as the lateral wall of the cranial hiatus and well below the posterior edge of the postor­ bital process of the frontal. The ventral edge of the medial wall of the gutter is rounded, very thick, and dense, and it forms the lateral edge of the enor­ mous cranial hiatus; it is related anteriorly to the extremely thickened medial edge of the alisphenoid. Immediately behind the suture with the alisphenoid, on the thickened medial wall of the squamosal gutter, a small crest probably represents the very reduced falciform process of the squamosal. At the posterior extremity of the medial wall of the squamosal is the medial ex­ tremity of the middle sinus fossa; the sinus was probably fairly well developed and may have extended onto most of the medial wall of the squamosal gutter. At the posteromedial corner of the bone, on the suture with the occipital, there is no articula­ tion area for the periotic but rather some squamous and spongy bone for attachment of the ligament that, as in the other Del­ phinida, held the periotic in place. Behind the saddle-shaped posterior extremity of the squamo­ sal gutter, and squeezed between it and the exoccipital, the pas­ sage for the external auditory meatus is long but very reduced; it opens dorsolaterally in the ventral part of the very small ster- nomastoid fossa. The latter is reduced and occupies the poster­ odorsal corner of the zygomatic process (or of the lateral wall of the squamosal gutter). It is limited dorsally by a small crest, which is much more reduced than its homologue in other del­ phinoids. The lateral side of the zygomatic process is wide and elongated anteroposteriorly. In other delphinoids this surface gives rise to the masseter muscle; therefore, the condition ob­ served in Odobenocetops suggests a strong masseter (for a ce­ tacean). On the lateral wall of the braincase, the suture of the squa­ mosal with the parietal is very unusual. In most other mam­ mals, and certainly in all the other odontocetes, the squamosal- parietal suture is mainly a subvertical plane facing laterally, and the part of the squamosal contacting the parietal is a squa­ mous lamina, which gives its name to the bone. In Odobenoce­ tops, the articulation is a rough zigzag line, and the joint sur­ face faces dorsally. This articulation, together with the general stoutness of the squamosal, suggests that large muscular ten­ sions were applied to the bone. In the posterolateral region of the squamosal, just anterior to the paroccipital process, the en­ tire region of the external auditory meatus is tremendously thickened, further exemplifying the exceptional stoutness of this bone in Odobenocetops peruvianus. Periotic: One isolated left periotic (MNHN SAS 1614) and one partial tympanic associated with another left periotic (MNHN SAS 1613) are referred to Odobenocetops peruvianus (Figures 9, 10). The taxonomic assignment of these isolated el­ ements is clearly confirmed by the recent discovery of a partial skeleton of Odobenocetops sp. (MNHN SAO 202) that in­ cludes, associated in situ with the skull, a tympanic and periotic like those described herein. In the following description the tympanic side of the periotic will be called ventral; the cerebral side, dorsal; the cochlear promontory side, medial; and the side opposite to the cochlear side, lateral. The nomenclature of the external anatomy of odontocete periotics follows Fordyce (1994). Both periotics are morphologically as unusual as the skull and do not closely resemble those of any other cetacean. The periotic of Odobenocetops is a large and stout bone whose proportions are in agreement with the very large cranial hiatus of the skull. It has a large, long, and robust anterior process with a blunt rounded apex. On the ventral side of the anterior process, a large lateral tuberosity is located lateral to the mal­ lear fossa. From the tuberosity, a wide and flat ventral rim ex­ tends anteriorly as far as the apex of the anterior process. It is NUMBER 93 235 FIGURE 9.—Odobenocetops peruvianus, left periotic (referred specimen, MNHN 1613): a, dorsal view; b, medial view; c, ventral view; d, lateral view. wider and flatter than in other delphinoids, especially in its an­ terior portion. This rim bears numerous fine parallel wrinkles, which are slightly concave anteriorly and oblique to the axis of the bone. This structure, although flattened, corresponds to the "bourrelet ventral" defined by Muizon (1988b), which is a syn­ apomorphy of the Delphinida. Three small fossae are observed on the ventral face of the an­ terior process, medial to the ventral rim. They are, from rear to front, the mallear fossa (or fossa capitis mallei), the epitubarian fossa, and the bullar facet. The mallear fossa receives the head of the malleus; it is oriented more ventrally than in the other Delphinoidea, especially the Monodontidae where the orienta- «ft 1 cm FIGURE 10.—Odobenocetops peruvianus, left periotic (referred specimen, MNHN 1614): a, dorsal view; b, medial view; c, ventral view; d, lateral view. tion is almost medial. The epitubarian fossa receives the pro­ cessus tubarius of the tympanic, also called the accessory ossi­ cle, unciform process, or uncinate process. When compared with that in other delphinoids, the epitubarian fossa is small relative to the size of the bone, but it is more concave. The third and most anterior fossa is the bullar facet, as defined by Fordyce (1994), and erroneously termed epitubarian fossa by Muizon (1987, 1988a, 1988b, 1988c, 1991). The bullar facet is a shallow fossa, slightly longer than it is wide, which occupies approximately the anterior one-third of the ventral face of the process. The bullar facet is surprisingly well developed for a delphi­ noid; however, it is noteworthy that a small bullar facet also is 236 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY observed in some Kentriodontidae (Muizon, 1988b: 172), such as Liolithax. The bullar facet is a common structure in several groups of odontocetes (Squalodontidae, Squalodelphidae, Pla- tanistidae, Waipatiidae, Ziphiidae, Physeteridae, and Li- potidae). The loss of the bullar facet (=epitubarian fossa sensu Muizon, 1987, 1988a, 1988b, 1988c, 1991) has been consid­ ered a synapomorphy of the monophyletic clade Delphinoidea + Inioidea (Muizon, 1988b). This clade is defined by three other synapomorphies (Muizon, 1988b: 165), so it is possible that loss of the bullar facet occurred several times in the evolu­ tion of the Delphinida. It also is possible that the presence of a bullar facet in Odobenocetops is related to the increase in length and thickness of the anterior process of the periotic, which is as highly specialized as the rest of the skull. In dorsal view, the anterior process shows several slits, fo­ ramina, and fissures, as is often the case in the Monodontidae. The fossa incudis (or fossa cms breve incudis) is located just posterior to the mallear fossa and anterior to the anterior ex­ tremity of the tympanic articular facet of the posterior process. It is small and shallow, and it faces more anteroventrally than in other delphinoids. Anterolateral to the fossa incudis is the ventral (or tympanic) opening of the facial canal, which transmits the facial nerve (VII). This foramen is small, approximating the size of that ob­ served in Delphinapterus, although the periotic of Odobenoce­ tops is almost twice the size of that of Delphinapterus. Be­ tween the mallear fossa, the fossa incudis, and the ventral opening of the facial canal is a small elongate fossa that faces posteriorly. Such a fossa is not present in all Delphinoidea; among the Holocene species, we have observed it in Pseud- orca. In some periotics of an undescribed fossil delphinapterid and of a fossil kogiine (Luo and Marsh, 1996) from Lee Creek Mine (North Carolina), a very small pit is observed ventrolat­ eral to the ventral opening of the facial canal. The fossa for the stapedial muscle is large but shallow and faces almost ventrally. It differs in this respect from the condi­ tion in Delphinapterus and most other odontocetes, where the fossa faces laterally or ventrolaterally. The groove for the facial nerve is not well separated from the stapedial muscle fossa as is generally observed in other Delphinoidea. Posteromedial to the ventral opening of the facial canal is the fenestra ovalis. Con­ trary to what is observed in Delphinapterus, Monodon, and generally in the Delphinoidea, where the fenestra ovalis is oval-shaped, in Odobenocetops it is almost circular, and the stapes (preserved in MNHN SAS 1613) is a small conical bone, inflated at its ventral extremity for articulation with the incus (Figure 12). It is not flattened as in most other odonto­ cetes. Posteromedial to the fenestra ovalis is the fenestra rotunda, which is clearly reniform in MNHN SAS 1613. The posterolat­ eral wall of the fenestra rotunda is thickened and shows a con­ spicuous elevation responsible for its oval shape. That mor­ phology is not observed in the other Delphinoidea, but it is common in Squalodon and in the Eurhinodelphidae (Muizon, 1988c, 1991). In MNHN SAS 1614, the fenestra rotunda is wider than in MNHN SAS 1613, a condition that could be pathologic or related to erosion. The pars cochlearis is extremely small considering the large size of the periotic. It is very low and somewhat recalls that of Squalodon, although it is not as compressed dorsoventrally in medial view, is not ventrally shifted, and has a rounded overall shape in ventral view. It differs strongly from the much higher profile (in dorsal or ventral view) pars cochlearis of other Del­ phinoidea and especially that of the Monodontidae. The poste­ rior process is relatively slender and bears a narrow, anteropos­ teriorly elongate articular facet for the tympanic. Lateral to the tympanic facet is a conspicuous anteroposteriorly oriented crest that bordered the external auditory meatus dorsally. On the dorsal face of the bone is the internal auditory win­ dow, composed of the tractus spiralis foraminosus (or fundus of the internal auditory meatus) (posteromedially), the foramen singulare, and the dorsal aperture of the facial canal for the fa­ cial nerve (anterolaterally). The window is large but relatively smaller than in the other Delphinoidea. It is approximately the size of that of Delphinapterus, although the periotic of Odo­ benocetops is twice as large. In particular, the tractus spiralis foraminosus is smaller than that of most large delphinoids. The window in Odobenocetops, however, differs from that of the other delphinoids in being much deeper, probably because of the thicker (in medial view) pars cochlearis. The same can be said concerning the aperture for the endolymphatic duct (dorsal opening of the vestibular aqueduct), which is a large conical pit posterolateral to the window. It is larger than that of Delphi­ napterus but resembles it in being widely open. The aperture for the cochlear aqueduct is a small pit facing dorsally, antero- medial to the aperture for the endolymphatic duct. The dorsal face of the periotic has an irregular aspect and presents numer­ ous exostoses made of dense bone; it does not show the smooth and regular morphology generally observed in the other Del­ phinoidea. In this respect, it resembles the periotics of Physe­ teridae. In lateral view, the periotic of Odobenocetops shows a very large and relatively flat surface. CT SCANS OF THE PERIOTICS METHODS.—The two periotics of Odobenocetops peruvi­ anus were scanned with an ultra-high resolution protocol de­ veloped for analysis of fossil material. The technique provides fine-level differentiation of mineralization characteristics, which in turn provides good delineation of inner-ear structures in CT images. Simple surface reconstruction algorithms were used to obtain three-dimensional reconstructions of the periotic shell. Segmentation algorithms were used to visualize particu­ lar features, e.g., the cochlear canal, vestibular aqueduct, facial nerve, and auditory nerve. In live ears, these algorithms utilize the attenuation coefficients for cochlear fluids and neural tis- NUMBER 93 237 FIGURE 11.—Odobenocetops peruvianus, left periotic (referred specimen, MNHN SAS 1613): reconstruction of inner ear spaces from CT data. (Scale bar= 1 cm; a=apex of cochlea; an=auditory nerve canal; b=basal turn; ca= cochlear aqueduct; fn=facial nerve canal; oosl=outer osseous spiral laminar ridge; va=vestibular aqueduct.) sues to produce three-dimensional representations of inner-ear anatomy; in fossils, the differences in X-ray attenuations of sediments or mineralization of bone versus neural and inner ear areas are used for the reconstructions. RESULTS.—Inner Ear: Both specimens are left periotics with well-preserved inner ears. Small differences in cochlear dimensions and turn number (Table 2) between these speci­ mens are within the range of normal interindividual variability for modern odontocete inner ears (Ketten and Wartzok, 1990). The cochleae have approximately 2.5 turns with the three clas­ sic hallmarks of odontocete cochleae: a ventrolateral apical ori­ entation, a substantial outer osseous lamina in the basal turn, and an exceptionally large cochlear aqueduct (Figure 11). The cochlear canal follows a conventional odontocete type II for­ mat (Ketten, 1984; Ketten and Wartzok, 1990), with closest morphometric affinities to Delphinapterus leucas and Mon- odon monoceros. The three-dimensional cochlear canal length for each specimen is approximately 44.4 mm, implying an ani­ mal length of 360-400 cm, assuming that the ratio of canal length to body mass follows the same allometry as extant odon­ tocetes (Table 1). A substantial outer osseous lamina (as indi­ cated by the laminar indentation in Figure 11) is present throughout the first 16 mm of the lower basal turn in both spec­ imens and is a clear indication of some ultrasonic hearing in Odobenocetops. The outer lamina covers approximately 35% of the cochlear length, again consistent with a type II odonto­ cete cochlear format. Cochleae of this type have maximum peak ultrasonic sensitivities below 80 kHz. In fact, the cochlear profile of these Odobenocetops cochleae is best approximated by species at the lower end of the type II group of odontocete ears, which have a peak frequency of 35-50 kHz. Although Odobenocetops, judging from its cochlear configu­ ration, could (like most mammals) perceive frequencies above 20 kHz, it is important to note that having ultrasonic hearing abilities is not synonymous with echolocation. Echolocation per se requires not only perception but also synchronized pro­ duction of ultrasonic signals coupled with the ability to analyze target features from the returning echoes. A notable deviation from typical extant odontocete inner ear configuration is the presence of well-defined semicircular ca­ nals (Figure 12), a large, reniform vestibular aqueduct, and, judging from the diameter of the residual VIIIth nerve canal (Table 3), relatively large vestibular and facial nerve fiber counts (Figure 13). Extant Odontoceti have vestibular volumes that are substantially less than their cochlear canal volumes. In TABLE 1.—Measurements (in cm) of holotype of Odobenocetops peruvianus (USNM 488252). Total length of skull from anteriormos extremity of pre- maxillae to posterior border of occipital condyle Width of postorbital constriction Width of orbit from apex of antorbital postorbital process Drocess to apex of Length of right premaxillary sheath from its apex to an- tero-dorsal hump of premaxilla Depth of right alveolus Bizygomatic width (estimated) Anteroposterior length of braincase Width of braincase (estimated) Length of mandibular gutter 45.7 11.6 10.5 31.7 30.7 15.4x2=30.8 18.3 10.2x2=20.4 10.2 238 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY FIGURE 12.—Odobenocetops peruvianus, left periotic (referred specimen, MNHN SAS 1613): reconstruction of inner ear spaces from CT data. (Scale bar=l cm; a=apex of cochlea; an=auditory nerve canal; b=basal turn; s= head of stapes; ssc=semicircular canal; va=vestibular aqueduct; vn=vestibular nerve bundle.) many species, the semicircular canals are sufficiently small and thread-like that the complete canal system cannot be traced in periodic histologic sections, much less detected by high-resolu­ tion CT (Gray, 1951; Ketten, 1992). This contrasts sharply with the condition in land mammals, in which vestibular canal vol­ umes rival cochlear volumes and are readily imaged by even conventional CT (Gray, 1951; Spoor, 1993). Although size is not a criterion for vestibular function, the majority of cetaceans appear to have uniquely small semicircular canals that are sig­ nificantly shorter and narrower than their cochlear canals (Gray 1951; Jansen and Jansen, 1969). This anomalous vestibular de- TABLE 2.—Cochlear morphometry of cetaceans. Measurement are in millime­ ters. (e=estimated.) Species Delphinapterus leucas Grampus griseus Lagenorhynchus albirostris Monodon monoceros Phocoena phocoena Physeter catodon Stenella attenuata Tursiops truncatus Odobenocetops peruvianus (MNHN SAS 1613) Odobenocetops peruvianus (MNHN SAS 1614) Turns 1.75 2.5 2.5 2.5 1.5 1.75 2.5 2.25 2.5 2.25 Scalae length 41.99 40.5 34.9 45.12 25.93 72.21 36.9 40.65 44.37 44.41 Basal diameter 10 8.73 8.74 11.6 5.25 14.3 8.61 9.45 10.1 10 Axial height 4.21 5.35 5.28 5.3 1.47 3.12 4.36 5.03 4.31 4.5 Body length 325 228 207 425 133 1361 185 259 380e 380e velopment is underscored by the fiber distributions of cetacean acoustovestibular nerves. Recent data show that the majority of odontocetes have 3%-5% of the total of VIIIth nerve fibers de­ voted to vestibular components, as compared with an average of 45% in land mammals—percentage ranges from about 20% in bats to about 60% in brachiating primates (Gao and Zhou, 1995). The three readily imaged semicircular canals and the well- defined vestibule of Odobenocetops are similar in dimensions to those of the beluga (Delphinapterus leucas) (Figure 12). Delphinapterus leucas has the highest semicircular canal ratio of any extant odontocete studied. In all mammals, cochlear size is best correlated with body mass, but the exact correlates of the semicircular-canal size in most mammals remain unclear. For primates, a positive relationship to locomotory behavior appears relatively robust (Spoor, 1993). It has been suggested that fusion of the cervical vertebrae in Cetacea resulted in lim­ ited head movements and substantially fewer inputs to the ves­ tibular system, leading to a loss of related receptors (Ketten, 1992). Better-developed semicircular canals in D. leucas are consistent with this species having a well-defined neck and greater rotational and lateral head movements (a full right an­ gle) than other living odontocetes. The diameters of the audi­ tory and vestibular canals in Delphinapterus, Monodon, and Odobenocetops are very close but not substantially different from those of other odontocetes. Cross-sectional areas of the canals for the vestibular component versus the auditory compo- NUMBER 93 239 FIGURE 13.—Odobenocetops peruvianus, left periotic (referred specimen, MNHN SAS 1613): reconstruction of inner ear spaces from CT data. (Scale bar= 1 cm; abbreviations as in Figure 11.) nent of the VIIIth nerve are somewhat higher for both Delphi­ napterus (7%) and Odobenocetops (6%-9%) than for most del­ phinids. Assuming that the diameters of neural fibers occupying those cross sections were equivalent in extinct and extant species, we speculate that Odobenocetops had a vestibu- lar-ganglion cell population between 4500 and 7000 neurons, with an average auditory ganglion population of approximately 75,000 to 80,000. There are, however, a great many caveats to this speculation, chief among them being that the available da­ tabase is too small for explicit conclusions about exact neu­ ronal levels. At this point, the most appropriate interpretation is that the exceptionally close morphometric resemblance be­ tween the Delphinapterus and Odobenocetops vestibular sys­ tems is striking and consistent with the idea that Odobenoce­ tops also may have had substantially greater head motions than most modern cetaceans. This motility could be related to the benthic feeding adaptations proposed herein. The significance of an enlarged vestibular aqueduct and pre­ sumably equally large endolymphatic duct is unclear. Large reniform aqueducts are found in some baleen whales, e.g., Bal- aena mysticetus. No detailed comparative studies of the aque­ duct across a variety of species are currently available for aquatic mammals, however. Equally striking is the relative size of the facial nerve canal in Odobenocetops compared with that of most odontocetes. With the exception of Monodon monoceros, the facial nerve is typically less than 1.5 mm in diameter. The facial nerve of the specimen of M. monoceros examined for this study, an adult TABLE 3.—Neural canal morphometry of cetaceans. Measurements are in millimeters. Species Delphinapterus leucas Grampus griseus Lagenorhynchus albirostris Monodon monoceros Phocoena phocoena Physeter catodon Stenella attenuata Tursiops truncatus Odobenocetops peruvianus (MNHN SAS 1613) Odobenocetops peruvianus (MNHN SAS 1614) Vestibular nerve diameter 1.20 - - 1.50 0.52 - - 1.00 1.15 1.00 Facial nerve diameter 1.47 - - 3.04 1.09 - - 1.41 2.7 2.5 Internal auditory canal diameter 5.6 4.7 4.6 6.22 3.7 11.3 4.9 5.6 5 5 Auditory nerve diameter 4.4 3.7 3.6 4.92 2.78 - - 4.5 3.85 4.0 240 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY FIGURE 14.—Odobenocetops peruvianus, left tympanic (referred specimen, MNHN 1613): a, medial view; b, ventral view. male, measured 3.1 mm, or approximately double the odonto­ cete average. The Odobenocetops facial nerve canals averaged 2.6 mm at mid-periotic level. Once again, it is not possible to draw definitive neural conclusions from the anatomy of one or two animals; however, it is noteworthy that in each of these species the tusks coincide with an unusually large volume of facial nerves. Middle Ear: Little can be said about the middle ear be­ cause of the lack of complete tympanies. One stapes remains in situ. It is monocrural with a large head and footplate, a configu­ ration common among both Odontoceti and Sirenia. The distal facet has a surface area that is large in comparison with the stapes footplate and is more ventrally directed in this specimen than in other odontocetes. Tympanic: The tympanic is incomplete, as only the involu- crum and a small portion of the ventral side are preserved (Fig­ ure 14). It shows, however, a typical feature of the Delphinida (Muizon, 1988b), namely the sigmoid morphology of the in- volucrum in medial view. This feature is very clear in the Li- potidae, Inioidea, and Delphinoidea. It is absent in the Platanis­ toidea (sensu Muizon 1987, 1991) and in the Physeterida. The involucrum is much more robust and stout in Odobenocetops than in any other Delphinoidea. It is not dorsoventrally flat­ tened as in the Holocene Delphinoidea, even in the most robust forms such as Orcinus. In some Kentriodontidae, such as Ato- cetus, the involucrum is relatively robust but much smaller. Occipital: The occipital is relatively lower than in other Delphinoidea but is very wide and convex (Figure 15). It has undergone some slight postmortem deformation, but this ap­ parently had little effect on the overall shape of the bone. The portion of the bone lateral to the occipital condyle is much wider and lower than in the Delphinidae and Phocoenidae and is strongly expanded laterally. It somewhat resembles the con­ dition in the Monodontidae, where the posterior crest of the temporal fossa (the lambdoid crest) is oblique and not vertical. In Odobenocetops the lambdoid crest is assimilated into the nuchal crest as a consequence of the strong modification of the temporal fossa. The paroccipital process, although partially broken, seems to have been very well developed, robust, and FIGURE 15.—Odobenocetops peruvianus. holotype (USNM 488252): occipital view of the skull. NUMBER 93 241 markedly expanded laterally. Its anterior face very probably re­ ceived a posterior sinus. It is thickened and fused to the poste­ rior thickened region of the squamosal. The very stout mor­ phology of this squamoso-paroccipital angle of the skull, as well as the robustness of the zygomatic process of the squamo­ sal, seems to indicate very strong musculature inserted on this region of the skull: the scalenus ventralis and scalenus dorsalis, which insert on the anteroventral and posteroventral regions of the paroccipital process respectively, and possibly the costo- humeralis and sternomastoideus, which insert in part on the posterolateral border of the process. In its dorsal region, the oc­ cipital crest does not seem to have been more prominent than in other delphinoids, and the muscle insertions are not as marked as in some Monodontidae (in some males of Monodon). The occipital shield is strongly convex, much more so than in other Delphinoidea. It differs from the somewhat flattened condition observed in the Monodontidae and is closer to the more convex condition of the Delphinidae and Phocoenidae. The right condyle only is preserved, and that not totally. It is much more convex and salient than in any Holocene delphi­ noid and adjoins a very deep supracondylar fossa. These struc­ tures are much more developed than in Delphinapterus, which is well known for having a very well-defined and flexible neck for a cetacean. Consequently, the morphology of the occipital condyle of Odobenocetops obviously indicates that it was ca­ pable of very ample movements of its head, which was cer­ tainly much more mobile and flexible than in Delphinapterus. Endocranial Cavity: The endocranial cavity is widely opened, as major parts of the posterolateral and ventrolateral regions of the left side of the skull are missing, allowing obser­ vation of the internal morphology of the braincase (Figure 16). Its major characteristic is the presence of ethmoidal fossae for a pair of small olfactory bulbs, separated by a small bony wall. The fossae are horizontal, slightly higher than they are wide, and approximately 4 mm wide, 6 mm high, and 8 mm deep. Such structures are very uncommon in the living odontocetes but have been mentioned in the Delphinidae (Sinclair, 1966) and Eoplatanistidae (sensu Pilleri and Ghir, 1981, 1982; Muizon, 1988c), and Muizon observed them in the Eurhinodel- phidae, in the Platanistidae (cf. Pomatodelphis sp., USNM 214759), and in the Squalodontidae (Squalodon tiedemani). Kellogg (1926) observed, in the posterior wall of the bony nares of the platanistid Zarhachis flagellator, two crescentic fo­ ramina that represent the exits of the olfactory nerves. Further­ more, Oelschlager and Buhl (1985a, 1985b) noted, in early on­ togenetic stages of Phocoena phocoena, an olfactory bulb that becomes reduced in later growth stages. The groove for the optic chiasma is located just below the ethmoidal fossae, and the optic nerves exited the skull at the very front of the brain, as in the living Monodontidae but un­ like the condition in the other odontocetes. The exits of the op­ tic nerves are located more medially (much closer to the mid- sagittal plane of the skull) and the chiasmatic groove is much narrower than in the other delphinoids, however. In Delphi­ napterus the distance between the exits of the optic nerves is generally two to three times greater in absolute value than in Odobenocetops. Posterolateral to the optic canal are the sphenorbital fissure and the foramen rotundum. They are large and the latter is su­ perposed to the former. They are separated by a bony wall (par­ tially broken) where they exit the skull. Among the Delphi­ noidea this morphology is present in Monodon. In the Delphinidae and Phocoenidae, the sphenorbital fissure and the foramen rotundum are most commonly confluent with the optic foramen. In Delphinapterus the optic foramen is isolated, but the sphenorbital fissure and foramen rotundum are confluent. The sphenorbital fissure of Odobenocetops is much larger than in Delphinapterus and approaches the size observed in the much larger Monodon, although in that genus the foramen ro­ tundum is smaller than in Odobenocetops. The sphenorbital fis­ sure is the passage for the nerves and blood vessels that provide innervation and blood supply to the anterior part of the skull, in­ cluding the oculomotor nerve (III), trochlear nerve (IV), oph­ thalmic branch of the trigeminal nerve (VI), abducens nerve (VI), anastomotic artery, and cavernous sinus. The foramen ro­ tundum is the passage for the maxillary branch of the trigeminal nerve (V2). The large size of these apertures in the skull of Odo­ benocetops is certainly related to the major modifications of the rostrum (large mobile eyes, premaxillary tusks, and inferred large upper lip), which would have required extensive blood and nerve supplies. The very large sphenorbital fissure in the walrus and the narwhal is very probably related to the growth of the tusks (and to the large upper lip, in the case of the walrus). The remainder of the internal view of the braincase shows mainly the cavities for the cerebral hemispheres, which appear to be proportionally shorter and wider than in the other Delphi­ noidea. Immediately dorsal to the lateral border of the foramen ovale, on the internal side of the braincase, is a longitudinal groove that seems to reach the foramen rotundum; it is absent from the other Delphinoidea, and its function is not clear. COMPARISONS AND AFFINITIES Comparing Odobenocetops peruvianus with other taxa is not easy, as most of the typical odontocete characters have been strongly modified. As already stated (Muizon, 1993a, 1993b), Odobenocetops is a delphinoid odontocete cetacean. Within the Delphinoidea, comparisons are not very productive because Odobenocetops differs from the other Delphinoidea in almost all the features observed. The exceptional morphology of the cetacean described herein, however, requires a statement of which characters allow a precise taxonomic definition of Odo­ benocetops. Five major features allow assignment of Odobeno­ cetops to the Cetacea: 1. The presence of large air sinuses in the auditory region, connected to well-developed pterygoid sinuses. All cetaceans have a peribullary sinus, but pterygoid sinuses are not found in all members of the order. Although they are absent in Pakicetus 242 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY FIGURE 16.—Endocranial cast of the holotype of Odobenocetops peruvianus (USNM 488252): a, dorsal view; b, anterior view; c, posterior view; d, lateral view. and in Protocetus, they are present in all other cetaceans. Air sinuses surrounding the auditory region and invading the ptery­ goid are not known in any other mammals. A pterygoid dupli­ cated into two laminae also is found in the Erinaceidae, the Macroscelidae, and the Tupaiidae. In these families, however, the fossa is always opened anteriorly, unlike what is observed in all cetaceans, and there are no peribullary sinuses. 2. The large supraorbital process of the frontal, which widely overhangs the orbital region. This feature is present in all adequately known cetaceans; however, it has not been ob­ served in Pakicetus because that part of the skull is unknown in this genus. 3. The narial fossae of Odobenocetops, which open dorsally and are not located at the apex of the skull. In the Sirenia the condition of the bony nares is significantly different from that in the cetaceans; the nares open anteriorly in a wide narial ba­ sin that opens dorsally and is located in the anterodorsal region of the skull. In the proboscideans, the narial fossae are not lo­ cated at the anterior end of the skull but they open anteriorly. A condition convergent with that of the cetaceans is found, how­ ever, in Macrauchenia, a Pleistocene South American lito- ptern. The cetacean condition is not found in any carnivores. 4. The absence of a true cribriform plate. This structure is located in the mesethmoid, at the anterior region of the cranial cavity, and gives passage to the olfactory nerves. The olfactory nerves have not totally disappeared in the cetaceans, however; small olfactory bulbs have been observed in several cetaceans, and Odobenocetops still retains small ethmoidal fossae (see NUMBER 93 243 above). The loss of the cribriform plate is obviously related to the aquatic life habits of the cetaceans. A well-developed crib­ riform plate is present in all the aquatic carnivores, but it is re­ duced in the Sirenia. 5. An immobile elbow. Although archaeocetes do not have an immobile elbow, this feature is constant in mysticetes and odontocetes. The partial forelimb illustrated in Figure 17 was found associated with a partial skull of Odobenocetops (as mentioned above) and shows the characteristic immobility of the elbow seen in the modern cetacean forelimb. Several other features allow us to classify Odobenocetops among the odontocetes, although the hyperspecialization of the skull partially hides the key character of the suborder. It has been noted elsewhere (Muizon, 1994) that the odontocetes are diagnosed by a posterior projection of the maxilla that covers the supraorbital process of the frontal totally or partially. In Odobenocetops, the maxillae have withdrawn medially and the supraorbital process is almost totally uncovered dorsally. Nonetheless, this key character of the odontocetes is still ob­ servable in Odobenocetops. The following features allow the assignment of Odobenocetops to the odontocetes. 1. Although strongly withdrawn, the maxillae still cover part of the medial part of the supraorbital processes medially, are expanded far backwards behind the nares, and are in contact with each other medially in that region. This condition is found neither in the Archaeoceti nor in the Mysticeti. 2. The fossa for the pterygoid sinus of Odobenocetops is greatly expanded dorsoventrally, and its dorsal limit is dorsal to the floor of the braincase. This condition is never found in any Archaeoceti or Mysticeti but is common in the odontocetes. In some very early taxa, such as agorophiids and Waipatia, the pterygoid fossa is not as developed as in the Holocene forms, but it is still clearly more developed than in the archaeocetes and mysticetes. 3. The premaxillary foramina are large. The premaxillary foramen of odontocetes is a supplementary passage for branches of the external carotid artery and for the maxillary di­ vision of the trigeminal nerve. This need for supplementary blood supply and innervation is related to the major specializa­ tions of the odontocete head, namely the melon and the air sac system. Enlarged premaxillary foramina are observed neither in the archaeocetes nor in the mysticetes. 4. The bones of the skull in the facial region are asymmetri­ cal, especially the premaxillae and the maxillae; the right bone is always more developed than the left. Although this asymme­ try is lacking or extremely reduced in several fossil groups (among others, Squalodontidae, Eurhinodelphidae, and Ago­ rophiidae), this pattern of asymmetry is known among mam­ mals only in the odontocetes. Furthermore, Odobenocetops presents a strong asymmetry of the tusks, as in the narwhal. 5. The narial passages are more vertical than those of ar­ chaeocetes and mysticetes. In Odobenocetops, because of the forward migration of the nares, this feature is not so well marked as in the other odontocetes. This is not a very satisfac- FiGURE 17.—Odobenocetops sp. (MNHN SAO 202): left forelimb, with partial humerus, partial ulna, and complete radius, in medial view. (Scale bar=3 cm.) tory character, however; in the early-diverging odontocetes it also is little developed, because the telescoping of the skull has not yet proceeded very far. As to the position of Odobenocetops within the odontocetes, some features allow its classification within the infraorder Del­ phinida (sensu Muizon, 1988c). The major synapomorphy of 244 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY the Delphinida is the protrusion of the pterygoid sinus within the palatine, which is thereby divided into a medial and a lat­ eral lamina. Such a structure is present in all eight families of the group (Lipotidae, Iniidae, Pontoporiidae, Kentriodontidae, Albireonidae, Monodontidae, Phocoenidae, and Delphinidae). The Delphinida also share the presence of a cranial hiatus that greatly enlarges the posterior lacerate foramen, which in turn is coalescent with the median lacerate foramen; in the Delphinida the periotic hangs in the hiatus, because it is connected to the skull by ligaments only. As mentioned above, Odobenocetops has a very large cranial hiatus. Furthermore, if the tympanic de­ scribed above actually belongs to Odobenocetops, the sigmoid morphology of its involucrum in medial view leaves little doubt that the walrus-like dolphin belongs to the Delphinida (see Muizon, 1988c: 163, fig. 3). The sigmoid involucrum of the tympanic of the Delphinida contrasts with the olive shape observed in the Platanistoidea (sensu Muizon 1987, 1991) and with the indented involucrum of the Eurhinodelphoidea and Ziphiidae. Odobenocetops peruvianus is a delphinoid because it bears a medial maxilla-premaxilla suture in the anterolateral edge of each narial fossa, which is formed by the maxilla. This charac­ ter, which was analyzed by Muizon (1988c: 199), is present in all the living Delphinoidea (with some individual variation), but it is not constant in the fossil groups, such as the Kentri­ odontidae. This feature is found only in the Delphinoidea, how­ ever, so its presence in Odobenocetops is evidence for its clas­ sification in that superfamily. Among the Delphinoidea, Odobenocetops is related to the Monodontidae by three derived features that have been re­ garded as synapomorphies of the Monodontidae (Muizon, 1988b:191): 1. A lateral lamina of the palatine that passes below the op­ tic gutter and joins the frontal laterally. 2. An alisphenoid and adjacent portions of the squamosal that are very thickened lateral to the foramen ovale and medial to the zygomatic process. 3. The presence of a very long and low temporal fossa (con­ trary to what Muizon (1988b:207) stated, it is clearly present in Denebola (Barnes, 1984, fig. 6)). This morphology corre­ sponds to an anteroposterior elongation of the middle region of the skull. Related to this modification are the anterior stretch­ ing of the supraorbital process of the frontal, the less transverse orientation of the optic gutter than in the other delphinoids, and the position of the exits of the optic nerves at the most anterior region of the brain and not posteroventral to the apex of the brain as in the other delphinoids. In fact, the transformation in the Monodontidae seems to be the consequence of the anterior migration of the eyes. That original modification is present to an extreme degree in Odobenocetops, where the supraorbital process of the frontal is tremendously stretched anteriorly, the optic gutter is oriented almost anteroposteriorly, the exits of the optic nerves are located at the anterior extremity of the brain, and the temporal fossa is greatly elongated anteroposteriorly. The morphology of Odobenocetops is obviously a much more derived stage of the condition observed in the Monodontidae. The Monodontidae also were diagnosed by two additional features: the extension of the maxilla-premaxilla suture on the lateral side of the nares (in the other Delphinoidea, the suture always remains on the anterior edge of the nares) and the re­ duction of the lateral lamina of the hamular process of the pterygoid (all the other delphinoids have well-developed lateral laminae) (Muizon, pers. obs.). These characters are not ob­ served in Odobenocetops, which is therefore less derived than the living Monodontidae for those features. The Monodontidae are further characterized by widely sepa­ rated pterygoid hamuli. This monodontid condition is greatly accentuated in Odobenocetops, where the pterygoids are very widely separated. This feature, however, also is known in some Delphinidae (generally as an individual variation) and in the Holocene Phocoenidae (it is absent in Piscolithax, a Tertiary phocoenid). In addition to the above evidence, the discovery of a partial skeleton (including skull and forelimb) of Odobenocetops sp. (MNHN SAO 202) in a slightly younger level at Sacaco con­ firms both that O. peruvianus is a cetacean and that its affinities lie with the Monodontidae. The forelimb shows the typical ce­ tacean modifications (immobility of the elbow), and its overall morphology recalls that of the Monodontidae (Muizon and Domning, 2002) (Figure 17). The monodontid affinities of Odobenocetops are clearly reinforced by the morphology of the inner ear, as shown above. In view of the exceptional specializations of Odobenocetops peruvianus, it was referred (Muizon, 1993a) to a new family, the Odobenocetopsidae, regarded as the sister group of the Monodontidae. Odobenocetops peruvianus shows obvious au­ tapomorphies, the most important of which are (1) loss of the typical cetacean rostrum and enormous thickening of the pre­ maxillae at the anterior region of the skull, (2) development on the premaxillae of large, downturned alveolar processes hous­ ing one large erupted right tusk and a small, probably unerupted left tusk, (3) extreme anterior position of the en­ larged nares and the dorsally facing orbit, (4) withdrawal of the frontal and maxilla from the posterodorsal angle of the skull, a condition that opens the temporal fossa dorsally, (5) reduction of the maxillae, which are excluded from the bony palate and only form part of the lateral wall of the skull, (6) extensive modification of the zygomatic process of the squamosal into the thick lateral wall of a large, anteroposteriorly elongated squamosal gutter, part of which houses the glenoid cavity and the middle sinus, (7) great reduction or absence of the premax­ illary sacs, and (8) a strongly thickened posteroventral region of the squamosal and a very solid, interdigitated parietal-squa- mosal suture. The occurrence of tusks in Odobenocetops is a convergence with Monodon; in the latter genus the large tusk of the male is NUMBER 93 245 implanted in the left maxilla, whereas in Odobenocetops the large tusk is implanted in the right premaxilla. Consequently, the tusk of Monodon and that of Odobenocetops are not homol­ ogous. Enlarged apical teeth or tusks are known in some other ceta­ ceans. Enlargement of apical teeth is common in squalodonts, where the anteriormost tooth of the premaxilla and of the den­ tary are enlarged and protrude anteriorly, being almost horizon­ tal (Dal Piaz, 1916; Kellogg, 1923). A similar condition is found in Kentriodon pernix from the middle Miocene of the Calvert Formation of Maryland. These teeth, in squalodonts and Kentriodon, do not show asymmetry. Kharthlidelphis diceros from the Oligocene of Georgian Republic (Mchedlidze and Pilleri, 1988) also possesses one pair of enlarged apical up­ per teeth that protrude anteriorly and horizontally. In this case the left tooth has a diameter approximately twice that of the right one, and the teeth are implanted in the maxillae. This con­ dition is basically similar to that of the narwhal, but the size difference between the teeth is much less and the right tusk is erupted. Kharthlidelphis is therefore very different from Odo­ benocetops, where the tusks are implanted in the premaxillae and the right tusk is much larger than the left one. Furthermore, the downturned tusks of Odobenocetops are unique among ce­ taceans. Functional Anatomy When compared with the other delphinoids, Odobenocetops shows drastic modifications of the skull that have been men­ tioned in the description above. These can be classified in three groups: (1) those related to the nasal sacs, basicranial sinuses, and auditory region, and consequently to sound production and reception; (2) those related to feeding; and (3) the tusks. In the following sections we analyze each of these groups of special­ izations. NASAL SACS, BASICRANIAL SINUSES, AND AUDITORY REGION Several authors have stated that the nasal sacs are at least partially implicated in sound production (Lawrence and Schev­ ill, 1956; Lilly, 1961; Lilly and Miller, 1961; Evans and Pres- cott, 1962; Norris, 1964, 1968, 1969; Norris and Evans, 1967). Mead (1975) critically reviewed the literature on the relations between the nasal diverticula and sound production and reached the conclusion that "the structures more likely to be in­ volved in sound production are those in the vicinity of the nasal plugs." Mead (1975) also stated that the premaxillary sacs could be used as sources for air during phonation. The role of the basicranial air sinus has been debated. Fraser and Purves (1960) stated that their function is the phonic isola­ tion of the periotic from bone-conducted sounds. The sound waves transmitted by bone conduction are reflected on the in­ terfaces between bone and soft tissues (which behave like liq­ uid in sound transmission) and the foamy filling of the air sinus (which behaves like air in sound transmission). The sound waves are therefore forced to enter the cochlea through the ex­ ternal auditory meatus, thus providing good directional hear­ ing. McCormick et al. (1970) rejected this idea, however, stat­ ing that the acoustic vibrations reach the ear through the tissues of the head, and Norris (1964, 1968, 1969) and Brill et al. (1988) proposed that the echolocation sounds may return to the body "by the way of the intramandibular fat body which acts as a passive wave guide and enter the middle ear via the tympanic bulla which transmits sounds directly to the ossicular chain and cochlea through the processus gracilis," which attaches the malleus to the tympanic. The melon is a fatty organ located on the dorsal side the ros­ trum anterior to the nasal complex. In some species, the melon reaches the apex of the rostrum; this is generally true, for ex­ ample, in the Globicephalinae. The melon is embraced by the medial portion of the rostral muscle (Mead, 1975), also called the pars labialis of the maxillonasolabialis (Lawrence and Schevill, 1956) and the nasolabialis profundus pars lateralis (Rodinov and Markov, 1992). Below the posterior part of the melon are the nasal plug muscle and the premaxillary sacs (Mead, 1975), in which it is partially imbricated. The most commonly suggested function of the melon is as an acoustic lens (Lilly, 1961; Norris, 1964, 1968, 1969; Wood, 1964). Al­ though it seems clear that the melon is involved in sound recep­ tion and transmission (Norris and Harvey, 1974; Mead, 1975), its function remains unclear. On the skull of Odobenocetops, three features can be ob­ served that denote lesser ability in sound production and/or transmission than in the other delphinoids: (1) the premaxillary sacs and nasal plug muscles were either absent or very reduced; (2) the melon was vestigial or absent; and (3) apparently the nasal diverticula of the nasal passages were very reduced or ab­ sent. The latter statement requires explanation. The nasal diverticula of the Delphinidae have been well de­ scribed by Lawrence and Schevill (1956) and Mead (1975). The nasal sacs and the communication between them are con­ trolled by the various layers of the pars nasalis of the maxillo­ nasolabialis muscle. Most of that muscle is attached to the dor­ sal side of the skull, lateral to the nasal opening and to the premaxilla, and posterior to the antorbital notch and the nuchal crest. It is divided into six layers according to Lawrence and Schevill (1956), and five layers according to Mead (1975). Rodinov and Markov (1992) recognized nine muscles, four in their nasolabialis group (which includes the nasal plug muscle, their nasolabialis profundus pars anterior medialis) and five in their maxillonasalis group. None of these layers or muscles is attached to the parietal. In Odobenocetops, the anterior with­ drawal of the frontal and maxilla and the loss of the roof of the temporal fossa are supposed to have considerably reduced the attachment area of the pars nasalis of the maxillonasolabialis. 246 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY This statement assumes that very few or no muscle fibers would have migrated onto the broadly uncovered parietal, pos­ teriorly. The five layers recognized by Mead (1975) are the pars posteroexternus (PE), the pars intermedius (I), the pars an- teroexternus (AE), the pars posterointernus (PI), and the pars anterointernus (Al). The first four layers (PE, I, AE, PI) attach to the lateral edge of the supraorbital process and to the tempo­ ral crest. The pars Al attaches mainly to the ascending process of the maxilla and to the lateral border of the premaxilla. Consequently, we can assume that the great reduction of the maxilla would have resulted in an extreme reduction (or loss) of Al. Furthermore, the disappearance of the temporal crest and the reduction of the lateral edge of the supraorbital process (which is very thin and strongly notched in dorsal view) also suggest a great reduction of the four layers PE, I, AE, and PI. Even if all the nasal diverticula totally disappeared, however, it is obvious that at least part of the maxillonasolabialis (pars an- teroextemus) must have been retained, as this layer is involved in opening the blowhole. Furthermore, Odobenocetops has large nasal apertures, and the problem is to determine whether it had nasal plugs or not. Given the size of the bony nasal passages, if nasal plugs were present, they would have been very large and consequently would have been moved by equally developed muscles. We ob­ served, however, that the nasal plug muscle, if present, was very small. Furthermore, Mead (1975:53) stated that the "PI and Al also serve to seat the nasal plugs in the orifice of the bony nasal passage." The Al originates on the posterior border of the supraorbital process and on the temporal crest, and the PI originates on the ascending process of the maxilla. As stated above, these layers were probably very reduced in Odobenoce­ tops, a condition that would be in agreement with the absence of nasal plugs. Furthermore, Lawrence and Schevill (1956) and Mead (1975) have shown that the nasal plugs are tightly imbri­ cated with the dorsal surface of the premaxillary sacs and with the melon and that there is a sort of histological continuity among these organs. Consequently, the absence or great reduc­ tion of the premaxillary sacs and the absence of a melon would be consistent with the absence or reduction of the nasal plugs. If the nasal plugs were actually present and well developed, there would be an inconsistency between their large size on the one hand and the reduction of the nasal plug muscle, premaxil­ lary sacs, and melon on the other. From this discussion it there­ fore follows that the nasal plugs were very probably absent or vestigial in Odobenocetops. There seems to be a consensus that the nasal plugs and the nasal diverticula are related to sound production (Lilly, 1961; Lilly and Miller, 1961; Norris, 1964, 1968, 1969; Norris and Evans, 1967; Kleinenberg et al., 1969; Diercks et al., 1971; Norris et al., 1971; Evans and Maderson, 1973; Mead 1975; Dormer, 1979; Ridgway et al., 1980). It also is possible that sound could be produced by the larynx (Lawrence and Schev­ ill, 1956; Purves, 1967; Schenkkan, 1973). Some authors have suggested that both the nasal sac system and the larynx are im­ plicated in sound production (Lilly and Miller, 1961; Evans andPrescott, 1962; Schevill, 1964; Evans, 1967). Ridgway et al. (1980), however, clearly rejected the possibility of any sound production in the larynx and stated that sounds are pro­ duced by the nasal system only. Consequently, Odobenocetops probably had no (or vestigial) nasal plugs, no (or vestigial) melon, and very reduced premax­ illary sacs, if any; and, as the maxillonasolabialis was appar­ ently very reduced in comparison to the other delphinoids, it also is likely that the other nasal diverticula were either re­ duced or totally absent. If so, then Odobenocetops had little (or no) ability to produce sounds (in the usual cetacean way) and therefore to echolocate. As indicated earlier, the inner-ear structure implies that Odo­ benocetops, like most mammals, was capable of perceiving ul­ trasonic sounds (frequencies >20 kHz), but whether Odobeno­ cetops could echolocate cannot be determined from ears alone. High-frequency facility is not synonymous with echolocation. The ability to echolocate implies the production of self-gener­ ated, beamed, gated signals and analysis of the corresponding echoes. Analysis of ambient sound is passive listening and is common to all mammals. Passive listening clearly provides considerable information about the immediate environment (e.g., directionality or the relative distance of two sound sources), but it does not provide spatial or textural information with the level of resolution commonly obtained through bioso- nar. Biosonar is an extraordinarily sophisticated acoustic imag­ ing system that conventionally uses very high frequencies be­ cause the level of detail that can be transduced is related to the wavelength of the ensonifying signal. The stronger the evi­ dence against the presence of a melon, which is directly impli­ cated in outgoing signal generation and control, the less sup­ port there is for functional echolocation by Odobenocetops. The absence of nasal plugs would be critical evidence against echolation, as that structure and the diagonal membrane seem to represent the major sound producers in odontocetes, al­ though it is not possible to refute definitively the hypothesis that Odobenocetops had a reduced air sac system and nasal plugs. Whatever the morphology was, it is very probable that Odo­ benocetops had little echolocational ability, if any—a condition that may have been compensated for by good binocular vision. The orbit of Odobenocetops is proportionally much larger than in any other cetacean. The distance between the apices of the postorbital and antorbital processes is close to that in Monodon even though Odobenocetops is an animal approximately 15% smaller. Furthermore, the dorsal shifting of the process and the depth of the orbital notch clearly indicate that dorsal or anterior binocular vision was possible. Furthermore, good dorsal or an­ terior binocular vision is in agreement with the probable feed­ ing posture of Odobenocetops (see "Scenario for the Feeding Mode of Odobenocetops peruvianus," below). NUMBER 93 247 FEEDING ADAPTATIONS The modifications of the skull of Odobenocetops are corre­ lated with major modifications of its musculature. Four aspects should be considered: mandibular movements, head move­ ments, morphology of the palate, and inferred presence of an upper lip and vibrissae. MANDIBULAR MOVEMENTS.—In Odobenocetops, the with­ drawal of the maxillae and frontals from the dorsal side of the braincase has opened dorsally the temporal fossa, which in most odontocetes is roofed. The parietals are widely uncovered dorsally and secondarily increased in size so that they occupy most of the dorsal surface of the braincase. In the other delphi­ noids, the temporal fossa is roofed by a lateral expansion of the frontal and the maxilla. The parietals are small bones restricted to the lateral sides of the skull and do not contact each other. The temporalis is a small weak muscle whose origin is located on the parietal, on the lateral side of the skull, and whose inser­ tion is on the coronoid process of the dentary. Its action is to raise the mandible to close the mouth. If muscle insertions and origins are at least partially related to particular bones (which is known to be not always true), the great development of the pa­ rietal of Odobenocetops might have led to a great extension of the temporalis on the dorsal side of the skull. Consequently, Odobenocetops probably was capable of much stronger adduc­ tion of the mandible than are other odontocetes. Another important elevator of the mandible is the masseter. In the dog, the masseter is divided into three layers. The super­ ficial and middle layers originate on the lateral side of the zy­ gomatic arch (on the jugal and squamosal), and the deep layer originates on its medial side (Miller et al., 1964). As for the od­ ontocetes, Howell (1927) described two layers in Neopho- caena, superficialis and profundus, but only the former had a bony origin, on the zygomatic process of the squamosal and on the postorbital process of the frontal. The superficial layer in­ serts on the posteroventral angle of the mandible and the deep layer on the dorsal edge of the dentary in front of the insertion of the temporalis. In Odobenocetops, the large distance exist­ ing between the postorbital process of the frontal and the zygo­ matic process might indicate that the masseter was divided into two clearly defined elements. Several features suggest, how­ ever, that the masseter did not originate on the postorbital pro­ cess, but rather that it was attached only on the strong zygo­ matic process of the squamosal. This structure is the very stout lateral wall of the squamosal gutter described above. It is a thick bony wall strongly attached to the skull all along its length. Furthermore, the squamosal-parietal contact is not squamous and vertical as in most other mammals; it is a very solid, transversely oriented, interdigitated suture apparently ca­ pable of resistance to strong anteroposterior stresses. Conse­ quently, the architecture of the areas of origin of the masseter indicates a muscle much stronger than in the other delphinoids. On the ventral side of the skull the pterygoid muscles also are mandibular adductors. In a terrestrial mammal (the dog, cf. Miller et al., 1964), both pterygoideus muscles (lateralis and medialis) originate on the lateral side of the skull, on the ptery­ goid but also on the orbitosphenoid. In Neophocaena, Howell (1927) observed a very reduced muscle originating on the pterygoid membrane and on the lateral edge of the pterygoid bone adjacent to it. The "insertion is not upon the mandible but on the tough tissue near the ear bone" (Howell, 1927:23). In Kogia, however, Schulte and Forest Smith (1918) observed two relatively strong pterygoid muscles (intemus and extemus) that insert upon the medial side of the dentary. Seagars (1982) found that in long-jawed delphinids with numerous teeth, the pterygoideus intemus (=medialis) is in fact the dominant ad­ ductor muscle, followed in importance by the temporalis, mas­ seter, and pterygoideus externus (=lateralis). In species with fewer teeth and broader jaws (Globicephala, Grampus), the temporalis is more powerful relative to the pterygoideus inter- nus, and in Orcinus the temporalis is actually the dominant muscle (Murie, 1870, 1973). In Odobenocetops, the pterygoid shows a very strong laterally convex crest in the position of the pterygoideus medialis of the dog and Kogia, and of part of the undivided pterygoid of Neophocaena. This crest is more devel­ oped than any structure observed in the other odontocetes and indicates, in Odobenocetops, a pterygoid muscle much stronger than in the other delphinoids. Together with the evidence of a large temporalis cited above, this seems consistent with the pat­ tern documented by Seagars (1982) in delphinids, which would lead us to expect a large temporalis and internal pterygoid in connection with the presumably short, toothless mandible of Odobenocetops. The main mandibular depressor muscle in terrestrial mam­ mals is the digastricus. In the dog it originates on the anterior side of the paroccipital process of the occipital and inserts upon the ventral border of the mandible (Miller et al., 1964). In the odontocetes, Howell (1927, 1930) termed it monogastricus, and the origin is on the anterolateral border of the thyrohyal and/or basihyal. Apparently there is no attachment to the skull. The tremendous thickening of the squamosal and, to a degree, of the occipital at the posterolateral angle of the skull in Odo­ benocetops, however, seems to indicate considerable muscular stress in that region. Although it is therefore possible that part or all of the digastricus muscle was attached to the posterolat­ eral angle of the skull, a condition that would have to be re­ garded as a reversal, the neck muscles (see below) are more probably responsible (at least partially) for the strong develop­ ment of that region of the skull. The masticatory musculature of Odobenocetops thus seems to have been much more powerful than in any other odonto­ cetes. Considering the lack of maxillary teeth, it indicates a very peculiar mode of feeding involving powerful movements of the lower jaw and perhaps of the tongue, gular, and hyoid musculature. Although there is no direct evidence of these lat­ ter muscle groups, the very deep and arched palate suggests a very large tongue and a consequently strong throat musculature (see below). 248 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY HEAD MOVEMENTS.—The muscles involved in head move­ ments are numerous and strong. In odontocetes, the most im­ portant are the semispinalis, rectus capitis, splenius, multifidus, sternomastoideus, longissimus, and scaleni. The first four mus­ cles insert on the occipital shield, and not much can be said about their morphology in Odobenocetops. The three latter muscles insert on the posterior border of the paroccipital pro­ cess of the occipital and/or on the zygomatic process of the squamosal. The sternomastoideus and the longissimus insert on the posterior region of the stemomastoid fossa and the lateral edge of the paroccipital process. In Odobenocetops this region is greatly expanded ventrally and forms the tremendously thickened posterolateral angle of the squamosal. The enlarge­ ment of the area of insertion of these muscles and the stoutness of the bone in this region suggest that the sternomastoideus and longissimus of Odobenocetops were very powerful muscles. The same could be said about the scaleni, which insert on the posterior surface of the paroccipital process and basioccipital crests. The bony origin of the sternomastoideus is on the ante­ rolateral edge of the sternum; the bony origin of the longissi­ mus is on the transverse process of the anterior thoracic verte­ brae (Pabst, 1990); origin of the scaleni (dorsalis and ventralis) is on the first ribs. These three muscles, if acting bilaterally, are depressors of the head, and if acting unilaterally, are lateral ro­ tators of the head. The region of the nuchal crest located behind the stemomastoid fossa also is extremely thickened. In Neophocaena and Monodon the splenius and semispinalis capi­ tis insert in this area. The lateral extremity of the insertion of these muscles also is on the posterior part of the upper part of the stemomastoid fossa. If acting bilaterally these muscles are levators of the head. Consequently, the assumed strength of the sternomastoideus, longissimus, and scalenus in Odobenocetops would indicate strong and active vertical and transverse movements of the head or at least a need for a strong control of these movements. This conclusion is in agreement with the morphology of the condyles and of the supracondylar fossa, which would allow significant vertical movements of the head (see above). It is further corroborated by the relatively large size of the semicir­ cular and vestibular canals, in Odobenocetops as in Delphi­ napterus, which are apparently related to the mobility of the neck (see above). PALATE MORPHOLOGY.—The palate of Odobenocetops is very wide, very deep, and arched. The morphology of its ante­ rior part is dominated by the exceptionally developed premax­ illae and the large vomer. The palate is very wide posteriorly because of the lateral expansion of the pterygoid, which forms a ventrolaterally oriented wing. The palatines also are very en­ larged in the posterior region of the palate. Between the alveo­ lar process of the premaxilla and the pterygoid wing, a deep, rounded notch forms the narrowest (middle) portion of the pal­ ate. This notch lies exactly in the path of a prolongation of the squamosal gutter and was obviously for the passage of the mandible. Seen ventrally, the notch is wide and almost semicir­ cular; viewed along the axis of the squamosal gutter, though, it is much narrower, as the lateral plane of the palate is oblique in relation to a line joining the squamosal gutter to the apex of the rostrum, the supposed axis of the dentary. If the dentary of Odobenocetops was straight as it is in all the other odontocetes, then the dentary in its middle portion was only slightly thicker than in the other delphinoids; however, it was probably very high, as it needed to be strong to bear the stresses of the power­ ful musculature. The anterior part of the mandible probably was greatly thickened to partially fill the anterior region of the palate. Another possibility is that at their anterior ends, the den­ taries were as high as the concave portion of the medial side of the premaxillary process (i.e., its proximal two-thirds, approxi­ mately 15 cm) and formed two divergent processes or expan­ sions complementary to those of the premaxillae. This mandib­ ular morphology would somewhat resemble that observed in the borhyaenoid marsupial Thylacosmilus from the Pliocene of Argentina. Such a morphology would allow the apex of the lower jaw to fit perfectly in the anterior region of the palate, thereby allowing a perfect closing of the mouth. An even slightly deep, wide, and/or vaulted palate is un­ known among other cetaceans, so the morphology of Odobeno­ cetops is unprecedented. In non-cetacean mammals, a deep, arched palate is known in the Odobenidae (Odobenus ros­ marus, Valenictus chul'avistensis) and in the Otariidae (Otaria flavescens and to a lesser extent Phocarctos hookeri). As men­ tioned above, the deep palate of Odobenocetops is consistent with a large tongue and consequently strong tongue muscles. Also, the gular musculature must have been fairly powerful, as it is related to the tongue and mandible movements. UPPER LIP AND VIBRISSAE.—Muizon (1993a, 1993b) has suggested the presence in Odobenocetops of a strong upper lip. That hypothesis is maintained herein. Lips are known in several other odontocetes and are especially well developed in the bel­ uga (Kleinenberg et al., 1969:80). Although these authors state that the lips of the beluga are devoid of musculature and not movable, figure 5 of Brodie (1989:132) seems to demonstrate considerable mobility of the beluga's lips. Furthermore, Mead (1975:39) mentioned that part of the lateral rostral muscle of the delphinoids is associated with the upper lip and stated that "[t]he lips of Cetacea are generally considered to be immobile, but in view of the great expanse of lip surface in animals such as Globicephala and Grampus and the relatively large mass of muscle inserting into the connective tissue of the lips in these forms, a certain amount of mobility may be present and may be important in feeding." A strongly developed, mobile upper lip in Odobenocetops could therefore result from the enlargement of a structure already existing in the other delphinoids. We mentioned above the strong muscle attachments on the dorsal and anterior face of the apex of the rostrum. Given that the medial rostral muscle is inserted into the melon (Mead, 1975:10), which Odobenocetops lacks, both the medial and lat­ eral portions of the rostral muscle might be implicated in upper lip movements in Odobenocetops. Consequently, the muscle arising from the subapical rims described above (passing later­ ally to the premaxillary foramina, converging at the apex of the NUMBER 93 249 rostrum, and enclosing the triangular area where, in other del­ phinoids, the nasal plug muscle is attached) would represent part of the medial portion of the rostral muscle. The remaining part of the medial portion of the rostral muscle, together with the lateral portion, would create the strong muscle attachments observed on the anterodorsal sides of the alveolar processes of the premaxillae. The whole rostral musculature of Odobenoce­ tops would therefore be related to upper lip movements. The occurrence of a large upper lip in Odobenocetops is con­ sistent with the presence of the large neurovascular foramina observed on the anterior and ventral borders of the alveolar process and with the presence of the very large palatine fora­ men, which could have provided extensive blood and nerve supply to the ventral region of the lip. The dorsal region of the lip was irrigated and innervated via the very large maxillary (and probably premaxillary) foramina, which also were related to the blood and nerve supply of the pulp cavities of the tusks. As noted above, the facial nerve of Odobenocetops is signifi­ cantly larger than in the other delphinoids except Monodon (Table 3). The large diameter of the facial nerve canal in the pe­ riotic is consistent with the presence of a large upper lip and vibrissae. Muizon (1993a, 1993b) suggested that the upper lip of Odo­ benocetops could have carried vibrissae, evolved convergently with those of the walrus. The extensive vascularization and in­ nervation of the upper lip and the very spongy nature of the bone of the anterior side of the rostrum suggest that such struc­ tures could possibly have been present. Furthermore, vestigial vibrissae are well known in some adult cetaceans (mysticetes, Inia) and have been observed in several newborn or fetal del­ phinoids and Pontoporia (Bourdelle and Grasse, 1955; Yablokov and Klevezal, 1962; Tomilin, 1967; Best and da Silva, 1989). The vascularization of the anterior side of the ros­ trum of Odobenocetops also could be related to the lip only, however, so the presence of vibrissae remains undemonstrated. COMPARISON WITH Odobenus.—Several features of Odo­ benocetops are convergent with the morphology of the walrus. They are (1) the large, deep, vaulted palate; (2) the presence of tusks; (3) the wide and low occipital, in posterior view, and the ventrolateral expansion of the ventrolateral angle of the skull; (4) the thickness and the extreme stoutness of the bone forming the ventrolateral angle of the skull and of the alisphenoid; (5) the well-developed upper lip and the presence of vibrissae (not certain in Odobenocetops); and (6) the presence of a rough, ir­ regular, and expanded area for attachment of tough or horny tissue on the edges of the mouth (inferred for Odobenocetops). The palate of Odobenocetops, however, differs strongly from that of Odobenus. It is much wider and deeper. It is composed of the premaxillae, palatines, vomer, and pterygoid, but the maxillae do not crop out on the palate as in most other mam­ mals. There are no maxillary teeth in Odobenocetops, although it is known that some fossil odobenine walruses (Valenictus chulavistensis) also lacked maxillary teeth (Demere, 1994). The vomer is very large and lozenge-shaped, as (but much larger than) in most other odontocetes. The tusks of Odobenoc­ etops are premaxillary and strongly asymmetrical, whereas they are maxillary and symmetrical in Odobenus. The poster­ oventral angle of the skull is made up of the paroccipital pro­ cess and the squamosal, whereas in Odobenus it is formed by the tremendously enlarged mastoid process of the periotic. The attachment area of tough tissue is located on the anterior edge of the premaxilla; in the walrus it is on the dorsal side of the mandibular symphysis. Muizon (1993a, 1993b) regarded Odobenocetops peruvianus as a walrus-convergent bottom feeder. We have noted above that Odobenocetops and Odobenus share several strong simi­ larities in their skull morphology. Functional comparisons will help determine whether the adaptations of Odobenocetops could fit with the walrus's mode of feeding. The living walrus feeds mainly upon benthic invertebrates, such as thin-shelled bivalves (Mya, other clams, and mussels), gastropods, molting crustaceans, holothurians, and tunicates, but rarely upon cephalopods, fish, or other vertebrates. Prey buried in the substrate, such as clams, are excavated by jetting water from the mouth (Kastelein and Mosterd, 1989; Kastelein et al., 1991). As suggested by Vibe (1950) and clearly ex­ pressed by Fay (1982), the walrus does not crush the shells of the mollusks it ingests; rather, it holds them between its lips and jaws at the front of the mouth and sucks out their siphons, using its mouth as a vacuum pump. Fay (1982:171) described the walrus mouth as follows: "The extraordinary vacuum pump of the walrus ... is powered by very large lingual retractors and depressors (M. styloglossus, hyoglossus, and genioglossus), complemented by the highly vaulted palate, the long firmly ankylosed mandibular symphysis, and the unusually powerful m. tensores veli palatini and m. buccinatorius, which provide rigidity to the walls of the 'cylinder.' The tongue is the 'piston.' The small rigid oral aperture, complemented by ample muscu­ lar lips, insure that the full effect of the vacuum is exerted only on objects held in the incisive area at the front of the mouth." Gordon (1984) confirmed that the vacuum is created by move­ ment of the tongue directly rearward. Fay (1982:167) also stated, concerning the tough tissue covering the dorsal surface of the mandibular symphysis: "The rough cornified surface seems admirably suited for grasping and holding slippery prey; I believe that it functions also to hold molluscan shells while their contents are removed by suction." When searching for food, the walrus moves along the bottom with its head down and its body at an angle to the bottom of about 10°-45° (Fay, 1982:164). The vibrissae are the sensitive organs that are constantly in contact with the bottom and trans­ mit information to the animal. As mentioned by Fay (1982), the walrus feeds in darkness and "the eyes are comparatively smaller than those of other pinnipeds;" "they may be the least important as sensors" and "the animals probably locate their prey by vibrissal contact." Fay (1982:164) also mentioned that excavation of buried animals may be achieved by "rooting" with the snout: "the pattern of vibrissal abrasion, the greater comification of the upper edge of the snout..., and the powerful cephalo-cervical musculature of the walrus are unlike those of 250 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY FIGURE 18 (left).—Odobenocetops peruvianus, holotype (USNM 488252): anterolateral view of the skull in a feeding position. V. -- * i „ > any other pinnipeds. Each points to frequent use of the snout as a digging organ." This mode of feeding creates long, sinuous furrows in the seafloor (Nelson and Johnson, 1987), each pre­ sumably representing a single bout of feeding by a walrus on a single dive. In Odobenocetops, although the mandible is unknown, sev­ eral aspects of the skull morphology are in perfect agreement with suction bottom feeding (see Figures 18, 20): The very large vaulted palate (even larger than in the walrus); the in­ ferred comifred (or strongly fibrous) edge of the upper lip; the fairly well-developed masticatory musculature (for a cetacean), close, in strength, to that of the walrus; the powerful occipital musculature (at least the scalenus, sternomastoideus, and long­ issimus); and the significant atlanto-occipital movements de­ noted by the very convex and salient occipital condyles and the large supracondylar fossae. Contrary to walrus morphology, though, the eyes of Odobenocetops were large, and good binoc­ ular vision compensated for the probably poor echolocational ability. Even walruses have good anterodorsal binocular vision and seem to forage visually when water clarity permits, how­ ever (Kastelein and Wiepkema, 1989; Kastelein et al., 1993). Odobenocetops had no maxillary teeth; like the living wal­ rus, it was a sucking feeder that did not have the masticatory apparatus necessary to crush any hard exoskeleton of its prey. Fay (1982:166) has shown that the teeth of the living walrus are not used for crushing, and it is noteworthy that some fossil odobenids (Demere, 1994) had lost the postcanine teeth alto­ gether. Another noteworthy feature of Odobenocetops is the strongly modified zygomatic process with a large squamosal gutter and a tremendously resistant parieto-squamosal suture. As it has no equivalent in any living mammal, this structure is difficult to interpret. If, however, this structure is related to the very strong masticatory musculature mentioned above, it likely permitted very powerful (though probably not very extensive) movements of the lower jaw. The exceptional strength of the mouth of Odobenocetops would thus have allowed it to main­ tain its prey extremely firmly in its mouth. The extreme devel­ opment of the squamosal gutter and its high position are diffi­ cult to explain because the mandible is unknown. Odobenocetops also shares with Odobenus the possession of tusks, but the large premaxillary alveolar processes of Odo­ benocetops are absent in the walrus. Fay (1982:136) stated that the tusks of the walrus are not related to feeding but instead NUMBER 93 25; play a social role. The same has been stated for the narwhal (Nishiwaki, 1972; Hay and Mansfield, 1989). Consequently, we assume that the tusks of Odobenocetops were used more as social organs than as feeding implements (see below for further discussion of the tusks). The deep premaxillary fossae border­ ing the anterior extremity of the vomer on the palate are absent in all the other odontocetes, and nothing similar is observed in the walrus. Their function has not been elucidated. COMPARISONS WITH MARINE MAMMALS OTHER THAN Odo­ benus.—Delphinapterus leucas: Kleinenberg et al. (1969: 89-90) presented an interesting interpretation of the functions of the tongue of the beluga for feeding. According to these au­ thors, the tongue is a very mobile and strong organ that has three functions. First the tongue orients the prey in the mouth. Second, the animal "draws its tongue back, pressing the prey to the palate and forcing it to the entrance of the pharynx." Third, "the tongue forces the water out of the mouth, preventing its entrance into the intestinal tract." The tongue thus appears to be a powerful and extremely specialized organ for food ingestion. It is therefore very probable that Odobenocetops, like all living odontocetes, had a powerful, enlarged (given the size of the palate) tongue, and that it could have been involved in a pump­ ing feeding action as is observed in the walrus. Bottom feeding is well established for belugas; worms and benthic animals are often found in their stomachs (Matthews, 1978). Belugas also are capable of powerful water-jetting simi­ lar to that observed in the walrus (D.P. Domning, pers. obs., and J.G. Mead, pers. comm., 1992). Bottom feeding also is known in Globicephala, and water-jetting in Orcinus and Glo- bicephala (Werth, 1992). Kolponomos clallamensis and K. newportensis: Tedford et al. (1994) described the skull of Kolponomos, an aquatic ursid from the early Miocene of Washington and Oregon. Among the most peculiar modifications of the skull of Kolponomos are the extremely enlarged mastoid processes, which are oriented ven- trolaterally as is observed in the walrus. According to Tedford et al., Kolponomos could have fed upon benthic hard-shelled invertebrates, mainly mollusks but also echinoids. Conse­ quently, the substantial and strong movements of the head must have necessitated strong neck musculature. Furthermore, Kolponomos had anterior binocular vision, which assisted its search for food, as in Odobenocetops and Odobenus (Kastelein et al., 1993). Tedford et al. (1994) also suggested that Kolponomos could have had enhanced tactile sensitivity of the lips and the muzzle, possibly possessing a large upper lip and tactile vibrissae like the walrus. If that was actually the case, then Odobenocetops also would resemble Kolponomos, in having both good binocular vision and an en­ larged, highly tactile upper lip. That bottom feeders could rely upon binocular vision and/or tactile sensibility of the upper lip and vibrissae is in agreement with the apparent lack (or reduc­ tion) of echolocational ability in Odobenocetops. Kolponomos, however, was very different from Odobenocetops in its mode of ingestion of prey. Kolponomos had very strong sea-otter-like crushing teeth, which were used to break the hard shells of benthic invertebrates (Tedford et al., 1994). Desmostylians and Sirenians: Like Odobenocetops, most marine mammals that are believed to have dug for food in the seafloor possess large ventrolateral expansions of the poster­ oventral angle of the skull, which presumably reinforced the neck when the animal was searching for its food, with its head down and the body in an oblique position in the water. Desmo­ stylians provide another example of this; they were evidently substrate-feeding herbivores (Domning et al., 1986) and all shared an enlarged paroccipital process (Ray et al., 1994). Fur­ thermore, the peculiar sirenian Miosiren, which has been sus­ pected of molluscivory on the basis of its dentition and rein­ forced palatal region, also has enlarged post-tympanic processes (Sickenberg, 1934). Other tusked sirenians, however, do not, and at least the most specialized of them may have used their jaw muscles instead of their neck muscles for digging (Domning, 1989). The Holocene sirenian Dugong dugon (Muller) superficially resembles Odobenocetops in having a pair of slender, relatively straight, and open-rooted upper incisor tusks that are largely enclosed in prominent, downward-directed premaxillae. In the dugong, however, these tusks are symmetrical, are not strongly divergent, project at most a few centimeters outside the gum, and are directed anteroventrad rather than posteroventrad. Al­ though tusks of fossil dugongids were surely used in feeding (probably to excavate sea grass rhizomes; Domning, 1989), those of the living species seem to have lost this function and have become relegated to purely social uses, such as fighting between males (Preen, 1989; Domning, unpubl. data). Evi­ dently as a result of this restriction in function, tusks of D. dugon (uniquely among sirenians) have become sexually di­ morphic, normally erupting only in the males (Marsh, 1980). It thus appears that the tusks of the herbivorous dugong can shed no additional useful light on the biology of the presumably car­ nivorous Odobenocetops. Scenario for the Feeding Mode of Odobenocetops peru­ vianus: Odobenocetops lived in the early Pliocene on the coast of Pern, in the shallow waters of the bay of Sud-Sacaco, probably close to the shore (Marocco and Muizon, 1988). It was a sucking feeder preying upon benthic invertebrates such as thin-shelled mollusks and/or molting crustaceans, which were abundant in the Pisco Formation (Muizon 1981; Muizon and DeVries, 1985; Carriol et al., 1986). It searched for food using a position similar to that of the walrus. The head main­ tained contact with the bottom while the body was at an angle to the bottom, probably close to the angle of the premaxillary process with the anteroposterior axis of the skull (-45°). The tail helped in keeping that position and, as in other cetaceans, provided propulsion (as mentioned by Fay (1982:164), the wal­ rus also uses the posterior part of the body to move forward, but by means of the hind limbs). The forelimbs probably had a stabilizing role, as in the walrus, and would have helped to maintain the upper lip and vibrissae in contact with the bottom. It used its very good dorsal binocular vision and its large and 252 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY FIGURE 19.—Odobenocetops peruvianus, two possible reconstructions of the head: a, hypothesis without vibrissae, a "cetacean-like" interpretation; b, hypothesis with vibrissae, a "walrus-like" and more speculative interpretation (sculptures by Mary Parrish, Department of Paleobiology, NMNH). highly sensitive upper lip, possibly assisted by strong vibrissae, to search for food (Figures 18-20). Most probably, little or no echolocation was involved. Once found, buried prey, such as clams, was probably excavated by water-jetting, as in the wal­ rus, then caught in the anterior part of the mouth and strongly held in this position by the powerful lips and jaws. Then dors­ oventral and/or anteroposterior movements of the tongue would have transformed the mouth into a sort of vacuum pump, and part of the prey was sucked out of its shell and in­ gested. Consequently, Odobenocetops was a highly specialized ceta­ cean that combined adaptations of both Odobenus (water-jet­ ting and sucking) and Kolponomos (good binocular vision), and therefore was probably the best-adapted known bottom- feeding marine carnivore. This proposed feeding scenario is not unexpected for an od­ ontocete. Suction feeding has been reported many times in od­ ontocetes (Kleinenberg, 1938; Wilke et al., 1953; Ray, 1966; Rae, 1973; Matthews, 1978; Crawford, 1981; Seaman et al., FIGURE 20.—Artist's reconstruction of Odobenocetops peruvianus in feeding position (original painting by Mary Parrish, Department of Paleobiology, NMNH). 1982; Martin, 1990; Heyning and Mead, 1996). An excellent review of the topic was presented by Werth (1992). According to Werth, it is in fact the absence of suction feeding that is un­ usual among odontocetes. Werth reported suction feeding in a number of species, both through direct observation: Delphi­ napterus leucas and Globicephala melaena; and inferred from anatomical or ecological evidence: Berardius bairdi, Mesopl­ odon layardi, Monodon monoceros, Phocoena phocoena, Pho- coenoides dalli, Physeter macrocephalus, and Tursiops trunca­ tus. Werth (1992:37) also reported a personal communication of W. Walker describing "a Tursiops stomach full of fresh, un­ digested siphons removed from clams; this type of diet (and in­ ferred feeding method) is strikingly similar to that of suction- feeding walruses." The observations made by Werth and others make the feeding adaptation inferred for Odobenocetops peru­ vianus much more likely than could be suspected at first glance. Possible Functions of the Tusks and Alveolar Processes The most striking and unexpected features of Odobenocetops are its tusks and the massive alveolar processes of the premax­ illae that support them—features paralleled in no other ceta- NUMBER 93 253 cean. These structures, in our opinion, embody the twofold im­ portance of Odobenocetops: Not only does the discovery of this new taxon dramatically extend the known morphological and ecological diversity of the Cetacea, but it also reopens the question of the function of tusks in tme walruses. After centuries of speculation and study, a consensus appears to have emerged that walruses use their tusks primarily in so­ cial (mostly agonistic) interactions, and secondarily for a vari­ ety of other purposes, but not to any significant extent in the course of their normal bottom-feeding (Fay, 1982; Nelson and Johnson, 1987; Kastelein and Mosterd, 1989; Kastelein et al., 1991). In other words, according to this view, walrus tusks are not feeding adaptations and do not form an integral part of the functional complex of feeding adaptations. All who have examined the skull of Odobenocetops have been immediately struck by its gross resemblance to that of a walrus, most obviously in regard to the tusks. Moreover, this study has confirmed that the feeding behavior of Odobenoce­ tops was walrus-like in that it consisted mainly of water-jetting and suction. This, however, raises a paradox: If walrus tusks are not feeding adaptations, then why should an animal whose feeding adaptations converge on those of a walrus be expected to have walrus-like tusks? Fay (1982:137-138) proposed a plausible explanation of how walruses evolved large tusks. He hypothesized that "posi­ tive selective pressures and the potential for tusk development probably have existed in all polygynous pinnipeds from the be­ ginning," but that the functional demands of pelagic piscivory, which required an unobstructed gape, precluded enlargement of the tusks beyond a certain point until odobenids took up feeding on benthic mollusks. Freed from the opposing selective pressure for a large gape, the animals were then able to respond to the preexisting social selection for hypertrophied tusks. This explanation, however, would probably not apply to Odobenocetops. Early odontocete cetaceans, unlike pinnipeds, did not have enlarged caniniform teeth with obvious "potential for tusk development." Moreover, although some odontocetes that have become specialized for suction feeding have devel­ oped tusk-like upper or lower teeth (narwhals, ziphiids), others have developed similar specializations without any enlarge­ ment of teeth (belugas); "selective pressures for rusk develop­ ment" thus seem not to have existed in all odontocetes. Fur­ thermore, if any of the suction-feeding specializations in belugas, narwhals, and Odobenocetops were inherited from a common ancestor, they obviously did not coincide in time of evolution with any enlargements of teeth, because tusks are ab­ sent in the first and nonhomologous in the other two. It ap­ pears, then, that any net social selection for tusk enlargement in odobenocetopsids did not result directly from, nor was it corre­ lated with, a shift from primitive piscivory to suction feeding. This also may have been true in walruses. Protodobenus japonicus, a new odobenine from the early Pliocene of Japan, seems to have been principally a piscivore with incipient adap­ tations for suction feeding, but its upper canine is not nearly so enlarged as in Odobenus (Horikawa, 1995). At least the begin­ nings of the evolution of suction feeding thus apparently pre­ ceded tusk enlargement. The tusk of Protodobenus is open- rooted, however, and apparently ever-growing, so it could be argued that this taxon also fulfills the prediction of Fay's hy­ pothesis that tusk enlargement should have followed immedi­ ately upon a shift away from piscivory. In Odobenocetops, tusk enlargement appears to have been correlated with the adoption of specifically Odobenus-\ike suc­ tion feeding, i.e., benthic suction feeding that involved continu­ ous direct contact of the snout with the substrate, possibly with major dependence on tactile vibrissae. If, as believed by stu­ dents of living walruses, this mode of feeding does not involve any use of, nor necessitate any enlargement of, the tusks, then the co-occurrence of benthic suction feeding and large tusks in both walruses and Odobenocetops is purely coincidental. If, on the other hand, this "explanation" is deemed unparsimonious, then we should look more closely for a functional connection between feeding and tusk enlargement in the modem walms. There is a wide range of possible functions for the tusks of Odobenocetops. Because the alveolar processes that support them are even more prominent than the tusks themselves, and differ from them in probably having been more nearly symmet­ rical, it is worthwhile to consider their possible functions sepa­ rately. Use of Tusks in Feeding: PRO: Tusks are teeth and primi­ tively serve for food-gathering. Even a single, asymmetrical tusk directed downward and backward could be used for stab­ bing prey, and walruses may kill seals on occasion (Fay, 1982:153). On the other hand, most, if not all, dental special­ izations for feeding are symmetrical, but the tusks are asym­ metrical. Asymmetry of feeding structures may occur in ceta­ ceans. Fin whales have asymmetrical color patterns around the mouth that are possibly associated with feeding (Gamble, 1985). Tursiops preferentially bottom-feed on the right side and have an asymmetrical larynx (with the piriform recess wider on the right) that may make it easier to swallow with that side down (Joy S. Reidenberg, pers. comm., 1995). CON: The tusks are asymmetrical; most, if not all, dental spe­ cializations for feeding are symmetrical. Asymmetry of feeding structures in cetaceans is not yet conclusively demonstrated and is not known to include the dentition. Tusks as Ballast: PRO: The relatively dense tusks, by add­ ing extra weight to the front of the head, may have helped keep the snout against the bottom during feeding. Walms tusks and their supporting bone also may serve this function. CON: If this was the primary function of the tusks, it is hard to explain why they were not equal in size. Also, it seems un­ likely that tusks as small as those of Odobenocetops would have had a significant ballast effect in an animal that could have approached the mass of a narwhal (between 800 and 1000 kg). The much larger tusk of a narwhal is not known to produce (either facultatively or obligately) a down-by-the-head attitude 254 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY in that animal, even though it has a much greater moment for doing so than would have been the case in Odobenocetops. Use of Tusks for Hauling Out: PRO: This is an important function of tusks in walruses. CON: Post-Eocene cetaceans do not haul out. Tusks used in hauling out would be most useful if they were symmetrical and stronger than observed in Odobenocetops. Use of Tusks in Piercing, Abrading, or Anchoring to Ice: PRO: Both walruses and narwhals use their tusks to make breathing holes, and walruses use them to anchor to ice (Fay, 1982:137). CON: Sea ice did not exist in Pern in the early Pliocene. Al­ though Odobenocetops may have lived in the Antarctic, it has not yet been found there. Social Role of Tusks: PRO: In many species tusks play so­ cial roles, such as in combat, in courting or mating, and in vi­ sual display. This is tme both in walruses and in narwhals, the sister group of Odobenocetops, in which they also are asym­ metrical. In walmses, they are used for ritualized dominance- threat displays by both males and females (Fay, 1982: 135-136). Use for combat would be consistent with the tusks' orientation, which is suitable for a slashing attack to an oppo­ nent's flank, using powerful muscles attached to the paroccipi­ tal processes. If the opponent were always approached head-on (right side to right side), this also would be consistent with the tusks' asymmetry. Use for social roles also would predict that the tusks would be larger in some individuals than in others. CON: The sex of the available specimen and the degree of in- trapopulational variation in the size of tusks are unknown. The tusks may have had a role in actual contact with other animals, but they seem rather slender to have been used for extremely forceful contact. Neither do their downward-and-backward di­ rection, unilateral enlargement, and slenderness seem opti­ mized for an impressive visual display, especially under water. Tusks as Primitive Retentions: PRO: The tusks may not have been adaptive in Odobenocetops itself, but merely re­ tained from an earlier evolutionary stage. CON: The weight, location, and external form of the tusks would probably have had deleterious hydrostatic and hydrody- namic effects on the animal's behavior and locomotion. Pre­ sumably, they would therefore have been eliminated quickly by selection in the absence of some positive selective value. Alveolar Processes as Support for the Tusks: PRO: Any use of the tusks would likely result in bending stresses, which sup­ porting sheaths would help resist. CON: The alveolar processes were probably relatively sym­ metrical, whereas the tusks are grossly asymmetrical. An unerupted tusk could not be used for anything and therefore would require no extra support. Walmses use their tusks very forcefully in a variety of ways and do not have such elongate bony sheaths to support them. Rather, they have been strength­ ened by being made thicker. In Odobenocetops they are rela­ tively slender, implying relatively little selection pressure for resistance to bending. Alveolar Processes as Hydrofoils or Diving Planes: PRO: The processes were prominent features of the head, at the very front end of the body where control surfaces would be most ef­ fective. Their oblique orientation resembles that of cetacean flippers. The bottom-feeding behavior of Odobenocetops might well have benefited from hydrofoils to keep the snout pressed against the substrate. CON: The processes may have been too small to have much effect as normal hydrofoils, especially if Odobenocetops was a relatively slow-swimming cetacean. Their angle of attack was not easily adjustable, so they would not have been as effective as the flippers. Other cetaceans manage without such control surfaces on the head. In cross section, the processes are thinner in front and thicker behind, hence not like that of an airfoil, so they would not have generated significant lift, even if they could have been held in a transverse rather than an oblique po­ sition. In bottom feeding, the alveolar processes would have lain against and parallel to the bottom (and hence in the same plane as the direction of advance) and would have had minimal effectiveness as hydrofoils. Moreover, their cross-sectional shape would, in that position, have tended to lift the head off the bottom rather than the contrary. If muscles or skin inserted on their posterior edges (see below), the processes plus at­ tached soft tissues might have had the size and shape necessary to function as hydrofoils, but the other objections would still apply. Alveolar Processes as Ballast: PRO: The extra mass of bone in the process, by adding weight to the front of the head, may have helped keep the snout against the bottom during feeding. Walrus tusks and their supporting bones also may serve this function. CON: The bone of the processes is not denser than the rest of the skull, unlike the bone in the rostrum of the walms. Further­ more, as is tme for the tusks, the mass of the alveolar processes was not great when compared with that of the animal. Use of Alveolar Processes for Sediment Displacement: PRO: The backward divergence of the processes would tend to pro­ duce a plowshare effect during feeding, shoving sediment to the sides and increasing the effective width of the path searched for prey. Cornified skin on the leading edge of the processes would be consistent with such use. CON: The dorsolaterad slope (in head-down position) of the lateral surface of the processes is unlike the ventrolaterad slope of a plowshare, and would press sediment downward rather than digging in and lifting it. Increasing by this means the area searched for prey would presuppose the presence of vibrissae along the whole length of each process, which is questionable (see below). Use of Alveolar Processes for Muscle Attachment: PRO: The processes would have provided highly effective lever arms for muscles (such as a modified platysma and/or cutaneus tmnci) that flexed the neck and/or turned the head, both impor­ tant actions in bottom feeding and in most of the conceivable uses of the tusks. The faint ridge on the posterior edge of the NUMBER 93 255 process might represent the insertion of a muscular aponeuro­ sis. CON: The large ventrolateral expansions of the occipital re­ gion would seem adequate for neck-muscle attachments, and such anterior muscle insertions are absent in walruses. Use of Alveolar Processes for Skin Attachment: PRO: The faint ridge on the posterior side of each alveolar process could even more plausibly mark the attachment of a sheet of skin ex­ tending backward from the process. Such a sheet might have served as a barrier to keep chumed-up mud and turbid water away from the eye during head-down bottom feeding. Al­ though Odobenus lacks such a sheet of skin (and a bony stmt to support it), it depends less upon vision during feeding, as sug­ gested by its more lateral placement of the eyes compared with Odobenocetops. Walmses do sometimes use their vision in for­ aging, but they do not always do so (Kastelein et al., 1993). CON: No contrary evidence is known. Alveolar Processes as Restricting Size of Mouth Opening for Suction Feeding: PRO: In walruses, known suction feeders, the short bony sheaths of the tusks form rigid sides for the mouth opening and help concentrate the suctional force (see Fay, 1982:171, fig. 106). CON: This also would be tme for the most proximal parts of the alveolar processes of Odobenocetops, but not the distal parts, and so would not explain the great elongation of the pro­ cesses. Both tusks of Odobenocetops seem large enough to have served this function even if supported only by short, wal­ rus-like processes. Alveolar Processes as Increasing Area and/or Breadth of Vibrissal Array: PRO: The symmetry and backward diver­ gence of the processes increase the total width of the snout, which could, therefore, support an increased number of vibris­ sae. Cornified skin on the leading edge of the process would be consistent with such use. CON: Large nutrient foramina are found only on the proxi­ mal, not the distal, portions of the alveolar processes (however, these foramina were not necessarily coextensive with the vibrissae). Moreover, it is not certain that vibrissae were in fact present. Alveolar Processes as Orientation Guides and Stabilizers for the Mystacium: PRO: More-or-less symmetrical tusk-like bony processes could serve as guides, somewhat in the manner of sled runners, to maintain the proper orientation of the myst­ acium and vibrissal array to the substrate while the animal searched for food and to reduce the exertion of neck muscles. In Odobenocetops, the symmetry and backward divergence of the processes would have given them more leverage in counter­ ing roll and yaw, as well as pitch. The right tusk, at least, would have enhanced this leverage (but the advantage of this must have been small, or else both tusks would have been of similar length). The cross-sectional shape of each process would have caused it to ride over sediment rather than digging in, likewise stabilizing the head against roll. Cornified skin on the leading edge of the process would be consistent with such use. Odobe­ nids with "walrus-like" morphotypes have analogous struc­ tures, i.e., long, symmetrical tusks with oval cross sections. CON: No contrary evidence is known. Use of Alveolar Processes in Combat: PRO: The large alve­ olar processes might have served a function in combat between males. This would be consistent with a social role for the tusks. The downward, backward, and lateral slope of the processes would serve, in a head-on collision, to guide an opponent's al­ veolar process and tusk ventrolaterad and away from the eye and flank. Cornified skin on the leading edge of the process would be consistent with such use. CON: The sex of the available specimen is unknown. Use of Alveolar Processes in Visual Display: PRO: This, too, would be consistent with a social role for the tusks. The processes are bulkier than the tusks themselves, and they would have accentuated and called attention to the presence and ori­ entation of the tusks. In head-on view, they also would have made the entire animal look larger. Use for display would pre­ dict that the processes would be larger in some individuals than in others. CON: The sex of the available specimen and the degree of in- trapopulational variation in the size of the processes are un­ known. Alveolar Processes as Primitive Retentions: PRO: Enlarged alveolar processes of the tusk-bearing bones are a necessity in animals with enlarged tusks. If tusks (however oriented) were present in the ancestors of Odobenocetops, enlarged alveolar processes also would have been present (as in narwhals); hence, enlargement of the processes may call for no special adaptive explanation. CON: For the reasons stated above, support for the tusks alone seems inadequate to account for the size of the processes. In any case, the enlarged processes of Odobenocetops relative to the rest of the skull are derived with respect to all other ceta­ ceans, and this enlargement calls for explanation. We conclude that the most plausible and important uses of the tusks of Odobenocetops were social ones. The alveolar pro­ cesses probably also played social roles, incidentally as a sup­ port for the empted tusk and perhaps more directly as weapons and/or shields in agonistic encounters, or in visual displays. Other functions, such as restricting the mouth opening for suc­ tion feeding or supporting an array of vibrissae, cannot be en­ tirely mled out and may even have been important in the early stages of evolutionary enlargement of the processes. Likewise, attached sheets of skin might have been useful in keeping sus­ pended sediment out of the field of vision in later stages of the processes' enlargement. At all stages of evolution, the in­ creased skeletal mass represented by both rusks and processes also would have had an incidental value in keeping the snout against the substrate during feeding. We suggest, however, that the single most important function of the alveolar processes, and the one that may have controlled their evolution, was that of orientation guides for the myst­ acium and vibrissal array. This is the only hypothesis that 256 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY seems to offer an explanation for the co-occurrence of benthic suction feeding and tusk-like structures in both Odobenocetops and walmses. Indeed, one of the most significant implications of the discovery of Odobenocetops is the fact that it reopens the question of whether the tusks of walmses are important in their feeding strategy. As noted above, the work of Fay (1982) and others seemed to have answered this question in the negative, and the consen­ sus today is that walms tusks serve primarily in social behav­ ior. In addition to their subsidiary functions, such as hauling out and opening holes in ice (Fay, 1982), perhaps walms tusks also serve as orientation guides for the mystacium and vibris­ sae. An analogy can be drawn with sled-mounted undersea cameras that are designed to be dragged along the seafloor; the sled runners stabilize the sled in roll, pitch, and yaw, keeping the lights and cameras pointed in the right direction. Similarly, tusk-like structures would stabilize an animal's head and help maintain the mouth and vibrissal array in a fixed attitude rela­ tive to the bottom, increasing the efficiency of search by the sense organs and, possibly, the efficiency and accuracy of wa­ ter-jetting and/or suction feeding. If such orientation guides are important in the odobenine style of bottom feeding (and it should be possible to test this experimentally in living walmses), then we would expect both sexes to have them (as is the case with the tusks of Odobenus). The external form of such guides rather than their internal structure would be functionally important, however, so they would not necessarily have to be genuine tusks; for example, tusk-like bony stmctures would serve just as well, provided they were reasonably symmetrical and sufficiently long. Such stmctures would make real tusks redundant, and vice versa, so we would not expect to find both long, symmetrical tusks and bony equivalents of them in the same species. This is bome out by fossil odobenines such as Valenictus (Demere, 1994), which resemble modern Odobenus in this regard—although this would of course be expected from their close relationship alone. This hypothesis also is consistent with the ontogeny of wal­ rus tusks. Weaning usually occurs in Odobenus at 14-27 months, and benthic feeding begins at 6-24 months (Fay, 1982:138-141). Tusk emption also begins in the first year of life; the tusk begins to show signs of discoloration and wear by the age of two years, and at this time the tip of the crown ex­ tends below the ventral side of the mandible and usually 2-4 cm below the edge of the upper lip (Fay, 1982:105-107, fig. 71). It therefore seems possible that the tusks could be starting to function as guides even at this age. What we find in Odobenocetops also is consistent with this interpretation. Although tusks are present, they are not sym­ metrical; indeed, only one of the pair may have erupted. In­ stead we find elongated, and probably much more symmetrical, alveolar processes of the premaxillae, which might very well have served as orientation guides for the snout. Their posterior divergence would have enhanced their effectiveness in this role. In fact, it is difficult to explain them merely as sheaths for the tusks, because walmses manage to wield much larger tusks very forcefully without the support of such elongated sheaths, and relatively symmetrical sheaths should not be needed for the highly asymmetrical (and, in the case of the unerupted tusk, functionless) teeth of Odobenocetops. We therefore posit a functional role for the alveolar processes independent of the tusks themselves. This still leaves unexplained the evolution of tusks in Odo­ benocetops. Given their asymmetry and the unlikelihood of functions such as hauling-out or ice-breaking in the case of a cetacean living in Pem, it is probably safe to make an analogy with narwhal tusks and attribute to them a primarily social function. It would not be surprising if sister groups such as monodontids and odobenocetopsids evolved such analogous stmctures in parallel, albeit developing the tusks from canines and incisors, respectively. Conclusions It appears, then, that the resemblance between Odobenoce­ tops and walmses is not merely superficial, but a genuine case of functional convergence in feeding adaptations—specifically, adaptations for suction feeding on infaunal benthic prey. Fur­ thermore, from the fact that stmctures resembling walms tusks evolved in an animal clearly convergent on walmses in other characters that are unquestionably feeding adaptations, we can surmise that the tusk-like structures are probably associated with feeding in both animals. Otherwise, the correlation of benthic suction feeding with tusk-like structures in both ani­ mals is merely a striking coincidence—a conclusion we regard as unparsimonious. Apart from the tusks and alveolar processes, the features in which Odobenocetops resembles a walrus are its hourglass- shaped skull, vaulted palate, enlarged paroccipital processes, possible vibrissae and sensitive upper lip, and possible homy covering of part of the upper lip. These features form a charac­ ter complex apparently unique to odobenids and odobenoce­ topsids, and in the former are functionally related to a style of feeding documented in no other animals: powerful suction feeding on infaunal benthic prey. This character complex is wholly or partly absent in animals that prey on benthic inverte­ brates but are not suction feeders (e.g., sea otters), as well as in ones that are suction feeders but prey on animals that are free- swimming or merely rest upon the bottom (e.g., many odonto­ cetes; A. Werth, pers. comm., 1992). It is thus only in associa­ tion with this "benthic suction feeding" character complex that stmctures resembling walrus tusks are found—specifically in derived odobenines and in Odobenocetops. If, instead, we attempt to explain the tusk-like stmctures in each case as having evolved mainly for social functions, we face an obvious problem. There is no apparent reason why fights or displays using such structures should occur always and only in benthic suction feeders. Moreover, it would be very NUMBER 93 257 surprising if both an odontocete and a pinniped were to evolve closely similar social displays, rituals, or modes of fighting that were used in and out of water, respectively, by animals of such very different body form. If Fay (1982:137) is correct in attributing a social selective pressure for tusk enlargement to all polygynous pinnipeds (and this may well be tme), then we might expect to find conver­ gences on the walms morphotype among other pinnipeds if we found them anywhere—yet this is not the case. We think Fay is correct in arguing that benthic suction feeding was the key adaptive shift that led to the enlargement of tusks in walmses. In view of the similar morphology seen in Odobenocetops, however, and the difficulty of applying the same social expla­ nation to this otherwise very different animal, we propose that the utility of the tusk-like stmctures in both cases was primarily in feeding (as orientation guides for the mystacium) and only secondarily in social interactions. In other words, we consider the tusk-like structures to be an integral part of the "benthic suction-feeding" character complex, because we see no other plausible explanation for this striking correlation of characters. Students of modem Odobenus should therefore reexamine its feeding behavior with the possibility in mind that the tusks may serve as orientation guides, as suggested herein. This hypothe­ sis should be amenable to experimental tests. Addendum This paper was already in the editorial process when three new skulls of Odobenocetops were collected from the Pisco Formation, which justified the publication of a preliminary note (Muizon et al., 1999). One of the skulls (SMNK 2491), referred to O. peruvianus, is from the Sud-Sacaco locality and from the SAS horizon (as was the holotype of O peruvianus), which is earliest Pliocene in age. It is almost symmetrical and bears two small tusks; al­ though the right is slightly larger than the left, it is still drasti­ cally smaller than the large tusk of the holotype. The alveolar sheaths are relatively small and of the same size. The two tusks are incomplete and their apices are missing. The right sheath is partly damaged at its apex, but because the preserved portion of the tusk is distinctly longer than the sheath as preserved, it is clear that the tusk was empted. The left sheath and tusk also are incomplete, and the tusk is broken in the alveolus. It is there­ fore not possible to determine whether the left tusk was empted. In the narwhal the unempted tusks of the female are generally similar in size and definitely smaller than the large left tusk of the male, so this new skull of Odobenocetops peru­ vianus (SMNK 2491) was identified as a female. The other two skulls (SMNK 2492 and MNHN SAO 202) come from the locality of Sacaco in the SAO horizon, which is slightly younger than the SAS horizon, in which the O. peruvi­ anus specimens were found. They were referred by Muizon et al. (1999) to a different and new species, O leptodon. The ho­ lotype (SMNK 2492) is a partially damaged skull that retains both tusks in situ. The right tusk is more than 1.35 m long as preserved (the apex was broken and worn during life), and its erupted portion measures 1.07 m. The left tusk is small and slender and was approximately 25 cm long. Its apex bears a distinct wear facet, which indicates that the tooth was empted. The presence of a very long right tusk and of an erupted left tusk in O. leptodon is an indication (not a proof) that this con­ dition also could be present in O. peruvianus. This has to be confirmed by the discovery of new specimens of O. peruvi­ anus, however. The morphology of the snout of O. leptodon differs from that of O. peruvianus in being more rounded and wider in dorsal view and in lacking large premaxillary foramina. Furthermore, the dorsal face of the premaxilla bears a fossa for a premaxil­ lary sac, and a pair of supplementary rostral bones is present at the anterodorsal apex of the snout. The anterodorsal edge of the orbit is only slightly concave in O. leptodon, whereas it is deeply notched in O. peruvianus. The presence of premaxillary sacs (which were probably absent in O peruvianus) is probably an indication of the presence of a melon in O. leptodon. This organ, related to echolocation, was probably absent or vestigial in O. peruvianus. As stated above in the text, the inferred ab­ sence of a melon in O. peruvianus was probably compensated for by good anterodorsal binocular vision. Muizon et al. (1999) concluded that binocular vision was either reduced or absent in O leptodon and that this was compensated for by echolocation abilities inferred from the probable presence of a melon. The three new skulls have at least one tympanic and one pe­ riotic in situ. The characteristic morphology of these bones (es­ pecially the massiveness of the anterior process of the periotic) definitely confirms the referral of the periotics and tympanic described above to Odobenocetops peruvianus. The holotype skull of Odobenocetops leptodon was associ­ ated with its atlas. The only occipital condyle preserved on the skull is damaged, which makes it difficult (but not impossible) to evaluate the position of the head relative to the axis of the body. An extrapolation could be done using the new skull of O. peruvianus (SMNK 2491), which has well-preserved condyles. The position of the atlanto-occipital articulation of Odobenoce­ tops indicates that in swimming position the head was bent ventrally, bringing the dorsal plane of the skull into an antero­ dorsal orientation. With the head in this position, the long tusk was almost parallel to the long axis of the body. This interpreta­ tion is compatible with the length of the large tusk, as it is un­ likely that such a long appendage could be at an angle of about 45° to the body during swimming (Muizon et al., 1999, fig. 2). The head of Odobenocetops peruvianus in Figure 20 therefore 258 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY should probably be bent slightly ventrally to make the long tusk parallel to the axis of the body. To conclude, the new specimens of Odobenocetops indicate the following: 1. The long right tusk of O. peruvianus was probably longer than initially thought, although this needs to be confirmed by the discovery of new specimens of this species. 2. The small left tusk was probably empted, but this also has to be confirmed by the discovery of new specimens. 3. The female of O. peruvianus had two small tusks, the right being only slightly larger than the left. This condition is likely to have also occurred in O. leptodon, although this has to be confirmed by additional specimens. 4. The new specimens confirm referral to O. peruvianus of the periotics and tympanic described above. 5. During swimming, the head of both species was bent ven­ trally and, in such a position, the long tusk was almost parallel to the axis of the body. Odobenocetops peruvianus (which has a deeply notched anterodorsal edge of the orbit) therefore had good dorsal (anterior) binocular vision when swimming. Literature Cited Barnes, L.G. 1984. Fossil Odontocetes (Mammalia, Cetacea) from the Almejas Forma­ tion, Isla Cedros, Mexico. PaleoBios, 42:1—46. Best, R.C., and V.M.F. da Silva 1989. Amazon River Dolphin, Boto: Inia geoffrensis (de Blainville, 1817). In S.H. Ridgway and R.J. Harrison, editors, Handbook of Marine Mammals, 4: River Dolphins and Larger Toothed Whales, pages 1-23. London: Academic Press. Boenninghaus, Georg 1904. Das Ohr des Zahnwales, zugleich ein Beitrag zur Theorie der Schal- leitung: Eine biologische Studie. Zoologische Jahrbucher, Abteilung fur Anatomie und Ontogenie der Tiere, 19(2): 189-360. Bourdelle, E., and P.-P. Grasse 1955. Ordre des Cetaces (Cetacea Brisson 1762; Cete Linne 1758). In P.-P. Grasse, editor, Trade de Zoologie, Anatomie, Systematique, Biolo­ gie, 17(l):341-450. Paris: Masson. Brill, R.L., M.L. Sevenich, T.J. Sullivan, J.D. Sustman, and R.E. Witt 1988. Behavioral Evidence for Hearing through the Lower Jaw by an Echolocating Dolphin (Tursiops truncatus). Marine Mammal Sci­ ence, 4(3):223-230. Brodie, P.F. 1989. The White Whale: Delphinapterus leucas (Pallas, 1776). In S.H. Ridgway and R.J. Harrison, editors, Handbook of Marine Mammals, 4: River Dolphins and Larger Toothed Whales, pages 119-144. Lon­ don: Academic Press. Brownell, R.L. 1989. Franciscana: Pontoporia blainvillei (Gervais et d'Orbigny, 1844). In S.H. Ridgway and R.J. Harrison, editors, Handbook of Marine Mammals, 4: River Dolphins and Larger Toothed Whales, pages 45-67. London: Academic Press. Carriol, R.R, C. de Muizon, and S. Secretan 1986. Les crustaces (Cirripedia, Decapoda) du Neogene de la formation Pisco (Perou). Annates de Paleontologie, 73(3): 137-164. Cheneval, J. 1993. L'avifaune Mio-Pliocene de la formation Pisco (Perou): Etude preliminaire. Documents des Laboratoires de Geologie de la Fac- ulte des Sciences de Lyon, 125:85-95. Colbert, E.H. 1944. A New Fossil Whale from the Miocene of Peru. Bulletin of the American Museum of Natural History, 83(3): 195-216. Crawford, T.W. 1981. Vertebrate Prey of Phocoenoides dalli (Dall's Porpoise) Associated with the Japanese High Sea Salmon Fishery in the Northern Pacific Ocean. Master's thesis, University of Washington, College of Fish­ eries, Seattle. Dal Piaz, G. 1916. Gli odontoceti del Miocene Bellunense; Squalodon. Memorie dell 'Is- tituti Geologico della (R) Universita di Padova, part 2, 4: 1-79. Demere, T.A. 1994. Two New Species of Fossil Walruses (Pinnipedia: Odobenidae) from the Upper Pliocene San Diego Formation, California. Proceed­ ings of the San Diego Society of Natural History, 29:77-98. Diercks, K.J., R.T. Trochta, CF. Greenlaw, and W.E. Evans 1971. Recording and Analysis of Dolphin Echolocation Signals. Journal of the Acoustical Society of America, 49:1732-1732. Domning, D.P. 1989. Fossil Sirenia of the West Atlantic and Caribbean Region, III: Xeno­ siren yucateca, gen. et sp. nov. Journal of Vertebrate Paleontology, 9(4):429-437. Domning, D.P., C E . Ray, and M.C. McKenna 1986. Two New Oligocene Desmostylians and a Discussion of Tethythe­ rian Systematics. Smithsonian Contributions to Paleobiology, 59: 1-56. Dormer, K.J. 1979. Mechanism of Sound Production and Air Recycling in Delphinids: Cineradiographic Evidence. Journal of the Acoustical Society of America, 65:229-239. Evans, W.E. 1967. Special Acoustic Problems in Cetacean Research, General Discus­ sion. In W.N. Tavolga, editor, Marine Bio-Acoustics, 2:395-398. New York: Pergamon Press. Evans, W.E., and P.F.A. Maderson 1973. Mechanisms of Sound Production in Delphinid Cetaceans: A Re­ view of Some Anatomical Considerations. American Zoologist, 13:1205-1213. Evans, W.E., and J.H. Prescott 1962. Observation of the Sound Capabilities of the Bottlenose Porpoise: A Study of the Whistles and Clicks. Zoologica, 47(3):121-126. Fay, F.H. 1982. Ecology and Biology of the Pacific Walrus, Odobenus rosmarus di- vergens Illiger. North American Fauna, 74:1-279. Washington, D.C: U.S. Department of Interior, Fish and Wildlife Service. Fordyce, R.E. 1994. Waipatia maerewhenua, New Genus and New Species (Waipatiidae, New Family), an Archaic Late Oligocene Dolphin (Cetacea: Odon­ toceti: Platanistoidea) from New Zealand. Proceedings of the San Diego Society of Natural History, 29:147-176. Fraser, F.C, and P.E. Purves 1960. Hearing in Cetaceans: Evolution of the Accessory Air Sacs and the Structure and Function of the Outer and Middle Ear in Recent Ceta- NUMBER 93 259 ceans. Bulletin of the British Museum (Natural History), Zoology, 7(1):1-140. Fried, W.A., P.E. Nachtigall, and P.W.B. Moore 1990. Taste Reception in the Pacific Bottlenose Dolphin (Tursiops trunca­ tus gilli) and the California Sea Lion (Zalophus californianus). In J.A. Thomas and R.A. Kastelein, editors, Sensory Abilities of Ceta­ ceans: Laboratory and Field Evidence, pages 447-^454. New York: Plenum Press. Gamble, R. 1985. Fin Whale Balaenoptera physalus (Linnaeus, 1758). In S.H. Ridg­ way and R.J. Harrison, editors, Handbook of Marine Mammals, 3: Sirenians and Baleen Whales, pages 171-192. New York: Academic Press. Gao, G., and K. Zhou 1995. Fiber Analysis of Vestibular Nerves in Small Cetaceans. In R.A. Kastelein, J.A. Thomas, and P.E. Nachtigall, editors, Sensory Sys­ tems of Aquatic Mammals, pages 447-453. Woerden, The Nether­ lands: De Spil Publishers. Gordon, K.R. 1984. Models of Tongue Movement in the Walrus (Odobenus rosmarus). Journal of Morphology, 182:179-196. Gray, 0 . 1951. An Introduction to the Study of the Comparative Anatomy of the Labyrinth. Journal of Laryngology and Otology, 65:681-703. Hay, K.A. 1980. Age Determination of the Narwhal, Monodon monoceros, L. In W.F. Perrin and C A . Myrick, editors, Age Determination of Toothed Whales and Sirenians. Reports of the International Whaling Com­ mission, Special Issue, 3:119-132. Hay, K.A., and A.W. Mansfield 1989. Narwhal Monodon monoceros Linnaeus, 1758. In S.H. Ridgway and R.J. Harrison, editors, Handbook of Marine Mammals, 4: River Dol­ phins and Larger Toothed Whales, pages 145-176. London: Aca­ demic Press. Heyning, J.E., and J.G. Mead 1996. Suction Feeding in Beaked Whales: Morphological and Observa­ tional Evidence. Contributions in Science, Natural History Museum of Los Angeles County, 464:1-12. Horikawa, H. 1995. A Primitive Odobenine Walrus of Early Pliocene Age from Japan. The Island Arc, 3:309-328. Howell, A.B. 1927. Contribution to the Anatomy of the Chinese Finless Porpoise Neom- eris phocaenoides. Proceedings of the United States National Mu­ seum, 70(13): 1-43. 1930. Myology of the Narwhal (Monodon monoceros). American Journal of Anatomy, 46(2): 187-215. Jansen, J., and J.K.S. Jansen 1969. The Nervous System of Cetacea. In H.T. Anderson, editor, The Biol­ ogy of Marine Mammals, pages 175-252. New York: Academic Press. Kastelein, R.A., N.M. Gerrits, and J.L. Dubbeldam 1991. The Anatomy of the Walrus Head (Odobenus rosmarus), Part 2: De­ scription of the Muscles and of Their Role in Feeding and Haul-Out Behaviour. Aquatic Mammals, 17(3): 156-180. Kastelein, R.A., and P. Mosterd 1989. The Excavation Technique for Molluscs of Pacific Walruses (Odo­ benus rosmarus divergens) under Controlled Conditions. Aquatic Mammals, 15(l):3-5. Kastelein, R.A., and P.R. Wiepkema 1989. A Digging Trough as Occupational Therapy for Pacific Walruses (Odobenus rosmarus divergens) in Human Care. Aquatic Mammals, 15(1):9-17. Kastelein, R.A., R.C.V.J. Zweypfenning, H. Spekreijse, J.L. Dubbeldam, and E.W. Born 1993. The Anatomy of the Walrus Head (Odobenus rosmarus), Part 3: The Eyes and Their Function in Walrus Ecology. Aquatic Mammals, 19(2):61-92. Kellogg, R. 1923. Kentriodon pern'ix, a Miocene Porpoise from Maryland. Proceed­ ings of the United States National Museum, 69( 19): 1-55. 1926. Supplementary Observations on the Skull of Zarhachis flagellator Cope. Proceedings of the United Stales National Museum, 67(28): 1-18. Ketten, D.R. 1984. Correlations of Morphology with Frequency for Odontocete Co­ chlea: Systematics and Topology. 335 pages. Doctoral dissertation, Johns Hopkins University, Baltimore, Maryland. 1992. The Marine Mammal Ear: Specializations for Aquatic Audition and Echolocation. In D. Webster, R. Fay, and A. Popper, editors, The Evolutionary Biology of Hearing, pages 717-754. New York: Springer-Verlag. Ketten, D.R., and D. Wartzok 1990. Three-dimensional Reconstructions of the Dolphin Ear. In J.A. Tho­ mas and R.A. Kastelein, editors, Sensory Abilities of Cetaceans: Laboratory and Field Evidence, pages 81-105. New York: Plenum Press. Kleinenberg, S.E. 1938. [Some Data on the Feeding of Tursiops tursio Fabr. in the Black Sea.] Byulleten Moskovskogo Obshchestva Ispytatelei Prirody, Ot- del Biologicheskii [Bulletin of the Moscow Society of Researchers on Nature, Section of Biology], 47:406-^413. [In Russian.] Kleinenberg, S.E., A.V. Yablokov, B.M. Bel'kovich, and M.N. Tarasevich 1969. [Beluga (Delphinapterus leucas): Investigation on the Species.} 376 pages. Academy of Science of the USSR, A.N. Severtsov Institute of Animal Morphology. [In Russian. English translation by Israel Program for Scientific Translations, Jerusalem.] Kuznetzov, VB. 1990. Chemical Sense of Dolphins: Quasi-Olfaction. In J.A. Thomas and R.A. Kastelein, editors, Sensory Abilities of Cetaceans: Laboratory and Field Evidence, pages 481-503. New York: Plenum Press. Lawrence, B., and W.E. Schevill 1956. The Functional Anatomy of the Delphinid Nose. Bulletin of the Mu­ seum of Comparative Zoology, Harvard College, 114(4): 103-151. Lilly, J.C. 1961. Man and Dolphin. 312 pages. Garden City, New Jersey: Doubleday &Co. Lilly, J .C, and A.M. Miller 1961. Sounds Emitted by the Bottlenose Dolphin. Science. 5:1689-1693. Lisson, CI . 1898. Los fosfatos de Ocucaje. Boletin de Minas, Industria y Construc- ciones, 14(5):45^47. Luo, Z., and K. Marsh 1996. Petrosal (Periotic) and Inner Ear of a Pliocene Kogiine Whale (Kogiinae, Odontoceti): Implications on Relationships and Hearing Evolution of Toothed Whales. Journal of Vertebrate Paleontology, 16(2):328-348. Marocco, R., and C de Muizon 1988. Los vertebrados del Neogeno de la costa Sur del Peru: ambiente y condiciones de fosilizacion. Bulletin de I'Institut Francais d'Etudes Andines, 27(2): 105-117. Marsh, H. 1980. Age Determination of the Dugong (Dugong dugon (Miiller)) in Northern Australia and Its Biological Implications. Reports of the International Whaling Commission, Special Issue, 3:181-201. 260 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY Martin, A.R. 1990. The Illustrated Encyclopedia of Whales and Dolphins. 192 pages. New York: Portland House. Matthews, L.H. 1978. The Natural History of the Whale. 219 pages. New York: Columbia University Press. McCormick, J.G., E.G., Wever, and J. Palin 1970. Sound Conduction in the Dolphin Ear. Journal of the Acoustical So­ ciety of America, 48(6): 1418-1428. Mchedlidze, G.A., and G. Pilleri 1988. Kharthlidelphis diceros, a New Genus and Species from the Oli­ gocene of Georgia (Caucasus, USSR). Investigations on Cetacea, 21:9-15. Mead, J.G. 1975. Anatomy of the External Nasal Passages and Facial Complex in the Delphinidae (Mammalia, Cetacea). Smithsonian Contributions to Zoology, 207:1-72. Miller, M.E., G.C. Christensen, and H.E. Evans 1964. Anatomy of the Dog. 941 pages. Philadelphia: W.B. Saunders. Muizon, C. de 1981. Les vertebres fossiles de la formation Pisco (Perou), premiere partie: Deux nouveaux Monachinae (Phocidae, Mammalia) du Pliocene de Sud-Sacaco. Travaux de I'lnstitut Francais d'Etudes Andines, 22: 160 pages. [Recherches sur les Grandes Civilisations, Memoire 6. Paris: A.D.P.F.] 1983a. Pliopontos littoralis, un nouveau Platanistidae (Cetacea) du Plio­ cene de la cote peruvienne. Comptes Rendus Hebdomadaires des Seances de I'Academie des Sciences (Paris) , series 2, 296: 1101-1104. 1983b. Un nouveau Phocoenidae (Cetacea) du Pliocene inferieur du Perou. Comptes Rendus Hebdomadaires des Seances de I'Academie des Sciences (Paris), series 2, 296:1203-1206. 1983c. Un Ziphiidae (Cetacea) nouveau du Pliocene inferieur du Perou. Comptes Rendus Hebdomadaires des Seances de I 'Academie des Sciences (Paris), series 2, 297:85-88. 1984. Les vertebres fossiles de la formation Pisco (Perou), deuxieme par- tie: Les odontocetes (Cetacea, Mammalia) du Pliocene inferieur de Sud-Sacaco. Travaux de I'lnstitut Francais d'Etudes Andines, 25: 188 pages. [Recherches sur les Grandes Civilisations, Memoire 50. Paris: A.D.P.F] 1986. Un nouveau Phocoenidae (Odontoceti, Mammalia) du Miocene superieur de la formation Pisco (Perou). Comptes Rendus Hebdoma­ daires des Seances de I'Academie des Sciences (Paris), series 2, 303(16): 1509-1512. 1987. The Affinities of Notocetus vanbenedeni, an Early Miocene Pla- tanistoid (Cetacea, Mammalia) from Patagonia, Southern Argentina. American Museum Novitates, 2904: 27 pages. 1988a. Les vertebres fossiles de la formation Pisco (Perou), troisieme par- tie: Les odontocetes (Cetacea, Mammalia) Miocenes. Travaux de I'lnstitut Francais d'Eludes Andines, 42: 244 pages. [Recherches sur les Grandes Civilisations, Memoire 78. Paris: A.D.P.F] 1988b. Le polyphyletisme des Acrodelphidae, odontocetes longirostres du Miocene europeen. Bulletin du Museum National d'Histoire Na­ turelle, series 4, section C, 10(l):31-88. 1988c. Les relations phylogenetiques des Delphinida. Annates de Paleon­ tologie, 74(4): 159-227. 1991. A New Ziphiid (Odontoceti, Mammalia) from the Early Miocene of Washington State (USA) and a Phylogenetical Analysis of the Ma­ jor Groups of Odontocetes. Bulletin du Museum National d'Histoire Naturelle, series 4, section c, 12:279-326. 1993a. Walrus-Like Feeding Adaptation in a New Cetacean from the Pliocene of Peru. Nature, 365:745-748. 1993b. Odobenocetops peruvianus. una remarcable convergencia de adapt- acion alimentaria entre morsa y delfin. Bulletin de I'lnstitut Francais d'Etudes Andines, 22(3):671—683. 1994. Are the Squalodonts Related to the Platanistoids? Proceedings of the San Diego Society of Natural History, 29:135-146. Muizon, C. de, and H. Bellon 1986. Nouvelles donnees sur l'age de la formation Pisco (Perou). Comptes Rendus Hebdomadaires des Seances de I 'Academie des Sciences (Paris), series 2, 303(5): 1401-1404. Muizon, C. de, and T.J. DeVries 1985. Geology and Paleontology of Late Cenozoic Marine Deposits in the Sacaco Area (Peru). Geologisches Rundschau, 74(3):547-563. Muizon, C. de, and D.P. Domning 2002. The Anatomy of Odobenocetops (Delphinoidea, Mammalia), the Walrus-like Dolphin from the Pliocene of Peru and Its Paleobiologi- cal Implications. Zoological Journal of the Linnean Society, 134(4):423-452. Muizon, C de, D.P. Domning, and M. Parrish 1999. Dimorphic Tusks and Adaptive Strategies in a New Species of Wal­ rus-like Dolphin (Odobenocetopsidae) from the Pliocene of Peru. Comptes Rendus de I 'Academie des Sciences, series 2 (Sciences de la Terre et des Planetes), 329:449-455. Murie, J. 1870. On Risso's Grampus, G. rissoanus (Desm.). Journal of Anatomy and Physiology, 5:118-138. 1873. On the Organization of the Caaing Whale, Globicephala melas. Transactions of the Zoological Society of London, 8(4):235—301. Nelson, C.H., and K.R. Johnson 1987. Whales and Walruses as Tillers of the Sea Floor. Scientific Ameri­ can, 256(2): 112-117. Nishiwaki, M. 1972. Genera] Biology. In S.H. Ridgway, editor, Mammals of the Sea: Biol­ ogy and Medicine, pages 1-204. Springfield, Illinois: C C Thomas. Norris, K.S. 1964. Some Problems in Echolocation in Cetaceans. In W.N. Tavolga, edi­ tor, Marine Bio-Acoustics. 1:317-355. New York: Pergamon Press. 1968. The Evolution of Acoustic Mechanisms in Cetaceans. In E.T. Drake, editor, Evolution and Environment, pages 297-324. New Haven: Yale University Press. 1969. The Echolocation of Marine Mammals. In H.T. Andersen, editor, The Biology of Marine Mammals, pages 391^423. New York: Aca­ demic Press. Norris, K.S., K.J. Dormer, J. Pegg, and G.J. Liese 1971. The Mechanism of Sound Production and Air Recycling in Por­ poises: A Preliminary Report. Proceedings of the Eighth Annual Conference on Biological Sonar and Diving Mammals, pages 113-129. Menlo Park, California: Stanford Research Institute. Norris, K.S., and W.E. Evans 1967. Directionality of Echolocation Clicks in the Rough-tooth Porpoise, Steno bredanensis (Lesson). In W.N. Tavolga, editor, Marine Bio- Acoustics, 2:305-316. New York: Pergamon Press. Norris, K.S., and G.W. Harvey 1974. Sound Transmission in the Porpoise Head. Journal of the Acoustical Society of America. 56(2):659-664. Oelschlager, H.A., and E.H. Buhl 1985a. Development and Rudimentation of the Peripheral Olfactory Sys­ tem in the Harbor Porpoise Phocoena phocoena (Mammalia: Ceta­ cea). Journal of Morphology, 184:351-360. 1985b. Occurrence of an Olfactory Bulb in the Early Development of the Harbor Porpoise (Phocoena phocoena L.). Fortschritte der Zoolo­ gie, 30:695-698. Pabst, D.A. 1990. Axial Muscles and Connective Tissues of the Bottlenose Dolphin. In S. Leatherwood and R.R. Reeves, editors, The Bottlenose Dolphin, pages 51-67. San Diego: Academic Press. Pilleri, G., editor 1989. Beitrage sur Paldontologie der Cetaceen Penis. Volume 1, 233 pages. Ostermundingen [Bern]: Hirnanatomisches Institut. NUMBER 93 261 1990. Beitrage sur Paldontologie der Cetaceen Perus. Volume 2, 248 pages. Ostermundingen [Bern]: Himanatomisches Institut. Pilleri, G., and M. Ghir 1981. The Brain (Endocranial Cast) of Schizodelphis sulcatus and the Cephalization of Eoplatanista italica (Cetacea): Paleoneurological and Paleoecological Considerations. Memorie degli Istituti di Geo­ logia e Mineralogia dell'Universitd di Padova, 34:387-440. 1982. The Brain and Cephalization of Schizodelphis sulcatus Gervais, 1861. Investigations on Cetacea, 13:13-25. Preen, A.R. 1989. Observations of Mating Behavior in Dugongs (Dugong dugon). Ma­ rine Mammal Science, 5:382-387. Purves, P.E. 1967. Anatomical and Experimental Observations on the Cetacean Sonar System. In R.-G. Busnel, editor, Animal Sonar Systems, Biology and Bionics, 1:197-270. Jouy-en-Josas: Laboratoire de Physiologie acoustique, INRA-CNRZ. Rae, B.B. 1973. Additional Notes on the Food of the Common Porpoise (Phocoena phocoena). Journal of Zoology, 169:127-131. Ray, C 1966. Round Table: Practical Problems. In K.S. Norris, editor, Whales, Dolphins and Porpoises, page 671. Berkeley: University of Califor­ nia Press. Ray, C.E., D.P. Domning, and M.C. McKenna 1994. A New Specimen of Behemotops proteus (Mammalia: Desmostylia) from the Marine Oligocene of Washington. Proceedings of the San Diego Society of Natural History, 29:205-222. Ridgway, S.H., D.A. Carder, R.F. Green, A.S. Gaunt, L.L. Gaunt, and W.E. Evans 1980. Electromyographic and Pressure Events in the Nasolaryngeal Sys­ tem of Dolphins during Sound Production. In R.-G. Busnel and J.F. Fish, editors, Animal Sonar Systems [NATO Advanced Science In­ stitute, series A (Life Science), Volume 28], pages 239-249. New York: Plenum Press. Rodinov, V.A., and V.I. Markov 1992. Functional Anatomy of the Nasal System in the Bottlenose Dolphin. In J.A. Thomas, R.A. Kastelein, and A.Ya. Supin, editors, Marine Mammal Sensory Systems, pages 147-177. New York: Plenum Press. Schenkkan, E.J. 1973. On the Comparative Anatomy and Function of the Nasal Tracts in Odontocetes (Mammalia, Cetacea). Bijdragen tot de Dierkunde, 43(2): 127-159. Schevill, W.E. 1964. General Discussion. In W.N. Tavolga, editor, Marine Bio-Acoustics, 1:395-398. New York: Pergamon Press. Schulte, H. von W., and M. de Forest Smith 1918. The External Characters, Skeletal Muscles, and Peripheral Nerves of Kogia breviceps (Blainville). Bulletin of the American Museum of Natural History, 38(2):7-72. Seagars, D.J. 1982. Jaw Structure and Functional Mechanics of Six Delphinids (Ceta­ cea: Odontoceti). 179 pages. Master's thesis, San Diego State Uni­ versity, San Diego. Seaman, G.A., L.F. Lowry, and K.J. Frost 1982. Foods of the Belukha Whales (Delphinapterus leucas) in Western Alaska. Cetology, 44(Sept): 1-19. Sickenberg, O. 1934. Beitrage zur Kenntnis Tertiarer Sirenen. Memoires du Musee Royal d'Histoire Naturelle de Belgique, 63:1-352. Sinclair, J.G. 1966. The Olfactory Complex of Dolphin Embryos. Texas Reports of Bio­ logical Medicine, 24:426-431. Spoor, F. 1993. The Comparative Morphology and Phylogeny of the Human Bony Labyrinth. Doctoral dissertation, Universiteit Utrecht, Utrecht. Tedford, R.H., L.G. Barnes, and C E . Ray 1994. The Early Miocene Littoral Ursid Carnivoran, Kolponomos: Sys­ tematics and Mode of Life. Proceedings of the San Diego Society of Natural History, 29:11-32. Tomilin, A.G. 1967. Cetacea. In Mammals of the USSR and Adjacent Countries, 9: 717 pages. [Original in Russian, published 1957. Translated by Israel Program for Scientific Translations, Jerusalem.] Vibe, C. 1950. The Marine Mammals and the Marine Fauna in the Thule District (North-West Greenland) with Observations on Ice Conditions in 1939^11. Meddelelser om Gronland. 150(6): 1—115. Werth, A.J. 1992. Anatomy and Evolution of Odontocete Suction Feeding, xiii+313 pages. Doctoral dissertation, Harvard University, Cambridge, Mas­ sachusetts. Wilke, F , T. Taniwaki, and N. Kuroda 1953. Phocoenoides and Lagenorhynchus in Japan, with Notes on Hunt­ ing. Journal of Mammalogy, 34(4):488-497. Wood, F.G. 1964. General Discussion. In W.N. Tavolga, editor, Marine Bio-Acoustics, 1:395-398. New York: Pergamon Press. Yablokov, A.V, and G.A. Klevezal 1962. [Whiskers of Whales and Seals and Their Distribution, Structure and Significance.] In S.E. Kleinenberg, editor, Morfologicheskie Os- obennosti Vodnikh Mlekopitayushchikh [Morphological Characteris­ tics of Aquatic Mammals], pages 48 -81 . Moscow: Izdatel'stvo "Nauka." [Translated from Russian by the Fisheries Research Board of Canada, 1969.] Paleopathology in a Miocene Kentriodontid Dolphin (Cetacea: Odontoceti) Susan D. Dawson and Michael D. Gottfried ABSTRACT A specimen of the kentriodontid dolphin Hadrodelphis cal­ vertense, from the Miocene Calvert Formation of Maryland, exhibits pathological lesions on the ribs, vertebrae, and mandible. These lesions were imaged with both plain film radiography and computed tomography. Appearance of the rib lesions is consistent with fracture-callus formation. Two lumbar vertebrae exhibit ver­ tebral spondylosis, which manifests itself as mild exostosis and lipping that does not involve the articular surfaces. These two pathologies are common findings in osteological collections of Holocene cetaceans. Such lesions are especially common among individuals that have reached skeletal maturity; however, this indi­ vidual is skeletally immature. The mandibular lesion consists of extensive periosteal new bone formation caudal to the tooth row, on the lateral aspect of the right mandibular ramus. Two large drainage fistulae are present, one of which reveals a bony sequestrum. New bone formation may result either from neoplastic or infectious processes—the etiology of the lesion appears to be infectious, possibly as a result of bacterial infection secondary to trauma. Mandibular osteomyelitis of this sort does not appear to be common in odontocetes, and new bone formation to this extent likely would have compromised echoloca­ tion. The combination of chronic infection and reduced echoloca­ tion abilities may have led directly to the demise of this individual. Introduction Contributions to the field of vertebrate paleopathology have increased dramatically in recent years (Rothschild and Martin, 1993) as a more sophisticated understanding of bone biology, along with modem veterinary and human medical techniques, has been applied to the fossil record. Relatively few pathologi­ cal conditions have been described in fossil cetaceans, how­ ever—these include the identification of vertebral growth re­ covery lines in a fossil dolphin (Sebes et al., 1990); vertebral exostoses on Miocene "cetotheres" (Kellogg, 1969); and evi- Susan D. Dawson, Department of Anatomy and Physiology, University of Prince Edward Island, Charlottetown, Prince Edward Island CIA 4P3, Canada. Michael D. Gottfried, MSU Museum, Michigan State University, East Lansing, Michigan 48824-1045, United States. dence of predation and/or scavenging by sharks (Demere and Cerutti, 1982; Cigala-Fulgosi, 1990). Our knowledge of disease processes in living cetaceans has been facilitated by stranding networks that recover cetacean carcasses, and by increasingly sophisticated observation and veterinary management of captive cetaceans. This knowledge base is an essential prerequisite for studies of pathologies in fossil cetaceans. Appropriate living models are necessary in or­ der to diagnose a specific disease process based solely upon os­ teological remains, and, despite our detailed medical knowl­ edge, the human model is not always appropriate. The pathological odontocete described herein was collected from the Miocene Calvert Formation in Charles County, Mary­ land. The specimen is referred to Hadrodelphis calvertense Kellogg, 1966, a monotypic genus erected on the basis of a par­ tial mandible. Hadrodelphis has been placed in the family Ken- triodontidae (Dawson, 1996), a diverse assemblage of primarily Miocene delphinoids whose monophyly has been questioned (Muizon, 1988). The lack of fusion of vertebral and forelimb epiphyses indicates that this specimen is a subadult individual. It exhibits three distinct pathological lesions—on the ribs, verte­ brae, and right mandible—that are the subject of this report. The lesions were examined and imaged with conventional radiogra­ phy as well as with computed tomography (CT) scans, and the osteological collection of Holocene cetaceans at the National Museum of Natural History (NMNH), Smithsonian Institution, which subsumed the collections of the former United States Na­ tional Museum (USNM), was surveyed to assess whether simi­ lar lesions occur in extant odontocetes. ACKNOWLEDGMENTS.—It is a great pleasure to participate in this volume commemorating Clayton Ray's many contribu­ tions to paleontology. Among his diverse activities, Clayton encouraged the development of a paleontology program at the Calvert Marine Museum (CMM), the repository for the speci­ men described herein and where much of the research for this paper was conducted. We are very grateful to D. Jarvi, P. Barth, J. Gibson, S. White, and A. Panos, all of the Radiology Department at Cal­ vert Memorial Hospital (Prince Frederick, Maryland) for radio- 263 264 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY graphing and scanning the specimen. G. Daleo (Children's Hospital, San Diego) contributed valuable advice on scanning techniques. B. Rothschild (Arthritis Center of Northeast Ohio) and V. de Buffrenil (Museum National d'Histoire Naturelle, Paris) offered helpful reviews that improved the manuscript. C. Potter (NMNH) facilitated access to the NMNH cetacean col­ lections. Financial support was provided by CMM's Lincoln Dry den Paleontology Fund. Systematic Paleontology Class MAMMALIA Order CETACEA Suborder ODONTOCETI Superfamily DELPHINOIDEA Family KENTRIODONTIDAE Subfamily LOPHOCETINAE Hadrodelphis calvertense Kellogg, 1966 REFERRED MATERIAL.—Calvert Marine Museum CMM-V- 11, an associated specimen consisting of the skull, mandibles, complete vertebral column, ribs, partial right scapula, and right humerus, radius, and ulna. LOCALITY.—CMM-V-11 was recovered in May, 1982, from Indian Creek, Charles County, Maryland, approximately 5 km northeast of the town of Charlotte Hall. The skeleton was par­ tially exposed in the creek in a shell layer assigned to bed 14 of the Calvert Formation, Chesapeake Group. The initial discov­ ery was made by H. Jackson and F. Granados, and the speci­ men was collected by them and D. Bohaska, N. Riker, C. Tay­ lor, and W. Ashby. GEOLOGICAL AND PALEOENVIRONMENTAL SETTING.—The Calvert Formation ranges from late early to middle Miocene in age, and in the Chesapeake region it is divided into 16 beds (re­ ferred to as "zones" by some authors), with bed 16 at the top of the formation. The nearshore marine clays, silts, and sands of the Calvert constitute one of the seven formations that make up the Chesapeake Group, a complex package of late Oligocene to Pliocene sediments deposited along the Middle Atlantic Coastal Plain. Climatic conditions during Calvert time were generally warmer than the region is at present, as reflected in the Calvert fauna, which includes very abundant and diverse invertebrates (especially mollusks), abundant sharks, and bony fish, sea turtles, crocodiles, seabirds, some land mammals, and an important assemblage of marine mammals. Overviews of the geology and biota of the Calvert Formation can be found in Vogt and Eshelman (1987) and Ward (1992). The cetacean fauna has been summarized by Gottfried et al. (1994). DESCRIPTION.—Ribs: Two ribs exhibit a circumscribed area of new bone formation located on each along the shaft of the rib. The larger of the lesions nearly doubles the diameter of the rib shaft, forming a distinct knobby swelling that blends smoothly with the normal cortex on the shaft proximal and dis­ tal to the lesion. The other lesion has a similar appearance but is smaller. The gross appearance of the cortex of the lesions is similar to that of the normal cortical bone. Radiographs of the larger lesion (Figure la) indicate periosteal new bone forma­ tion surrounding a central area of bone lysis. Radiographs also indicate a radiolucent line perpendicular to the long axis of the bone, around which is organized the periosteal new bone for­ mation; this line may represent the original fracture site. Grossly similar lesions are commonly present on ribs of extant cetaceans (MDG, pers. obs., USNM collection). The rib lesions on CMM-V-11 are consistent with fracture-callus formation. A series of short ( -6-9 mm), shallow, subparallel, transverse grooves incised into the rib shaft near the lesion shown in Fig­ ure \a indicate possible predation and/or scavenging by sharks. This is a relatively common occurrence on fossil cetacean re­ mains (e.g., Demere and Cerutti, 1982; Cigala-Fulgosi, 1990), and similar scars also are present on other postcranial elements of the specimen. Vertebrae: Two lumbar vertebral centra (about numbers 24 and 25) exhibit exostoses on their right dorsolateral surfaces. The exostosis on number 24 is relatively small and confined to a portion of the posterior right dorsolateral surface; that on number 25 is larger and extensively covers the right anterior dorsolateral surface of the centrum (Figure lb). This excess bone growth does not involve the articular surfaces but extends outward from each of the affected centra. Grossly, the affected bone has an irregular, globular appearance. CT imaging of the larger exostosis on centrum 25 indicates that the new bone is dense and compact. The incidence of vertebral pathology in cetaceans has pro­ voked a fair amount of discussion (Paterson, 1984; Walker et al., 1986; Kompanje, 1993a). Vertebral exostosis and fusion in the lumbar and caudal regions is frequently observed in extant cetaceans. This condition has variously been termed ankylos­ ing spondylosis, ankylosing spondylitis, spondylosis defor­ mans, or spondylitis deformans (Kompanje, 1993a). These le­ sions are most commonly seen in older individuals. In humans, such lesions occur as part of the aging process and may be a- symptomatic. The incomplete skeletal fusion, however, indi­ cates that CMM-V-11 is a subadult rather than a mature indi­ vidual. Mandible: The right mandible exhibits an extensive area of new bone formation on the lateral aspect of the ramus (Figure 2a) posterior to the tooth row. The mandibles of Hadrodelphis are fused anteriorly along an extensive symphysis (equal to one-third of total mandibular length) and bear 19 alveoli each. The lesion appears to involve the posteriormost alveolus, but it is not localized around that point and extends about 20 cm caudad from the last alveolus. There is no evidence that the le­ sion led to tooth loss—37 of the total complement of 38 man- NUMBER 93 265 FIGURE 1.—a. Detail from radiograph of rib lesion of Hadrodelphis calvertense (CMM-V-11). This view shows a knob-like area of periosteal new bone formation surrounding a central area of lysis, and a radiolucent line (arrow) approximately transverse to the long axis of the rib, both consistent with a fracture callus. (Scale bar=l cm), b, Anterior articular view of lumbar vertebral centrum (about number 25) from CMM-V-11; the area of dense exostosis (arrow) is visible along the right dorsolateral margin of the centrum (note it does not include the articular face of the centrum). (Scale bar=l cm.) dibular teeth were found either in situ on the jaws or in the sur­ rounding matrix. The lesion, which blends relatively smoothly into the normal bone, involves the entire depth of the mandible, resulting in a very swollen and outwardly bowed appearance in the affected area (Figure 2a). The texture of the new bone is more porous than that of the normal mandible anterior to the lesion and on the corresponding portion of the normal left mandible. There is 266 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY FIGURE 2.—a, Articulated mandibles of Hadrodelphis calvertense (CMM-V-11) in dorsal view; note swollen and bowed appearance of right mandible (arrow) (Scale bar=10 cm.) b,c, Detail of the mandibular lesion: in lat­ eral (b) and dorsal (anterior at top) (c) view; note drainage fistulae openings (arrows) in both views, and bony sequestrum within fistular opening in c. (Scale bars= 10 cm.) a subcircular opening, approximately 28 mm across and 20 mm high, located on the caudolateral aspect of the lesion (Figure 2b), and a similar, more nearly circular, opening approximately 26 mm in diameter on the dorsal surface of the lesion (Figure 2c). Both of these are interpreted as drainage fistulae, a conclu­ sion supported by the presence of a sequestrum of isolated ne­ crotic bone visible within the dorsal fistula (Figure 2c). CT imaging (Figures 3,4) of the lesion confirms that the new bone formation is confined to the lateral aspect of the mandi­ ble; in a dorsal-view CT scan, the original mandibular bone is clearly visible, surrounding the mandibular canal (Figure 3). The periosteal new bone formation is extensive and encloses a NUMBER 93 267 FIGURE 3.—Dorsal view computed tomography (CT) image (anterior to bottom) of mandibles of Hadrodelphis calvertense (CMM-V-11), showing extensive periosteal new bone formation on the lateral aspect of the right mandible caudal to the tooth row. Arrows indicate original normal lateral margin of mandible. Normal appear­ ance and thickness of the bone can be seen on the corresponding portion of the left mandible. (Scale bar=5 cm.) network of drainage tracts; the extent of new bone formation is particularly striking when compared with the normal appear­ ance of this part of the mandible on the left ramus (Figure 3). The network of drainage tracts seen on CMM-V-11 is as ex­ pected in cases of chronic infectious processes, with new drain­ age tracts forming as older ones become blocked. Lesions involving extensive new bone formation are typi­ cally the result of either neoplastic or infectious processes. Through radiologic and gross examination it is possible to dis­ tinguish between these two etiologies (Rothschild and Martin, 1993; Rothschild and Rothschild, 1995). We consider neoplasia the less likely of the two possibilities. Neoplasia may be caused by primary~or secondary bone tumors: osteosarcoma, or pri­ mary bone tumor, is seldom reported in modern cetaceans and often involves lysis of bone; secondary bone tumors (me­ tastases from other neoplastic lesions) are typically multifocal in distribution, but CMM-V-11 does not exhibit any other simi­ lar lesions. The presence of a bony sequestrum and the two drainage fis­ tulae strongly suggest osteomyelitis, an infectious process that may be bacterial or fungal in origin, as the cause of the man­ dibular lesion. One form of mandibular lesion called lumpy jaw (Jubb et al., 1993), most commonly seen in cattle, results from an actinomycosis infection thought to progress from trauma to periodontal tissue. Lumpy jaw does involve extensive peri­ osteal new bone formation, but the gross appearance of the le­ sion in Hadrodelphis is inconsistent with lumpy jaw. A more likely pathogenesis is bacterial infection, introduced either by fracture or trauma to the teeth or periodontal tissue. Fracture of the jaw itself can be excluded as the cause of the mandibular lesion. There is no displacement of segments of the mandible, nor is there any disruption of the continuity of the original outline of the mandible in the CT images (Figures 3, 4), or in radiograph, as would be expected in a fracture. The le­ sion does not appear to have a locus of origin within an alveo­ lus, and there is no evidence that it resulted from the bite of a predator. This process of elimination suggests trauma to peri­ odontal tissue as the likeliest route of infection. Discussion The lesions in Hadrodelphis calvertense (CMM-V-11) repre­ sent three separate etiologies: healed rib fractures, idiopathic vertebral spondylosis, and chronic mandibular osteomyelitis following periodontal trauma. Gross and radiologic (Figure la) appearance of the rib lesion is consistent with a fracture callus. De Smet (1977) discussed skeletal pathologies observed in mu­ seum specimens of recent bottlenose dolphins (Tursiops trun­ catus); although the sample size was small (n=l2), rib fracture calluses were noted on half of both stranded and captive indi- 268 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY FIGURE 4.—a-d, Successive transverse cross-sectional CT scans through the mandibular lesion on Hadrodelphis calvertense (CMM-V-11). The lesion was imaged at 1 mm intervals; these are representative slices through the lesion, progressing from caudal to rostral (lateral to the left in each). Small inset images in b-d show location of those slices. The large, laterally positioned fistula is clearly visible in a (arrow), and the more dorsal fistula in d (arrow); isolated areas of dense bone (which show as bright white) in b-d represent sequestra. (Scale bar=5 cm.) viduals. This high incidence of healed fractures suggests that rib fracture is not necessarily debilitating in odontocetes. Kom­ panje (1991) described a stranded killer whale (Orcinus orca) with exostoses on five ribs, which he concluded were caused by osteomyelitis secondary to fracture—evidence that rib frac­ tures are not always resolved without complication. Not only is exostosis on post-cervical vertebrae common in modern delphinids, but it also has been reported in fossil ceta­ ceans, and it provided the etymological basis for the "ce- tothere" Thinocetus arthritus Kellogg, 1969. Kompanje (1993a) summarized some of the terms used in the literature to describe vertebral exostosis in cetaceans and made a convinc­ ing argument for diagnosing the condition as spondylosis de­ formans. He noted that in humans this condition is related to the aging process and is found in almost all geriatric human skeletons. The etiology of spondylosis involves degeneration of one or more vertebral disks, with new bone formation asso­ ciated with sites of ligamentous attachments. Kellogg (1969) referred to the exostoses on Thinocetus ar­ thritus as both spondylitis deformans and osteophytosis (these were later referred to as ligamentous ossifications by Roths­ child, 1987). Walker et al. (1986), in their study of the Pacific NUMBER 93 269 white-sided dolphin (Lagenorhynchus obliquidens), referred to vertebral pathologies as osteonecrosis, although their illustra­ tion appears consistent with exostosis indicative of spondylosis (sensu Kompanje). In a relatively large sample («=63, of which 39 were stranded specimens), Walker et al. found a similarly high incidence of vertebral necrosis (54%). They could not as­ sociate the pathologies with any predisposing factor, including age, however, and concluded that the observed pathologies did not contribute to the strandings. Paterson (1984) suggested that, although extreme cases of new bone formation in verte­ brae may cause vertebral canal stenosis and associated neural deficits, most cases in cetaceans represent part of the normal aging process. Although the condition in Hadrodelphis appears most consistent with vertebral spondylosis (sensu Kompanje), the specimen is subadult, so non-age-related factors must have contributed to the new bone formation. Although clearly less common than rib or vertebral patholo­ gies, cetacean mandibular pathologies have been reported in the literature. Kompanje (1993b) described mandibular osteo­ myelitis in five stranded harbor porpoises (Phocoena phoc­ oena) from the Netherlands. The cause of the infection in three of these cases can be attributed to fracture, as evidenced by clear discontinuities in the bones at the point of injury. In hu­ mans, infection introduced through dental or periodontal tis­ sues (Topazian et al., 1987)—not fracture—is the most com­ mon cause of mandibular osteomyelitis. One of the porpoises described by Kompanje (1993b) may have become infected via periodontal tissues, which also are the likely route in Hadrodel­ phis; neither of these specimens exhibits tooth loss or any in­ volvement of the alveoli. Possible scenarios for the initial periodontal trauma in Hadrodelphis include predation attempts by sharks or some other predator, or trauma during feeding. Although shark pre­ dation on fossil and living cetaceans is well documented, we believe this is a relatively unlikely explanation. A comprehen­ sive study of stranded bottlenose dolphins (Tursiops truncatus) found that most predatory attacks targeted the ventral region of the caudal peduncle, around the genital region (Mead and Pot­ ter, 1990). Corkeron et al. (1987) found evidence of shark bite wounds adjacent to the dorsal fins of a number of T. truncatus (in a study done as part of a photo identification project using dorsal-fin morphology). Long (1991) described a pygmy sperm whale (Kogia breviceps) that had been attacked on the anterior part of its caudal peduncle by a white shark (Carcharodon car- charias). Cigala-Fulgosi (1990) discussed a Pliocene dolphin (Tursiops cortesii) skeleton that, based upon the distinctive tooth marks, also had very likely been attacked by a white shark. Cigala-Fulgosi's description is perhaps the best-docu­ mented instance in the fossil record of a shark attack on a ceta­ cean; the author found that the majority of the wounds were to the thoracic region (ribs, right forelimb, and right scapula) and to the lumbar vertebrae. No bites to the skull or jaws were re­ corded. Taken together, these reports suggest that shark attacks on cetaceans typically concentrate on the trunk and tail regions, not the head or jaws. Shark predation also seems unlikely as the cause of death for CMM-V-11. As mentioned above, tooth marks are present on several elements of the skeleton, and the tip of a shark tooth is embedded in the transverse process of a lumbar vertebra. The lack of remodeling on the bone surrounding these marks im­ plies postmortem scavenging, probably of a mild nature given the relative completeness of the skeleton. Unlike the rib and vertebral pathologies, the mandibular le­ sion may well have contributed to this individual's demise. Chronic infection is often debilitating, and a painfully infected jaw may have inhibited feeding. Additionally, the extensive secondary bone formation on the lateral border of the mandible likely interfered with this part of the jaw's function as an "acoustic window" in echolocation, further affecting the ani­ mal's ability to capture prey and also causing general disorien­ tation. Both debilitation and disorientation are factors in strand­ ing, and in the case of CMM-V-11 the relatively complete nature of the skeleton is consistent with a stranded individual that was not subjected to extensive post-mortem disturbance. Although living cetaceans are the most appropriate model for interpreting pathologies in fossil cetaceans, there are other fac­ tors must be considered when evaluating epidemiological fea­ tures of disease frequency. Modern stranding networks are an invaluable source of basic information on cetacean biology, in­ cluding parasites, toxicology, and incidence of trauma. Osteo­ logical lesions, however, are obviously the most relevant in a paleopathological context, and many such lesions may go un­ detected during typical necropsy procedures. Skeletal prepara­ tion is necessary to assess the frequency of skeletal lesions with confidence. This is particularly true for healed, nohdisplaced fractures and for idiopathic lesions, such as vertebral spondylo­ sis. Unfortunately, complete skeletal preparation is. not feasible in many cases. Another complicating factor is the bias introduced by using museum collections as a source of specimens. The stranded specimens that frequently form the bulk of collections are often debilitated by disease and/or trauma; studies of disease fre­ quencies based upon stranded specimens therefore probably do not reflect natural conditions in wild populations. Studies that include wild-caught or incidental-catch specimens, or that compare data between wild-caught and stranded specimens, are more likely to prove meaningful. It may be trivial to conclude that more study is needed, but clearly that is the case in ceta­ cean paleopathology. Comprehensive comparative studies of pathologies in extant cetaceans will allow for more robust in­ terpretations of disease processes in their fossil relatives. 270 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY Literature Cited Cigala-Fulgosi, F. 1990. Predation (or Possible Scavenging) by a Great White Shark on an Extinct Species of Bottlenosed Dolphin in the Italian Pliocene. Ter­ tiary Research, 12(l):17-36. Corkeron, P.J., R.J. Morris, and M.M. Bryden 1987. Interactions between Bottlenose Dolphins and Sharks in Moreton Bay, Queensland. Aquatic Mammals, 13(3): 109-113. Dawson, S.D. 1996. A Description of the Skull and Postcrania of Hadrodelphis cal­ vertense Kellogg, 1966, and Its Position within the Kentriodontidae (Cetacea; Delphinoidea). Journal of Vertebrate Paleontology, 16(1): 125-134. Demere, T.A., and R.A. Cerutti 1982. A Pliocene Shark Attack on a Cetotheriid Whale. Journal of Pale­ ontology, 6(6): 1480-1482. Gottfried, M.D., D.J. Bohaska, and F.C. Whitmore, Jr. 1994. Miocene Cetaceans of the Chesapeake Group. In A. Berta and T.A. Demere, editors, Contributions in Marine Mammal Paleontology Honoring Frank C. Whitmore, Jr. Proceedings of the San Diego So­ ciety of Natural History, 29:229-238. Jubb, K.V.F., P.C. Kennedy, and N. Palmer 1993. Pathology of Domestic Animals. Fourth edition, volume 1, 780 pages. San Diego: Academic Press. Kellogg, R. 1966. Fossil Marine Mammals from the Miocene Calvert Formation of Maryland and Virginia, Part 4: A New Odontocete from the Calvert Formation of Maryland. Bulletin of the United States National Mu­ seum, 247:99-101. 1969. Cetothere Skeletons from the Miocene Choptank Formation of Maryland and Virginia. Bulletin of the United Stales National Mu­ seum, 294:1^10. Kompanje, E.J.O. 1991. Een oud Geval van Osteomyelitis bij een Orka Orcinus orca. Lutra. 34:71-76. 1993a. Vertebral Osteophytosis in Cetacea: Spondylosis or Spondylitis? Zeitschrift fiir Saugertierkunde, 58:316-318. 1993b. Osteomyelitis of the Mandible in Harbor Porpoises Phocoena phoc­ oena from the Netherlands. Lutra, 36:39—46. Long, D.J. 1991. Apparent Predation by a White Shark Carcharodon carcharias on a Pygmy Sperm Whale Kogia breviceps. United States Fishery Bulle­ tin, 89:538-540. Mead, J.G., and C.W. Potter 1990. Natural History of Bottlenose Dolphins along the Central Atlantic Coast. In S. Leatherwood and R.R. Reeves, editors, The Bottlenose Dolphin, pages 165-196. New York: Academic Press. Muizon, C. de 1988. Les relations phylogenetiques des Delphinida (Cetacea, Mammalia). Annates de Paleontologie, 74(4): 159-227. Paterson, R.A. 1984. Spondylitis Deformans in a Bryde's Whale (Balaeonoptera edeni Anderson) Stranded on the Southern Coast of Queensland. Journal of Wildlife Diseases, 20(3):250-252. Rothschild, B.M 1987. Diffuse Idiopathic Skeletal Hyperostosis as Reflected in the Paleon­ tologie Record: Dinosaurs and Early Mammals. Seminars in Arthri­ tis and Rheumatism, 17:119-125. Rothschild, B.M., and L.D. Martin 1993. Paleopathology: Disease in the Fossil Record. 386 pages. Boca Ra­ ton, Florida: CRC Press. Rothschild, B.M., and C. Rothschild 1995. Comparison of Radiologic and Gross Examination for Detection of Cancer in Defleshed Skeletons. American Journal of Physical An­ thropology, 96:357-363. Sebes, J.L, B.M. Rothschild, J.W. Langston, and M.L. Gavant 1990. Growth Recovery Lines in Fossil Vertebrae. Journal of Vertebrate Paleontology, supplement, 19(3):41A. Smet, W.M.A. de 1977. The Fate of Old Bottle-nosed Dolphins, Tursiops truncatus. in Na­ ture As Revealed by the Condition of Their Skeletons. Aquatic Mam­ mals, 5(l):78-86. Topazian, R.G., and M.H. Goldberg, editors 1987. Oral and Maxillofacial Infections: Osteomyelitis of the Jaws. Sec­ ond edition, 473 pages. Philadelphia: W.B. Saunders. Vogt, P.R., and R.E. Eshelman 1987. Maryland's Cliffs of Calvert: A Fossiliferous Record of Mid-Mio­ cene Inner Shelf and Coastal Environments. Geological Society of America Centennial Field Guide, Northeast Section, pages 9-14. Walker, W.A., S. Leatherwood, K.R. Goodrich, W.F. Perrin, and R.K. Stroud 1986. Geographical Variation of the Pacific White-sided Dolphin, Lageno- rhynchus obliquidens, in the Northeastern Pacific. In M.M. Bryden and R. Harrison, editors, Research on Dolphins, pages 441-466. Oxford: Oxford Scientific Publishers. Ward, L.W. 1992. Molluscan Biostratigraphy of the Miocene Middle Atlantic Coastal Plain of North America. Virginia Museum of Natural History Mem­ oir, 2: 159 pages Paleontology of the Late Oligocene Ashley and Chandler Bridge Formations of South Carolina, 2: Micromysticetus rothauseni, a Primitive Cetotheriid Mysticete (Mammalia: Cetacea) Albert E. Sanders and Lawrence G. Barnes ABSTRACT Previously named fossil mysticete species in the extinct genus Cetotheriopsis Brandt, 1871, are C. lintianus (von Meyer, 1849), from Austria, and C. tobieni Rothausen, 1971, from Germany, both of late Oligocene (lower Chattian) age. Although each spe­ cies is known only by a single braincase without a rostrum, both have been understood to be baleen-bearing, archaic mysticetes of the extinct family Cetotheriidae. Fossil baleen whales of this age are generally rare worldwide, and no Oligocene cetotheriids have been named previously from the eastern coast of North America. Elsewhere, we have established the family Eomysticetidae to accommodate two very primitive new mysticetes from the late Oligocene Chandler Bridge Formation near Charleston, South Carolina (Sanders and Barnes, 1999, 2002). In addition, two mys­ ticete braincases without rostra from the underlying Ashley For­ mation near Charleston represent a new genus and new species in the Cetotheriidae. These specimens document a new taxon that is markedly different from Cetotheriopsis and is herein named Micromysticetus rothauseni, new genus, new species It is most similar to Cetotheriopsis tobieni Rothausen, 1971, which is herein referred to the new genus Micromysticetus. Micromysticetus roth­ auseni is a significant addition to the diverse Oligocene marine vertebrate assemblages from the Oligocene beds near Charleston. It constitutes the oldest known western North Atlantic records of the family Cetotheriidae and provides additional evidence of the archaeocete ancestry of the mysticetes. Introduction Tertiary marine deposits of South Carolina have yielded im­ portant fossil cetacean remains since 1845, when Robert W. Gibbes described the archaeocete Dorudon serratus from Eocene beds in the vicinity of the Santee River. Subsequent Albert E. Sanders, The Charleston Museum, 360 Meeting Street, Charleston, South Carolina 29403. Lawrence G. Barnes, Natural His­ tory Museum of Los Angeles County, 900 Exposition Boulevard, Los Angeles, California 90007. discoveries have included two odontocetes, Agorophius pyg­ maeus (Muller, 1849) and Xenorophus sloanii Kellogg (1923b), both of which are known from holotype specimens found in the late Oligocene (ca. 30 Ma) Ashley Formation. In recent years, studies emanating from The Charleston Museum (ChM) have revealed a previously unrecognized rock unit overlying the Ashley Formation. Named the Chandler Bridge Formation by Sanders et al. (1982), this unit is slightly younger than the Ashley Formation, is also of late Oligocene age (ca. 28 Ma), and has yielded a wealth of cetacean material that is pro­ viding critical new information about the evolution and sys­ tematics of Oligocene whales of the western hemisphere (Whitmore and Sanders, 1976; Sanders, 1980; Sanders et al., 1982; Weems and Sanders, 1986; Sanders and Barnes, 1989, 1991, 1999, 2002; Barnes and Sanders, 1990). With the exception of the toothed mysticetes of the genera Aetiocetus Emlong, 1966, and Chonecetus Russell, 1968, from the northwest coast of the United States, Oligocene-age mys­ ticetes from the Northern Hemisphere were formerly known only from Europe. In 1975 the remains of a primitive, archaeo- cete-like baleen whale were found in the late Oligocene Chan­ dler Bridge Formation near Charleston, South Carolina, and were excavated by the Charleston Museum (Sanders and Barnes, 1989, 1999, 2002; Barnes and Sanders, 1990). We now describe and name an additional mysticete from Oligocene de­ posits near Charleston, a preliminary report of the holotype (ChM PV4844) having been made by Sanders and Barnes (1991). A partial braincase (ChM PV5933) referable to this taxon also has been found, and both specimens are conserved in the Charleston Museum collections. The purpose of this study is to report this newly recognized whale from South Carolina and to comment on its taxonomic affinities. We recog­ nize that the family Cetotheriidae and the subfamily Cetotheri- opsinae are greatly in need of review and definition, but that task is beyond the scope of this paper. 271 272 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY AREA ENLARGED jjjP FIGURE 1.—Map showing the localities for the holotype (A) and the paratype (B) of Micromysticetus rothauseni, new genus and new species, in Dorchester County, South Carolina, approximately 20 miles north of Charleston. (Base map from Sanders, 1980, courtesy of National Geographic Society.) ACKNOWLEDGMENTS.—We appreciate this opportunity to recognize the contributions of our longtime friend and col­ league Clayton Ray. For more than 25 years he has been ever helpful to us and, in his quiet, tactful manner, has often sug­ gested an alternative answer to a question in such a way that one would wonder why that solution had not seemed evident in the first place. A thorough, meticulous worker, he also is one of the finest writers who has ever graced this profession, having the rare ability to transmit his findings with a lucidity that leaves little doubt about his interpretation of the data that he has presented. A veritable bulldog when in pursuit of an ob­ scure or elusive subject, he leaves virtually no stone unturned until he either has found what he is seeking or is certain that it probably won't be found. One can truly say that if Clayton Ray has been over the ground, there probably isn't much left to find there. More importantly, he is a splendid human being, possess­ ing many of the qualities that distinguished Joseph Leidy. We consider it a privilege to have known him and to have worked with him, for it is doubtful that we will see his like again in our lifetimes. We thank Vance McCollum, the collector of the holotype of Micromysticetus rothauseni, and Billy Palmer, who collected the paratype, for their generosity in making these specimens available to science. We are especially grateful to Karlheinz Rothausen, now retired from the Institut fiir Geowissen- schaften-Palaontologie, Johannes Gutenberg-Universitat, Mainz, Germany, for the loan of his original drawings, photo­ graphs, and negatives of Micromysticetus (Cetotheriopsis) to­ bieni, for his review of our manuscript, and for many other fa­ vors. Bernhard Gruber of the Oberosterreichisches Landes- museum in Linz, Austria, kindly permitted K. Rothausen and the first author to examine the holotype of Cetotheriopsis lint- ianus and other specimens in his care and made arrangements for the excellent photographs of the holotype of C. lintianus by the museum photographer Bernhard Ecker. Sanders extends thanks to the Deutsche Forschungsmeinschaft in Bonn, Ger­ many, for financial assistance in a grant obtained by K. Roth­ ausen to cover lodging expenses incurred in the study of the European material involved in this study and in another study in which he and Rothausen are collaborating. The European segment of this study also was funded in part by a National Geographic Society grant to the first author in connection with his studies of other cetacean specimens at the Oberosterreichi­ sches Landesmuseum in Linz. We also are grateful to the Na­ tional Geographic Society for permission to use as the base map for our Figure 1 the map published as figure 1 in the first author's report of paleontological excavation activities carried out under the grants (numbers 954 and 1074) from the society (Sanders, 1980:603). NUMBER 93 273 The Charleston Museum and the Natural History Museum of Los Angeles County and its foundation provided facilities, sal­ aries, and travel funds that made this study possible. AES pro­ duced the drawings. Bryan Stone produced all of the photo­ graphs except Figure 13c, made by Terry Richardson; Figures llA,B, 12A-C, and 14, made by AES; and Figures 15-17, made by Bernhard Ecker. We thank David P. Whistler for comments on the manuscript and Zhexi Luo for his review of the text and figures of our description of the periotics of Micromysticetus rothauseni. Clayton E. Ray and Frank C. Whitmore, Jr., pro­ vided access to fossil cetacean material in the National Mu­ seum of Natural History and were helpful in numerous other ways. MATERIAL AND METHODS Terminology for cranial anatomy follows Kellogg (1938) and Fraser and Purves (1960). With some modifications re­ quired by the morphology of the specimens, cranial measure­ ments follow Perrin (1975) and Kellogg (1936, 1965). The cra­ nial reconstructions are in part mirror images of opposite sides of the specimens, and those of the two European taxa are based upon photographs of the holotypes and upon notes made from direct observations by Sanders. Broken lines represent areas of the skull not preserved in the actual specimen and thus are in­ ferred or hypothetical configurations. Geological interpretations of the rock units that yielded the holotype specimens are in accordance with the original de­ scription of the Chandler Bridge Formation (Sanders et a l , 1982) and subsequent observations (Weems and Sanders, 1986). Direct comparison of a cast of the holotype of Cetotheriopsis tobieni Rothausen, 1971, with the holotype of Cetotheriopsis lintianus (von Meyer, 1849) was made by K. Rothausen and one of us (AES) in July, 1992, and that material was compared with photographs of the holotype of the new taxon from South Carolina described in this paper. Where helpful, the fossils de­ scribed herein have been compared with specimens of certain extant taxa. ABBREVIATIONS.—The following abbreviations are used for institutions housing specimens used in this study: ChM The Charleston Museum, Charleston, South Carolina FMNH Field Museum of Natural History, Chicago, Illinois JGU Institut fur Geowissenschaften-Palaontologie, Johannes Guten- berg-Universitat, Mainz, Germany LACM Natural History Museum of Los Angeles County, Los Angeles, California UCMP University of California Museum of Paleontology, Berkeley, California UO University of Oregon, Eugene, Oregon USNM Collections of the National Museum of Natural History, Smith­ sonian Institution, Washington, D.C. (including the collections of the former United States National Museum) ABBREVIATIONS.—The following anatomical abbreviations are used in this study: Al be Boc Bs ch Eoc fm fps Fr gf hpt jn Mc mea Na Oc Pa Pal Pgi pop Pt pts Soc Sq sqf sqp tf Vo zps alisphenoid basioccipital crest basioccipital basisphenoid cranial hiatus exoccipital foramen magnum foramen pseudovale frontal bone glenoid fossa hamular process of pterygoid jugular notch mesethmoid external auditory meatus nasal occipital condyle parietal palatine postglenoid process paroccipital process pterygoid fossa for pterygoid sinus supraoccipital squamosal squamosal fossa squamosal prominence temporal fossa vomer zygomatic process of squamo GEOLOGIC SETTING Throughout the area near Charleston, the Chandler Bridge Formation is underlain by the Ashley Formation (Figure 2). Both formations are late Oligocene (early Chattian) marine units that are rich in fossil vertebrate material. The Chandler Bridge Formation was laid down approximately 28 million years ago (Sanders et al., 1982), and the Ashley Formation is considered to be about 30 million years old. The holotype material of Micromysticetus rothauseni, new genus and new species (ChM PV4844), was found in the Ash­ ley Formation at the bottom of a channelized stream only a few hundred feet west of the type locality of the Chandler Bridge Formation. The other specimen (ChM PV5933) was found rel­ atively nearby, also in the Ashley Formation. The uppermost portion of the Chandler Bridge Formation (Bed 3) is a beach facies from which a large number of marine vertebrate remains were recovered in a major excavation con­ ducted by the Charleston Museum (Sanders, 1980). The site of that excavation is the type locality for the Chandler Bridge For­ mation and is 2.1 km southwest of the type locality of another Oligocene mysticete, the archaic Eomysticetus whitmorei Sanders and Barnes (2002). The Chandler Bridge Formation unconformably overlies the late Oligocene Ashley Formation (ca. 30 Ma), a calcarenite that underlies the entire Charleston area. Together, these two forma­ tions have produced a comprehensive assemblage of Oli- gocene-age marine-associated vertebrates. Included are an ex­ tensive fish fauna; five taxa of sea turtles; a large crocodile, 274 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY Ground level Wando Formation (Late Pleistocene) Pebble bed x ca. 0.8 m FIGURE 2.—Generalized stratigraphic section of late Oligocene (lower Chat­ tian) beds along Chandler Bridge Creek and Eagle Creek near bridge on Lad- son Road (County Road 230) in Dorchester County, South Carolina. Both the holotype and the paratype crania of Micromysticetus rothauseni, new genus and new species, were found in the Ashley Formation. Gavialosuchus carolinensis Erickson and Sawyer (1996); sev­ eral species of littoral and pelagic birds (including the largest and most nearly complete pseudodontorn yet found); at least two taxa of dugongids; and one of the largest and most diverse assemblages of cetacean remains yet recovered from any Oli­ gocene deposit. Representatives of Mysticeti and Odontoceti have been found in the Chandler Bridge Formation, but squal- odontoids and other primitive odontocetes are by far the most numerous, with mysticetes being relatively uncommon. To date, there has been no evidence of faunal assemblages or indi­ vidual taxa that are positively restricted to either the Chandler Bridge or the Ashley Formation, although Micromysticetus rothauseni is presently known only from the Ashley. An early Chattian age for the Chandler Bridge Formation is indicated by the presence of undescribed squalodonts of the same evolution­ ary grade as Eosqualodon langewieschei Rothausen, 1968 (see Whitmore and Sanders, 1976; Sanders, 1980), which is from Eochattian sands (Chattian A) at Doberg, Germany. The Chat­ tian A beds at Doberg have been referred to nannoplankton zone NP24 (Martini and Muller, 1975) and are considered to be of early Chattian age (Curry et al., 1978:46). A detailed ap­ praisal of the age of the Chandler Bridge Formation is given by Sanders etal. (1982). E. Martini (pers. comm., June 1990) examined nannoplank­ ton from the underlying Ashley Formation at the type locality of the Chandler Bridge Formation and found the Ashley For­ mation to be referable to zone NP24. Because the same evolu­ tionary grades and many of the same genera are represented among the cetaceans, sea turtles, and other vertebrate faunas of both the Ashley and the Chandler Bridge Formations, we con­ sider that these two formations belong to the same biostrati- graphic interval (NP24) and that very little time (perhaps only about 2 My) elapsed between the deposition of these two units. Systematic Paleontology Class MAMMALIA Linnaeus, 1758 Order CETACEA Brisson, 1762 Suborder MYSTICETI Flower, 1864 Superfamily BALAENOPTEROIDEA (Gray, 1868) Family CETOTHERIIDAE (Brandt, 1872) Miller, 1923 Subfamily CETOTHERIOPSINAE Brandt, 1872 Micromysticetus, new genus Cetotheriopsis Brandt, 1871 [in part].—Rothausen, 1971:135. DIAGNOSIS.—A genus of cetotheriopsine cetotheriids differ­ ing from Cetotheriopsis in the following characteristics: smaller size; cranium with an occipital shield shaped like a broad equilateral triangle, not an anteroposteriorly elongate tri­ angle, medial crest on midline of occipital shield less promi­ nent, squamosal fossa short anteroposteriorly, shallow, and in­ terrupted by a protuberance—herein termed the squamosal prominence—extending posterolaterally to its posterior margin; exoccipital thick anteroposteriorly and narrow transversely; zy­ gomatic process of squamosal elongate, deep, arched, and ex­ tending beyond apex of supraoccipital; glenoid fossa broad transversely with rounded posterointernal margin; basioccipital between basioccipital crests flat to slightly convex; sulcus for external acoustic meatus short and broad. Adults with no prominent sulcus dorsal to occipital condyles on occipital shield, ETYMOLOGY.—From mikros (Greek), small, and mystax (Greek), moustache, in reference to baleen; and from ketos (Greek), whale. TYPE SPECIES.—Cetotheriopsis tobieni Rothausen, 1971. INCLUDED SPECIES.—Micromysticetus tobieni (Rothausen, 1971) and Micromysticetus rothauseni, new species. Micromysticetus tobieni (Rothausen, 1971), new combination FIGURES 3-6 Cetotheriopsis tobieni Rothausen, 1971:135, figs. 2, 3, plates 1, 2. EMENDED DIAGNOSIS.—A small species of Micromysticetus probably not exceeding 4.5 m in total length, differing from M. rothauseni, new species, by having large occipital condyles, occupying more than 45% of distance between outer margins NUMBER 93 275 / v. \ ,-\ f\ X Fr \ / / \ FIGURE 3.—Micromysticetus tobieni (Rothausen, 1971): A, holotype braincase (JGU P1289), dorsal view (from original photograph used in Rothausen 1971, table 1: fig. 1; courtesy of K. Rothausen); B, reconstruction. (Dashed lines indicate hypothetical configurations; abbreviations are explained in "Material and Methods.") of exoccipitals. Lateral margins of supraoccipital almost straight in anterior two-thirds, then forming a slight angle in posterior portion; inner margins of basioccipital crests rounded and curving away from each other posteriorly, the space be­ tween them being slightly convex. HOLOTYPE.—Partial braincase, JGU P1289, collection of In­ stitut fiir Geowissenschaften-Palaontologie, Johannes Guten- berg-Universitat, Mainz, Germany. Previously in the private collection of Fritz von der Hocht, Krefeld, Germany. TYPE LOCALITY.—Kiesgrube Wilhelm Frangen, Lank-La- 276 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY choana posterio TD process, 10_cm pts Lamina lateralis, Pt Lamina mesia Al fragment of periotic periotic pop Eoc I Boc foramen Q C air sinus system FIGURE 4.—Micromysticetus tobieni (Rothausen, 1971), holotype braincase (JGU P1289): A, posterior view (Rothausen 1971, table 1: fig. 3); B, left lateral view (Rothausen 1971, table 2: fig. 4). (From original photo­ graphs courtesy of K. Rothausen; abbreviations are explained in "Material and Methods.") turn, WNW Dusseldorf-Kaiserwerth, Nordrhein-Westfalen, Germany FORMATION AND AGE.—Meeressande, upper "Chattian A" beds, nannoplankton zone NP24 (Martini and Muller, 1975), lower Chattian, late Oligocene. REMARKS.—The holotype cranium of Micromysticetus tobi­ eni lacks both of the zygomatic processes, but it seems appar­ ent that had the anterior ends of those processes been preserved they would have extended beyond the anteriormost extent of the apex of the supraoccipital, as in M. rothauseni, new species, NUMBER 93 277 Soc Eoc FIGURE 5.—Micromysticetus tobieni (Rothausen, 1971), holotype braincase (JGU P1289): A, posterior view (Rothausen 1971, table 1: fig. 3); B, left lateral view (Rothausen 1971, table 2: fig. 4). (From original photo­ graphs courtesy of K. Rothausen; abbreviations are explained in "Material and Methods.") (Figures 3B, 7). The shape of the supraoccipital shield in the holotype is almost that of an equilateral triangle, the axes of the lateral margins intersecting the plane of the posterior margin of the occipital condyles at an angle of approximately 60°. The squamosal prominences are partly eroded but clearly evident and divide the floor of the squamosal fossa into two levels, as in M. rothauseni, new species, but not as conspicuously. The occipital condyles are quite large (Figure 8), occupying 46% of the space between the outer margins of the exoccipitals, but the condyles in M. rothauseni, new species, occupy a slightly 278 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY p g l FIGURE 6.—Micromysticetus tobieni (Rothausen, 1971), holotype braincase (JGU P1289), anterior view. (Abbre­ viations are explained in "Material and Methods"; photograph courtesy of K. Rothausen.) greater portion of the width of the exoccipital region (48% in the holotype skull). Ventrally, the inner margin of the glenoid fossa bordering the cranial hiatus is more rounded than in Ce­ totheriopsis lintianus, in which the corresponding border is sharply angular (Figures 4, 16). Micromysticetus rothauseni, new species FIGURES 7-14 Cetothere "possibly related to Cetotheriopsis."—Sanders and Barnes, 1991. DIAGNOSIS.—A species of Micromysticetus differing from Micromysticetus tobieni in the following characteristics: cra­ nium with more rounded posterior margin of squamosal fossa; median crest on supraoccipital not extending close to dorsal margin of foramen magnum; lateral margin of occipital shield more convex; exoccipital thicker anteroposteriorly and pro­ truding more posteroventrally; basioccipital crests angular and not rounded as in M. tobieni, and with space between them flat and not convex. HOLOTYPE.—ChM PV4844, braincase lacking the entire ros­ trum and the right zygomatic process of the squamosal; both periotics; and axis vertebra. Collected by Vance McCollum in March 1986. TYPE LOCALITY.—South Carolina, Dorchester County; bot­ tom of channelized bed of Chandler Bridge Creek in Hickory Hills housing development, -0 .66 km (0.41 mi.) north of County Road 230 near S.C. Route 642, approximately 32 km (20 mi.) north of Charleston (Figure 1). FORMATION AND AGE.—Ashley Formation, late Oligocene, early Chattian correlative, nannoplankton zone NP 24, ca. 30 Ma. The skull and related specimens were found in the top of the Ashley Formation exposed in the streambed, approximately 30 cm below the contact with the overlying Chandler Bridge Formation, and only a few hundred meters west of the type lo­ cality of the Chandler Bridge Formation, marked as "Excava­ tion Site" in Figure 1. ETYMOLOGY.—The specific name is a patronym honoring Karlheinz Rothausen, in recognition of his many contributions to our knowledge of the Oligocene cetaceans of Europe. PARATYPE.—ChM PV5933, partial braincase missing left side and most of basicranium. South Carolina, Dorchester County; bank of Eagle Creek, -152 m (500 ft.) northeast of confluence with Chandler Bridge Creek; County Road 230, -1.04 km (0.65 mi.) southwest of type locality of Micromys­ ticetus rothauseni. Collected by Billy Palmer, November 1997. DESCRIPTION.—Cranium: Although missing most of the zygomatic process of the right squamosal and a large section of the supraoccipital (Figure 7A) , the holotype cranium of Mi­ cromysticetus rothauseni is well preserved otherwise, except where some of the dorsal surfaces have been eroded, especially along the lambdoidal crests and portions of the exoccipitals, occipital condyles, and zygomatic processes of the squamosals. Micromysticetus rothauseni has a broad, triangular brain­ case; large and widely flaring zygomatic processes of the squa­ mosals; and a narrow squamosal with a well-developed squa­ mosal prominence (Figure 7). The parietals are joined at the midline and form the posterior portion of an intertemporal con­ striction with a narrowly rounded sagittal crest. This portion of the skull roof is preserved for a distance of 48 mm anterior to the apex of the supraoccipital and shows evidence of interdigi­ tation with the frontals; thus, it is probable that most, if not all, of the parietal portion of the intertemporal region is preserved NUMBER 93 279 FIGURE 7.—Micromysticetus rothauseni, new genus and new species: A, holotype braincase, ChM PV4844, dor­ sal view; B, reconstruction. (Dashed lines indicate hypothetical configurations; abbreviations are explained in "Material and Methods.") 280 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY 5 cm tr+i?-- \ 10 cm FIGURE 8.—Micromysticetus rothauseni, new genus and new species: A, holotype braincase, ChM PV 4844, ven­ tral view; B, reconstruction. (Dashed lines indicate hypothetical configurations; abbreviations are explained in "Material and Methods.") in the holotype. Details of that region anterior to the parietals are not known, but it seems likely that the frontals would have continued the constriction anteriorly for approximately the same distance as the length of the parietal portion. A moder­ ately long, narrow, intertemporal region in the skull of this ani­ mal would have been necessary for the postorbital process of the frontal to clear the anterior tip of the zygoma. The occipital shield is triangular and relatively broad and is slightly concave anterior to the foramen magnum and medial to the nuchal crest. Dorsal to the occipital condyles the surface of the occipital shield is flat to convex. The lateral and apical mar­ gins of the supraoccipital are missing entirely, but the squamo­ sal and parietal rims of the lambdoidal crests provide the out­ line of its original form. Laterally, they flare outward and overhang the temporal fossa. The axis of the left lambdoidal crest departs the plane of the posterior margins of the occipital condyles at an angle of approximately 75° (at first nearly 90° in M. tobieni, then bending anteriorly at an angle of about 60°). Anterodorsally, the margins of the lambdoidal crests curve in­ ward toward the apex of the supraoccipital. In M. rothauseni, therefore, the margins of the supraoccipital are noticeably con­ vex as they approach the apical region, whereas in M. tobieni NUMBER 93 281 FIGURE 9.—Micromysticetus rothauseni, new genus and new species, holotype braincase, ChM PV 4844, posterior view. (Abbreviations are explained in "Material and Methods.") the marginal curvature is more subtle and much less noticeable. Dorsally, there is a medial crest that extends posteriorly from the apical portion of the supraoccipital for about one-half the distance to the dorsal margin of the foramen magnum. In M. to­ bieni there is a similar crest that extends posteriorly for approx­ imately three-fourths of the distance to the foramen magnum. The occipital condyles are very large for a skull of such mod­ est dimensions and are exceptionally broad transversely, occu­ pying 48% of the distance (256.6 mm) between the outer mar­ gins of the exoccipitals. Despite their massive size, the condyles do not protrude prominently from the occipital shield. The plane of the dorsalmost margins of the condyles is on a level with the posterior margin of the floors of the squamosal fossae. In certain areas where they are least eroded, the articu­ lar surfaces of the condyles are punctate, suggesting that the specimen does not represent a fully mature individual. Al­ though there is considerable variation in the amount of hori­ zontal space taken up by the condyles in other species of ce- totheres, that variation does not appear to be entirely a function of individual body size. In the holotype skull of the middle Mi­ ocene cetothere Cophocetus oregonensis Packard and Kellogg, 1934 (UO 305), the condyles occupy a mere 28% of the exoc­ cipital region, although the anteroposterior length (1.2 m) of this specimen is only one-half the length (2.5 m) of the holo­ type skull (LACM 882) of the late Miocene cetotheriid Mixo- cetus elysius Kellogg, 1934a, in which the condyles occupy 34% of the exoccipital region. In the smallest known mysticete, represented by a partial cranium (UCMP 26502) from the late Miocene Towsley Formation (Barnes, 1977:329) of California and described as Nannocetus eremus by Kellogg (1929), who considered it to be a small cetothere, the condyles take up 49% of the distance between the outer margins of the exoccipitals (182 mm). In the considerably larger form M. rothauseni, how­ ever, they compose 48% of that distance (256.6 mm), nearly the same as in Nannocetus. It thus appears again that larger species do not necessarily have significantly smaller condyles than more diminutive taxa. There may be some ontogenetic variation in the size of the condyles relative to the transverse diameter of the exoccipital region, as suggested by the paratype of M. rothauseni (ChM PV5933), in which the condyles oc­ cupy 50% of the space between the exoccipital margins. That specimen is decidedly a much younger individual than the ho­ lotype. Nevertheless, the exceptionally large condyles of M. rothauseni seem to stand alone in comparison with other known cetotheres except Nannocetus and thus may be tenta­ tively regarded as diagnostic for this species until proven other­ wise by additional material. The condyles of M. tobieni also are comparatively large (46% of the width of the exoccipital re­ gion), which suggests that large condyles may in fact be char­ acteristic of the genus Micromysticetus. Even if an allowance is made for allometric considerations, the condyles of these two FrGURE 10.—Micromysticetus rothauseni, new genus and new species: A, holotype braincase, ChM PV4844, left lateral view; B, reconstruction. (Dashed lines indicate hypothetical configurations; abbreviations are explained in "Material and Methods.") 282 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY FIGURE 11.—A, Micromysticetus rothauseni, new genus and new species, holotype right and left periotics, ChM PV4844; B, left periotic of the archaeocete Zygorhiza kochii (ChM PV5065) (missing posterior process) from the late Eocene Harleyville Formation in Dorchester County, South Carolina (left), and the holotype left periotic of Micromysticetus rothauseni, new genus and new species (right). taxa still are proportionately larger than those of most of the other cetotheres noted above. Lateral to the condyle, the surface of each exoccipital is not concave as in Cetotheriopsis lintianus. The exoccipital portion of the occipital shield is thick, as is typical of most fossil and extant baleen whales. In its ventral part, however, the paroccip­ ital process is exceptionally thick, and it extends posteroven­ trally. Between the braincase and the zygomatic process, the squa­ mosal fossa is shallow, anteroposteriorly short, and wide. Aris­ ing from the center of the floor of the squamosal fossa there is a knobby protuberance that we formally term the "squamosal prominence" (sqp). This prominence forms a scarp-like ridge that divides the floor of the squamosal fossa into two levels, the posterior level being elevated approximately 15 mm above the anterior level. Anteriorly from the anterior margin of the prom­ inence, the floor of the fossa forms a narrow gutter that slopes ventrally to the anterior margin of the squamosal. The posterior level of the fossa is on a plane approximately parallel to the sagittal plane of the skull. A modified extension of the lamb­ doidal crest, the squamosal prominence is joined to the poste­ rior part of the zygomatic process of the squamosal. The lateral wall of the braincase is nearly vertically oriented and slightly concave near its middle. It is not bowed laterally as is typical of most balaenopterids. The zygomatic process of the squamosal is elongate, is very large relative to the size of the cranium, and diverges anterolat- erally from the midline of the cranium. It arches anterodorsally in a broad curve and bends ventrally at its anterior extremity. The process is thick dorsoventrally and is of uniform thickness for most of its length, terminating in a rounded point anteriorly. The medial surface is uniformly smooth. The lateral surface is uniformly bowed outward, making its lateral side smoothly convex, and it flares slightly ventrolaterally at the ventral mar­ gin, especially so at the posterior end of the process, which is rounded dorsally and flattened ventrally to receive the posterior end of the jugal. The zygomatic process of the squamosal is slightly expanded dorsoventrally at its posterior end and arches slightly anteroposteriorly to create the large, anterodorsally in­ clined glenoid fossa. At the posterior end of the lateral surface of the zygomatic process is a prominent, vertically oriented stemomastoid fossa. This fossa is inclined anterodorsally and is located immediately dorsal to the large paroccipital process. The latter is thick an­ teroposteriorly, is narrow transversely, and projects promi­ nently ventrally and posteriorly from the occipital shield. NUMBER 93 283 7-0 mm 25 mm pa.co. 20 mm 25 mm 20 mm 25 mm fe.r. pr.p. FIGURE 12.—Micromystice­ tus rothauseni, new genus and new species, holotype left periotic, ChM PV4844: A, ventral view; B, cerebral view; C, dorsolateral view. (Abbreviations: a.c. = aque- ductus cochlearis; c.n.=fora- men for cochlear nerve; f.i.= fossa incudis; f.n. = foramen for facial nerve; fe.o. = fenes­ tra ovalis; fe.r. = fenestra rotunda; f.t.t. = fossa for ten­ sor tympani muscle; i.a.m.= internal auditory meatus; n.f. = notch for exit of facial nerve; pa. co.=pars cochle­ aris; po.=promontorium; pr.a. =anterior process; pr.p.=pos- terior process; s.f.=subarcu- ate fossa; v.n. = foramen for vestibular nerve.) The recess for the external auditory meatus is deep, but short transversely and broad anteroposteriorly and is clearly defined between the postglenoid process of the squamosal and the par­ occipital process. The cranial hiatus, the recess between the squamosal and the basioccipital in which the periotic lies, is large and elongated anteroposteriorly. Anterolateral to the cranial hiatus and imme­ diately posterior to the large, obliquely oriented foramen pseudovale is the falciform process. This process is thick and is connected to the postglenoid process of the squamosal by a very slender and low, transverse crest of bone. This crest marks the lateral border of a large pterygoid sinus extending anteri­ orly from the cranial hiatus. The ventral surface of the basioc­ cipital is nearly flat. The basioccipital crests are broad, rectan­ gular, and flattened ventrally, their inner margins paralleling the midline axis of the skull before breaking into divergent an­ gles that terminate in narrow, rounded points posteriorly (Fig­ ure 7). As depicted by Rothausen (1971:137, pl. 3: fig. 5), the basioccipital crests (Alae Basioccipitalis in Rothausen, 1971) of Micromysticetus tobieni are rounded and knob-like, their in­ ner margins curving away from each other posteriorly. The glenoid fossa is broad transversely, with a nearly square articular surface, which is poorly delimited anteriorly. The in­ nermost margin bordering the cranial hiatus is rounded, as in Micromysticetus tobieni. The postglenoid process is canted posterolaterally and is thickest laterally, where its ventral mar­ gin is convex, then becomes thinner medially, where it termi­ nates at the margin of the pterygoid sinus. Periotics: The holotype left and right periotics of Micro­ mysticetus rothauseni (ChM PV4844) are well preserved (Fig­ ures 11, 12). The left periotic is missing only the dorsalmost extension of the superior process, which is present in the right periotic and has been utilized in the reconstruction of the left one (Figure 12). The right one is chipped in several places. From the anteriormost point on the anterior process to the pos­ teriormost point on the posterior process, the left periotic is 73 mm in length. The pars cochlearis is 28 mm in dorsoventral di­ ameter, 36 mm anteroposteriorly, and 34 mm in transverse di­ ameter. Its external face is strongly rugose. Immediately adja- 284 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY cent to the dorsal end of the subarcuate fossa is a much smaller fossa that is reduced to a tiny pit in the right periotic; other­ wise, these two bones are mirror images of each other and their measurements do not differ significantly. The anterior process is short, laterally compressed, and broadly expanded dorsoven­ trally. The posterior process also is short and is much smaller than both the anterior process and the pars cochlearis, quite un­ like the periotics of most Miocene cetotheres, in which the pos­ terior process is massive and greatly extended posteriorly (Kellogg, 1934a,b 1965, 1968a,b, 1969; Packard and Kellogg, 1934). The periotic of M. rothauseni bears little resemblance to those of Miocene cetotheres, and in ventral aspect it is most similar to the periotic of the late Eocene archaeocete Zygorhiza kochii (Reichenbach, 1847) as figured by Kellogg (1936:116, fig. 35a) and as seen in ChM PV5065, a left periotic of Zy­ gorhiza (missing the posterior process) from Eocene beds in the Giant Portland Cement quarry near Harleyville, Dorchester County, South Carolina (Figure 11B). The anterior process of the periotic is more laterally compressed in Micromysticetus than in Zygorhiza and the posterior process is much smaller than in the latter (Kellogg, 1936:116, fig. 35a), but the relative positions of the fossa for the tensor tympani muscle, the fenes­ tra ovalis, the fenestra rotunda, the canal for the facial nerve, the fossa incudis, and the subarcuate fossa are much the same in PV4844 and PV5065. In cerebral (internal) aspect, the ante­ rior process is hatchet-shaped, like that of Zygorhiza, and in the left periotic it is 38.8 mm in dorsoventral diameter. The poste­ rior process is small and narrow, measuring only 15.5 mm dor­ soventrally. The great disparity in the size of these processes is exactly the opposite of their dimensions in most Miocene ce­ totheres, in which the anterior process has been drastically re­ duced and the posterior process greatly expanded. Axis Vertebra: The holotype axis vertebra of Micromys­ ticetus rothauseni (ChM PV4844, Figure 13, Table 1) is essen­ tially complete, with only minor erosion of certain surfaces. Like the periotic, the axis vertebra most closely resembles its counterpart in the archaeocete Zygorhiza kochii, as figured by Kellogg (1936:133, fig. 41) from USNM 4679. The transverse processes are extremely short and, like those of the axis in Z. kochii, are "directed more backward than forward" (Kellogg 1936:132). They are extremely narrow anteroposteriorly and thus lack the vertebrarterial canal present in the axis of Zy­ gorhiza. The anterior face of the axis slopes upward and poste­ riorly at a very slight angle. The anterior facets for articulation with the atlas vertebra are flat along their outer portions but curve upward onto the odontoid process. They are separated above the odontoid process by an interval of approximately 40 mm. The odontoid process is broadly rounded ventrolaterally and slopes upward from the centrum to its anteriormost extent, where it abruptly assumes a vertical face 16.5 mm in width that ascends to the level of the dorsal surface of the centrum. On either side of the narrow rectangular vertical face there is a curvilinear excavation extending laterally to the inner margin of the articulating facets. In the upper inner corner of this exca­ vation there is a declivity that notches out the area directly be­ hind the anterior face. Both the dorsal surface of the centrum and the roof of the neural canal slope upward and backward. The neural canal is circular in profile, and in vertical diameter it is 35.3 mm anteriorly and 36.7 mm posteriorly. The posterior face of the centrum is 74.5 mm in transverse diameter and 60.6 mm in vertical diameter and is slightly concave. The epiphysis is well ankylosed to the centrum, but its outer margins have not yet reached the outer margins of the centrum. On the ven­ tral face of the centrum there is a rectangular, tab-like promi­ nence measuring 29.2 mm transversely and 22.4 mm antero­ posteriorly. The neural spine is short, low, and conspicuously thickened at its base, which is anteroposteriorly elongated so that it overhangs the odontoid process, as in Zygorhiza kochii (Kellogg, 1936:133, fig. 41b). Its posterior face is vertically inclined, does not extend beyond the posterior face of the cen­ trum, and bears a medial carina. Dorsally, the neural spine is divided medially by a sulcus that is narrow anteriorly but be­ comes increasingly wider and deeper posteriorly until it has di­ vided the posteriormost portion of the dorsal surface of the spine into two low, knob-like prominences. The pedicles of the neural arch are thinnest anterolaterally and are flattened poste­ riorly below the postzygapophyses, which are inclined ob­ liquely upward and have circular ventral postzygapophysial facets. The punctate surfaces of the anterior articular facets and the failure of the epiphysis to reach the margins of the pos­ terior face indicate that this specimen was not a mature indi­ vidual. PARATYPE.—The paratype partial cranium (ChM PV5933, Figure 14) is missing virtually the entire left side except the condyle and upper portion of the parietal and the anterior mar­ gin of the supraoccipital. All of the basioccipital is missing ex­ cept for the badly eroded remnant of the right basioccipital crest and the posterior portion of the pterygoid. This specimen duplicates all of the characters of the holotype except in one regard. Approximately 37 mm anterior to the dor­ sal margin of the foramen magnum, a small sulcus about 5 mm in transverse diameter begins to traverse the surface of the su­ praoccipital posteriorly along the midline until it reaches a fo­ ramen that penetrates the occipital shield about 13 mm anterior to the dorsal edge of the foramen magnum. This feature is com­ pletely absent in the holotype braincase (obviously that of an older individual); the only trace of it is a small, shallow, almost unnoticeable depression at the midline. DISCUSSION.—The archaeocete-like periotic and axis verte­ bra of Micromysticetus rothauseni suggest that it and M. tobi­ eni are more primitive than the preserved elements of their cra­ nial architecture might indicate. As previously noted, the length of the preserved portion of the intertemporal region in M. rothauseni is almost certainly only half of its original NUMBER 93 285 B FIGURE 13.—Micromysticetus rothauseni, new genus and new species, holotype axis vertebra, ChM PV4844: A, left lateral view; B, anterior view; c, posterior view. (Abbreviations: ns.=neural spine; od.=odontoid process; pz.=postzygapophysis; tr.=transverse process.) length, estimated to be approximately 95 mm, and it seems probable that M. tobieni had an intertemporal constriction of similar length proportionate to its size. An intertemporal region of even greater proportionate length is present in the primitive Oligocene baleen whale family Eomysticetidae (Sanders and Barnes, 2002), a group that firmly establishes Archaeoceti as ancestors of Mysticeti. The striking similarities between the pe­ riotic and axis vertebra of M. rothauseni and those of Zy­ gorhiza kochii provide additional evidence of archaeocete- mysticete relationships and argue strongly in favor of a dorud- ontine archaeocete ancestry of the Mysticeti. TABLE 1.—Measurements (in mm) of the holotype axis vertebra of Micromys­ ticetus rothauseni, new genus and new species (ChM PV4844), from the late Oligocene of South Carolina. (*=measurement is an estimate). Anteroposterior diameter of centrum Transverse diameter of centrum, anteriorly Vertical diameter of centrum, anteriorly Tip of neural spine to ventral face of centrum, anteriorly Greatest vertical diameter of neural canal, anteriorly Greatest transverse diameter of neural canal, anteriorly Greatest distance between outer margins of transverse processes Least anteroposterior diameter of right pedicle of neural arch Greatest transverse diameter of centrum, posteriorly Greatest vertical diameter of centrum, posteriorly 49* 132 66 134 35 42 147 13 76 59 286 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY 5 cm FIGURE 14.—Micromysticetus rothauseni, new genus and new species, paratype braincase, ChM PV5933: A, dorsal view; B, ventral view. Cetotheriopsis Brandt, 1871 Balaenodon Owen, 1846 [in part].—Von Meyer, 1849:550; 1850:201. Stenodon Van Beneden, 1865 [based upon composite assemblage of specimens belonging to one or more unrelated animals]. Cetotheriopsis Brandt, 1871:566.—Brandt, 1873:40, pl. 19: figs. 1-4; 1874: 6-11, pl. 1: figs. 1-16.—Winge, 1910:17-19.—Kellogg, 1923a:22-23; 1931: 307; 1934a:81.—Rothausen, 1971:131, 135, etc.—Whitmore and Sanders, 1976:318.—Fordyce, 1977:265; 1984, fig. 5b. Aulocetus Van Beneden, 1875:537-539. [First latinized use of the name. "Au- locete" was used previously by Van Beneden (1861; see Kellogg, 1923a: 22-23). It is ajunior synonym of Cetotheriopsis Brandt, 1871, because it has as its type species Balaenodon lintianus (see Winge, 1910:17-18; Kellogg, 1923a: 22-23); therefore, both genera, Aulocetus and Cetotheriopsis, have the same type species (see Kellogg, 1923a:22-23; 1931:307).] DIAGNOSIS.—A genus of the subfamily Cetotheriopsinae differing from Micromysticetus in the following characteristics: larger size; cranium with occipital shield shaped like an antero­ posteriorly elongate triangle; exoccipital broad transversely; glenoid fossa elongate anteroposteriorly and with an angular posterointerior margin; sulcus for external acoustic meatus long; zygomatic processes of squamosals not extending beyond apex of supraoccipital. TYPE SPECIES.—Balaenodon lintianus von Meyer, 1849. INCLUDED SPECIES.—Cetotheriopsis lintianus (von Meyer, 1849), late Oligocene, Austria. Cetotheriopsis lintianus (von Meyer, 1849) FIGURES 15-17 Balaenodon lintianus von Meyer, 1849:550.—Von Meyer, 1850:201. "Aulocete" Van Beneden, 1861. Stenodon lintianus Van Beneden, 1865:73-79, text fig. 2 [see Kellogg, 1923a: 22-23]. Cetotheriopsis lintianus Brandt, 1871:196.—Brandt, 1873:40, pl. 19: figs. 1-4; 1874:6-11, pl. Lfigs. 1-16.—Kellogg, 1923a:22-23; 1928:185-187, fig. 18; 1931:307; 1934a:81; 1969:1.—Slijper, 1936:209.—Rothausen, 1971:132.— Whitmore and Sanders, 1977:314.—Fordyce, 1984, fig. 5b [photo of dorsal view of holotype]. Cetotheriopsis linziana Brandt, 1873 [in part, based upon the holotype of Ce­ totheriopsis lintianus and other specimens].—Brandt, 1874. Squalodon linzianus Brandt, 1871. [A tympanic bulla that Brandt (1873) reas­ signed to Squalodon ehrlichi and Abel (1913) referred to Patriocetus ehrli- chi. Aulocetus lintianus Van Beneden, 1861, 1875. [=Balaenodon lintianus; see Winge, 1909:17-18; Kellogg, 1923a:22-23; 1931:307.] Aulocetus linzianum Van Beneden, 1875 [see Kellogg, 1923a:22-23]. HOLOTYPE.—Cranium missing the entire rostrum; possibly associa ted atlas and another ver tebra (see Kellogg, 1923a:22-23). The tooth and tympanic bulla that were origi­ nally included in the type material by von Meyer (1849) were later removed from it (von Meyer, 1850:202; Kellogg, 1923a:22-23); they were thought by Van Beneden (1865) to belong to a squalodont and subsequently were referred to "Squalodon" ehrlichii by Brandt (1873, 1874). Oberosterreich- isches Landesmuseum, Linz/Donau, Austria. TYPE LOCALITY.—The holotype was found in the Sandgrube Bauernberg near Linz, Austria (Rothausen, 1971:140). FORMATION AND AGE.—Linz Sands, upper Chattian, upper­ most Oligocene, nannoplankton zone NP25 (Rabeder and Steininger, 1975:177). EMENDED DIAGNOSIS.—A relatively small mysticete proba­ bly not exceeding 5 m; supraoccipital with well-developed me­ dial crest on anterior half and deep, basin-like depression in posterior half; apex narrow and acute. Crest lying below level of adjacent lambdoidal crests in medial trough extending from depressed center of supraoccipital anteriorly almost to apex; apex elevated slightly above posterior portion of supraoccipital (Figure 16B). In posterior view, vertex of braincase strongly de­ pressed below lambdoidal crests (Kellogg, 1928:187); lamb­ doidal crests prominently developed and overhanging temporal NUMBER 93 287 V y / \ Fr S' }0 "'>\tl \ FIGURE 15.—Cetotheriopsis lintianus (von Meyer, 1849): A, holotype braincase, dorsal view (Oberosterreichisches Landesmu- seum photograph by B. Ecker); B, reconstruction. (Dashed lines indicate hypothetical configurations; abbreviations are explained in "Material and Methods.") B 10 cm 0c fossae. Exoccipitals broadly rounded and sloping posteroven­ trally from posterior margin of squamosal fossae, but outer margins too badly eroded to permit reliable transverse mea­ surement. Occipital condyles massive and with dorsal margins situated well below level of floors of squamosal fossae; squa­ mosal fossae elongate and deeply excavated. Zygomatic pro­ cesses of squamosal roughly parallel to axis of skull midline and not conspicuously directed outward. REMARKS.—The holotype skull is so badly eroded in the squamosal and exoccipital regions that it is difficult to deter­ mine their details. The ventral side of the skull (Figure 16A) has not been prepared, and virtually the entire basicranium is 288 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY Soc 5 cm FIGURE 16.—Cetotheriopsis lintianus (von Meyer, 1849), holotype braincase: A, ventral view; B, posterior view (Oberosterreichisches Landesmuseum photographs by B. Ecker). (Abbreviations are explained in "Material and Methods.") missing. The specimen is covered with fine quartz sand grains like those that have long hindered studies of the odontocetes Patriocetus ehrlichi (Van Beneden, 1865) and Agriocetus in- certus (Brandt, 1874), also from the Linz Sands. Kellogg (1928:187) noted that this species has "elongate zygomatic processes, that do not reach forward to the level of the apex of the supraoccipital shield." These processes are so badly eroded that it is difficult to be certain about their original length, but in view of the greatly elongated supraoccipital and relative length of the zygomatic processes in Eomysticetus (Sanders and Barnes, 2002), it seems highly unlikely that the zygomata of C. lintianus extended beyond the apex of the supraoccipital, if in fact they even reached that point. Probably through deforma­ tion by the weight of overlying sediments, the left squamosal has been twisted out of its proper orientation, the lateral face having been turned upward into the dorsal plane of the skull, thus causing the glenoid fossa to be abnormally visible in lat­ eral aspect (Figure 17). In our reconstruction of this specimen (Figure 15B), we have followed the orientation suggested by the right squamosal and the remains of its zygomatic process; the form and lateral angle of the left squamosal in the actual specimen and the reconstruction are therefore noticeably differ­ ent (Figure 15A,B). Our reconstruction of the portion of the skull anterior to the apex of the supraoccipital is based upon NUMBER 93 289 FIGURE 17.—Cetotheropsis lintianus (von Meyer, 1849), holo­ type braincase, left lateral view (Oberosterreichisches Landesmu- seum photograph by B. Ecker). !&> Oc features manifested in the holotype cranium of the archaic mysticete Eomyticetus whitmorei (Sanders and Barnes, 2002), which has an elongate intertemporal region and a supraoccipi­ tal thrust strongly forward like that of C. lintianus. Discussion Until recently, Micromysticetus tobieni (Rothausen, 1971), from lower Chattian beds near Dusseldorf, Germany (Figures 3-6), and Cetotheriopsis lintianus (von Meyer, 1849), from up­ permost Oligocene sands at Linz, Austria (Figures 15-17), rep­ resented the only toothless baleen-bearing, fossil mysticete specimens reported from Oligocene beds in the Northern Hemisphere. Despite the absence of rostra on these specimens, they have been accepted as true, baleen-bearing mysticetes and have been classified in the family Cetotheriidae Brandt, 1872. The family Cetotheriidae differs from the families Balaenop- teridae, Eschrichtiidae, and Balaenidae by having a skull with an elongate intertemporal region formed by the parietals and a gradual lateral slope from the cranial vertex to the dorsal sur­ face of the supraorbital process. The laterally projecting zygo­ matic processes in all three of the species of the subfamily Ce- totheriopsinae discussed herein suggests the presence in each of a wide supraorbital process and a broad rostrum consistent with the development of baleen and the filter-feeding method used by the true mysticetes. In view of the elongate intertempo­ ral region in all four specimens, their assignment to the family Cetotheriidae corresponds with previous classifications of such animals and is supported by our present, although somewhat limited, knowledge of the group. A comparison of the measurements of the holotype skulls of Micromysticetus tobieni and Cetotheriopsis lintianus with those of Micromysticetus rothauseni (Table 2) reveals three principal points about these three species. First, Micromystice­ tus tobieni and Micromysticetus rothauseni are approximately comparable in size, M. tobieni being slightly larger, and both are smaller than the holotype of Cetotheriopsis lintianus. Sec­ ond, the skull of C. lintianus is the largest of all three of the ho- lotypes. Third, and most important, the anteroposterior lengths of the supraoccipitals separate the three specimens into two sets, one in which the supraoccipital is narrow and greatly elon­ gated (C. lintianus) and the other in which it is considerably shorter anteroposteriorly (M. tobieni and M. rothauseni). The set of curves in Figure 18 graphically demonstrates the morphometric separation between the two groups of speci- o.io c D Micromysticetus tobieni Micromysticetus rothauseni Cetotheriopsis lintianus Ratios from: Distance from dorsal margin of foramen magnum to apex of supraoccipital into Transverse diameter of foramen magnum Distance between lateral margins of occipital condyles Occipital condyles to apex of supraoccpital Distance between lateral margins of exoccipitals FIGURE 18.—Morphometric comparison of the holotype crania of Micromys­ ticetus tobieni (Rothausen, 1971), Micromysticetus rothauseni. new genus and new species, and Cetotheriopsis lintianus (von Meyer, 1849), using ratios derived from five selected cranial measurements. 290 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY TABLE 2.—Measurements (in mm) of the holotype (ChM PV4844) and paratype (ChM PV5933) skulls of Micromys­ ticetus rothauseni, n. gen. n. sp., from the late Oligocene of South Carolina, and of the holotype skulls of Micromys­ ticetus tobieni (Rothausen) (JGU P1289) and Cetotheriopsis lintianus (von Meyer) from the late Oligocene of Europe. Except as noted below, measurements of Af. tobieni and C. lintianus are from Rothausen (1971, table 1). (Measurements in parentheses are estimates; *=measurement from cast of Micromysticetus tob ieni;f=measurement from holotype of Cetotheriopsis lintianus: -=no data.) Character Occipital condyles to plane of anteriormost end of specimen, as preerved Dorsal edge of foramen magnum to apex of supraoccipital Occipital condyles to apex of supraoccipital Occipital condyles to posterior end of vomer Greatest width across zygomatic processes of squamosals Distance between lateral margins of exoccipitals Distance between lateral margins of occipital condyles Greatest distance between lateral margins of basioccipital processes (Alae Basioccipitalis in Rothausen, 1971) Transverse diameter of foramen magnum Vertical diameter of foramen magnum Vertical diameter of right occipital condyle Greatest transverse diameter of right occipital condyle Distance from inner margin of right occipital condyle to outer edge of right exoccipital Maximum length of left zygomatic process, extremity of postglenoid process to anterior end of zygoma Width of temporal fossa Greatest internal height of braincase at midline M. rothauseni M. rothauseni (holotype) 256.4 182.5 224.5 - (398) 256.6 124.4 94.7 44.4 36.3 67 41 109.6 192 122 82 (paratype) 214.2 (156) 200.2 - (370) (230) 115 - (43) - (54) (37) (97) - (103) (70) M. tobieni (P1289) 267 199 247 110 403 278* 126* 116 54 45 78.5* 49* 112 - - C. lintianus 340 2 8 7 | 325 - 505 - 145 - 55.5f 41t 84t 55t - - - - mens. To obtain these ratios, we divided measurement A, the distance from the dorsal margin of the foramen magnum to the apex of the supraoccipital, into four other measurements: B, the transverse diameter of the foramen magnum; C, the distance between the lateral margins of the occipital condyles; D, the distance from the occipital condyles to the apex of the supraoc­ cipital; and E, the distance between the lateral margins of the exoccipitals. Although poor preservation prevents the taking of measurement E in the holotype of Cetotheriopsis lintianus, measurements of B, C, and D place the curve for that species well away from those for Micromysticetus tobieni and Mi­ cromysticetus rothauseni. Although M. rothauseni and M. tobieni are similar in general appearance, there is considerable difference in the shape of the basioccipital crests in these two specimens. In M. tobieni the basioccipital crests have rounded margins, whereas in M. roth­ auseni the margins of these processes are decidedly angular, their inner margins paralleling each other before breaking into divergent angles that terminate in narrow, rounded points pos­ teriorly. Micromysticetus rothauseni thus represents a species that is distinct from M. tobieni, but the variability between other species in polytypic mysticete genera and particularly the presence of squamosal prominences in both forms support their referral to the same genus. Conclusions Two mysticete braincases without rostra, from the late Oli­ gocene Ashley Formation (lower Chattian) near Charleston, South Carolina, represent a new genus and new species herein named Micromysticetus rothauseni. Micromysticetus rothaus­ eni is distinct from the European Cetotheriopsis tobieni at the specific level, but the presence of a previously unrecognized character uniting these two species requires that C. tobieni be reassigned to the new genus Micromysticetus. Micromysticetus tobieni (Rothausen, 1971) and M. rothaus­ eni differ from Cetotheriopsis lintianus (von Meyer, 1849) principally in having an anteroposteriorly shorter supraoccipi­ tal, the apex of which terminates behind the anteriormost extent of the zygomatic process of the squamosal. That feature is in contrast with the condition in the skull of C. lintianus, in which the apex of the supraoccipital seems almost certainly to have reached the level of the anterior tip of the zygoma. Micromys­ ticetus also is characterized by the presence of a knob-like emi­ nence (the squamosal prominence) on the squamosal. Both genera are classified in the family Cetotheriidae. Micromysticetus rothauseni constitutes the oldest western North Atlantic records of the family Cetotheriidae; previous specimens of cetotheres from this region were found in Mi­ ocene sediments (Kellogg, 1924, 1965, 1968a-c, 1969). It fur­ ther documents that this genus of cetotheriids was present on both sides of the North Atlantic during late Oligocene time. In their archaeocete-like morphologies, the periotic and axis ver­ tebra of Micromysticetus rothauseni greatly resemble those structures in the Eocene archaeocete Zygorhiza kochii (Rei­ chenbach, 1847), lending further support to the idea that the an­ cestry of the suborder Mysticeti is tied to the subfamily Dorud- ontinae of the archaeocete family Basilosauridae. NUMBER 93 291 V / / \ /" \ - 'V, • i ' "1 - ' I ^^f.juijoLllJ/jl. lr,, / FIGURE 19.—Reconstructions of crania in dorsal aspect: A, Micromysticetus tobieni (Rothausen, 1971); B, Micromysticetus rothauseni, new genus and new species; and C, Cetotheriopsis lintianus (von Meyer, 1849). (Dashed lines rep­ resent hypothetical configurations; abbreviations are explained in "Material and Methods.") Literature Cited Abel, 0 . 1913. Die Vorfahren der Bartenwale. Denkschriften der Akademie der Wis­ senschaften, Wien, Mathematisch-Naturwissenschaftliche Klasse, 90:155-224, plates 1-12. Bames, L.G. 1977 ("1976"). Outline of Eastern North Pacific Fossil Cetacean Assem­ blages. Systematic Zoology, 25(4):321-343. [Date on title page is 1976; actually published in 1977.] Barnes, L.G., and A.E. Sanders 1990. An Archaic Oligocene Mysticete from South Carolina. Journal of Vertebrate Paleontology, supplement, 9(3): 14A. Beneden, P.J. Van 1861. Un mammifere nouveau du Crag d'Anvers. Bulletin de I'Academie Royale des Sciences de Belgique, series 2, 12:22-28. 1865. Recherches sur les ossements provenant du Crag d'Anvers: Les squalodons. Memoires de I'Academie Royale des Sciences, des Lettres et des Beaux-Arts de Belgique, series 2, 35(3): 1-85, plates 1-4. 1875. Le squelette de la baleine fossile du Musee de Milan. Bulletin de I'Academie Royale des Sciences de Belgique, series 2, 40:736-758. Brandt, J.F. von 1871. Bericht iiber den Fortgang meiner Studien iiber die Cetaceen, 292 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY welche das grosse zur Tertiarzert von Mitteleuropa bis Centralasien hinein ausgedehnte Meeresbecken bevolkerten. Bulletin de I'Ac­ ademie Imperiale des Sciences de St. Petersbourg, series 3, 16:563- 566. 1872. Uber eine neue Classification der Bartenwale (Balaenoidea) mit beriicksichtigung der untergegangenen Gattungen derselben. Bulle­ tin de I'Academie Imperiale des Sciences de St. Petersbourg, series 3, 17:113-124. 1873. Untersuchungen fiber die Fossilen und Subfossilen Cetaceen Euro- pas. Memoires de I'Academie Imperiale des Sciences de St. Peters­ bourg, series 7, 20(1): vii+372 pages, 24 plates. 1874. Ergangzunge zu den fossilen Cetaceen Europas. Memoires de I'Ac­ ademie Imperiale des Sciences de St. Petersbourg, series 7, 21(6): iv + 54 pages, 5 plates. Brisson, M.J. 1762. Regnum animate in classes IX distributum sive synopsis methodica. Revised edition, 296 pages Leiden: Theodorus Haak. Curry, D., C.G. Adams, M.C. Boulter, F.C. Dilley, F.E. Eames, B.M. Funnell, and M.K. Wells 1978. A Correlation of Tertiary Rocks in the British Isles. Geological So­ ciety of London, Special Report, 12:1-72. Emlong, D.R. 1966. A New Archaic Cetacean from the Oligocene of Northwest Oregon. Bulletin of the Museum of Natural History. University of Oregon, 3: 51 pages. Erickson, B.R, and G.S. Sawyer 1996. The Estuarine Crocodile Gavialosuchus carolinensis n. sp. (Cro- codylia: Eusuchia) from the Late Oligocene of South Carolina, North America. Monograph of the Science Museum of Minnesota, Paleontology, 3: 47 pages. Flower, W.H. 1864. Notes on the Skeletons of Whales in the Principal Museums of Hol­ land and Belgium, with Description of Two Species Apparently New to Science. Proceedings of the Zoological Society of London. 1864:384-420. Fordyce, R.E. 1977. The Development of the Circum-Antarctic Current and the Evolu­ tion of the Mysticeti (Mammalia: Cetacea). Palaeogeography, Palaeoclimatology, Palaeoecology, 21:265-271. 1984. Evolution and Zoogeography of Cetaceans in Australia. In M. Ar­ cher and G. Clayton, editors, Vertebrate Zoogeography and Evolu­ tion in Australasia, pages 929-948. Carlisle, Australia: Hesperian Press. Fraser, F.C, and P.E. Purves 1960. Hearing in Cetaceans: Evolution of the Accessory Air Sacs and the Structure and Function of the Outer and Middle Ear in Recent Ceta­ ceans. Bulletin of the British Museum (Natural History), Zoology, 7(1): 1-140 pages, frontispiece, plates 1-53. Gray, J.E. 1868. Synopsis of the Species of Whales and Dolphins in the Collection of the British Museum. 10 pages, 37 plates. London: Bernard Quaritch. Kellogg, R. 1923a. Description of Two Squalodonts Recently Discovered in the Calvert Cliffs, Maryland, and Notes on the Sharktoothed Cetaceans. Pro­ ceedings of the United States National Museum, 62(16): 1-69, plates 1-20. 1923b. Description of an Apparently New Toothed Cetacean from South Carolina. Smithsonian Miscellaneous Collections, 76(7): 7 pages, 2 plates. 1924. Description of a New Genus and Species of Whalebone Whale from the Calvert Cliffs, Maryland. Proceedings of the United States Na­ tional Museum, 63: 14 pages, 6 plates [publication 2482]. 1928. The History of Whales: Their Adaptation to Life in the Water. Quar­ terly Review of Biology, 3:29-76, 174-208. 1929. A New Cetothere from Southern California. University of Southern California Publications, Bulletin of the Department of Geological Sciences, 18(15):449—457. 1931. Pelagic Mammals from the Temblor Formation of the Kern River Region, California. Proceedings of the California Academy of Sci­ ences, series 4, 19(12):217-397. 1934a. The Patagonian Fossil Whalebone Whale, Cetotherium moreni (Ly­ dekker). In Contributions to Paleontology: Marine Mammals. Car­ negie Institution cf Washington Publication, 447:63-81. 1934b. A New Cetothere from the Modelo Formation at Los Angeles, Cali­ fornia. In Contributions to Paleontology: Marine Mammals. Car­ negie Institution of Washington Publication, 447:83-104. 1936. A Review of the Archaeoceti. Carnegie Institution of Washington Publication, 482: xv + 366 pages, 37 plates. 1938. Adaptation of Structure to Function in Whales. Carnegie Institution of Washington Publication, 501:649-682. 1965. Fossil Marine Mammals from the Miocene Calvert Formation of Maryland and Virginia, Part 1: A New Whalebone Whale from the Miocene Calvert Formation. Bulletin of the United States National Museum. 247:1-45, plates 1-21. 1968a. Fossil Marine Mammals from the Miocene Calvert Formation of Maryland and Virginia, Part 5: Miocene Calvert Mysticetes De­ scribed by Cope. Bulletin of the United States National Museum, 247:103-132, plates 46-48. 1968b. Fossil Marine Mammals from the Miocene Calvert Formation of Maryland and Virginia, Part 6: A Hitherto Unrecognized Calvert Cetothere. Bulletin of the United States National Museum, 247: 133-161, plates 49-57. 1968c. Fossil Marine Mammals from the Miocene Calvert Formation of Maryland and Virginia, Part 7: A Sharp-Nosed Cetothere from the Miocene Calvert. Bulletin of the United States National Museum, 247:163-173, plates 58-63. 1969. Cetothere Skeletons from the Miocene Choptank Formation of Maryland and Virginia, Part 1: The Skeleton of a Miocene Choptank Cetothere. Bulletin of the United States National Museum, 294: 1-24, plates 1-15. Martini, E., and C. Muller 1975. Calcareous Nannoplankton from the Type Chattian. Proceedings of the Sixth Congress on Neogene Mediterranean Stratigraphy, Brat­ islava, Yugoslavia, 1975,1:37-41. Meyer, H. von 1849. [Untitled.] Neues Jahrbuch fiir Mineralogie, Geognosie, Geologie, und Petrefaklen-Kunde (Stuttgart), 1849:547-550. 1850. [Untitled.] Neues Jahrbuch fiir Mineralogie, Geognosie, Geologie, undPetrefakten-Kunde (Stuttgart), 1850:195-204. Miller, G.S., Jr. 1923. The Telescoping of the Cetacean Skull. Smithsonian Miscellaneous Collections. 76(5): 1-71, 8 plates. Muller, J. 1849. Uber die fossilen Reste der Zeuglodonten von Nordamerica, mit Rucksicht auf die europdischen Reste aus dieser Familie. iv + 38 pages, 27 plates. Berlin: Verlag von G. Reimer. Owen, R. 1846. A History of British Fossil Mammals and Birds, xlvi + 560 pages, 237 figures. London: John Van Voorst. Packard, E.L., and A.R. Kellogg 1934. A New Cetothere from the Miocene Astoria Formation of Newport, Oregon. In Contributions to Paleontology: Marine Mammals. Car­ negie Institution of Washington Publication, 447:1-62. Perrin, W.F. 1975. Variation of Spotted and Spinner Porpoise (Genus Stenella) in the Eastern Pacific and Hawaii. Bulletin of the Scripps Institute of Oceanography, University of California, 21:1-206. Rabeder, G., and F. Steininger 1975. Die direkten biostratigraphischen korrelationsmoglichkeiten von Saugetierfaunen aus dem Oligo-Miozan der Centralen Paratethys NUMBER 93 293 Proceedings of the Sixth Congress on Mediterranean Neogene Stratigraphy, Bratislava, Yugoslavia. 1:177-183. Reichenbach, H.G.L. 1847. /" C.G. Cams, Resultate geologischger, anatomischer und zoologis- cher Untersuchungen iiber das unter dem Namen Hydrarchos von der A.C. Koch zuerst nach Europa gebrachte und in Dresden aus- gestellte grosse fossile Skellet, pages 1-15, plates 1-7. Dresden and Leipzig. Rothausen, K. 1968. Die systematische Stellung der europaischen Squalodontidae (Od­ ontoceti, Mamm.). Paldontologische Zeitschrift, 42(l/2):83-104, plates 11, 12. 1971. Cetotheriopsis tobieni n. sp., der erste palaogene Bartenwal (Ce­ totheriidae, Mysticeti, Mamm.) nordlich des Tethysraumes. Abhand­ lungen des Hessische Landesamtesn fur Bodenforschung, 60:131- 148, plates 1-3. Russell, L.S. 1968. A New Cetacean from the Oligocene Sooke Formation of Vancou­ ver Island, British Columbia. Canadian Journal of Earth Science, 5:929-933. Sanders, A.E. 1980. Excavation of Oligocene Marine Fossil Beds near Charleston, South Carolina. National Geographic Society Research Reports. 12:601-621. Sanders, A.E., and L.G. Barnes 1989. An Archaic Oligocene Mysticete from South Carolina, U.S.A. Ab­ stracts, Eighth Biennial Conference on the Biology of Marine Mam­ mals, Pacific Grove, California, December. 1989, page 58. 1991. Late Oligocene Cetotheriopsis-Like Mysticetes (Mammalia, Ceta­ cea) from near Charleston, South Carolina. Journal of Vertebrate Paleontology, supplement, 11(3):54A. 2002. Paleontology of the Late Oligocene Ashley and Chandler Bridge Formations of South Carolina, 3: Eomysticetidae, a New Family of Primitive Mysticetes (Mammalia: Cetacea). In R. Emry, editor, Cen­ ozoic Mammals of Land and Sea: Tributes to the Career of Clayton E. Ray. Smithsonian Contributions to Paleobiology, 93:313-356. Sanders, A.E., R.E. Weems, and E.M. Lemon, Jr. 1982. Chandler Bridge Formation: A New Oligocene Stratigraphic Unit in the Lower Coastal Plain of South Carolina. U.S. Geological Survey Bulletin. 1529-H: 105-124, figures 24-27. Slijper, E.J. 1936. Die Cetaceen: Vergleichend-anatomisch und systematisch. Capita Zoologica, 6(1-2): xv + 590 pages. Weems, R.E., and A.E. Sanders 1986. The Chandler Bridge Formation (Upper Oligocene) of the Charles­ ton, South Carolina, Region. In T.L. Neathery, editor, Geological Society of America Centennial Field Guide, Southeastern Section, pages 323-326. Whitmore, F.C, Jr., and A.E. Sanders 1977. Review of the Oligocene Cetacea. Systematic Zoology, 25(4):304-320. Winge, H.A. 1909. Om Plesiocetus og Sqvalodon fra Danmark. Videnskabelige Med- delelser, Dansk Naturhistorisk Foreninq, Kobenhavn, 7(1 ):1—38, plates 1, 2. Balaena ricei, a New Species of Bowhead Whale from the Yorktown Formation (Pliocene) of Hampton, Virginia James W. Westgate and Frank C. Whitmore, Jr. ABSTRACT Fossil species of Balaena have previously been named on the basis of material from the Western Hemisphere, but all are founded upon undiagnostic material. The holotype of B. ricei new species consists of a partial skull, partial mandible, all major flip­ per bones, and representatives of all types of vertebrae, allowing comparison with the two best-known European species. This study, together with the known occurrence of Balaena spe­ cies in the Pliocene of Europe, strengthens the conclusion that bowhead whales were present on both sides of the North Atlantic Ocean at that time. By contrast, remains of Balaena have not been reported from Miocene deposits of the Atlantic coast of North America. Introduction In September 1960 William Rice discovered the skull and skeleton of a large bowhead whale while excavating materials from his borrow pit in Hampton, Virginia (Figure 1). This was the first recorded occurrence of a bowhead whale from the Yorktown Formation (U.S. Geological Survey, 1962, 1965). Another bowhead whale was found in 1985 at the old Hampton Landfill on Magruder Boulevard (G.H. Johnson, pers. comm., 1985). It consists of a set of fused cervical vertebrae, two man­ dibles, the rib cage, six dorsal vertebrae, and both scapulae, ul­ nae, and radii. The specimen, of late Yorktown age, is in the collections of the Department of Geology, College of William and Mary, Williamsburg, Virginia. The family Balaenidae, the bowhead and right whales, has a narrow, highly arched rostrum and fused cervical vertebrae that James W. Westgate, Department of Geology, Lamar University, Beau­ mont, Texas 77710. Frank C. Whitmore, Jr., Research Associate, De­ partment of Paleobiology, National Museum of Natural History, Smithsonian Institution, Washington, D.C. 20560-0121. allow ready identification of the members of the family; both characters are seen in the present specimen. The fossil record of the family goes back to the earliest Miocene (McLeod et al., 1993), but the record is sparse until the late Miocene and the Pliocene, when fossil occurrences indicate widespread, perhaps global, distribution of balaenids. Despite the wide distribution of the family, its generic and specific diversity have always been low. Today the family is represented by two monotypic genera: Eubalaena, the right whale, and Balaena, the bowhead. Several extinct genera of Balaenidae have been described, but only one, Balaenula Van Beneden (1872), is now generally ac­ cepted. The specimen described differs from Eubalaena in having a larger head, whose profile has a continuous curve from the oc­ cipital condyles to the anterior end of the rostrum, and in pos­ sessing a coracoid process on the scapula (True, 1904; McLeod et al., 1993). It differs from Balaenula in its greater size. Many fossil species have been described and assigned to the genus Balaena. Most of these were published in the nineteenth century, and many are based upon undiagnostic material. Of the many nominal species, McLeod et al. (1993) recognized four as valid species of Balaena: B. affinis Owen, 1844; B. etr­ usca Capellini, 1873; B. montalionis Capellini, 1904; and B. primigenias Van Beneden, 1872. In modern times, bowhead whales lived in great numbers along the Atlantic coast of North America. In spite of this, no balaenid remains have yet been found in the Miocene Calvert, Choptank, St. Marys, or Eastover Formations, or in the late Pliocene Chowan River Formation, all of the Chesapeake Group (Table 1). The occurrence of the Rice's Pit bowhead whale in the upper Yorktown Formation, however, establishes proof that bowhead whales inhabited the waters off eastern North America during middle Pliocene times. ACKNOWLEDGMENTS.—Thanks are due to the Research In­ tern Program, Office of Academic Studies, Smithsonian Insti­ tution, which provided funds and related expenses for this 295 296 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY FIGURE 1.—Rice's Pit, Hampton, Virginia. (W.M. Rice photo.) study during the summers of 1972 and 1974; to the late Will­ iam M. Rice for his cooperation in the removal of the specimen from his borrow pit and for his generous donation of the speci­ men to the National Museum of Natural History; to Gerald H. Johnson of the Department of Geology, College of William and Mary, for his time and assistance with writing the manuscript; to David J. Bohaska for his help in preparing the manuscript for publication, and especially for his advice in orienting and labeling the ear ossicles; to Samuel A. McLeod for a helpful re­ view of the manuscript; to the late Robert McKinney of the United States Geological Survey, who photographed all of the bones except the ear ossicles, which were taken by Victor Krantz; to the members of the Department of Paleobiology, Smithsonian Institution, who put up with the inconveniences a specimen of this size creates; and to each person who aided in the tedious job of typing the various stages of the manuscript. Thanks must also be given to Clayton E. Ray of the Depart­ ment of Paleobiology, Smithsonian Institution, who acted as Westgate's advisor during the two summers in which he was a participant in the Research Intern Program. The Yorktown Formation exposed in Rice's Pit is a fully ma­ rine sequence that coarsens and becomes more carbonate-rich upward. At the base of Rice's Pit, the Yorktown is a sparsely fossiliferous, quartzose, clayey, fine sandy silt with less than 20% carbonate. It grades to a moderately fossiliferous, mas­ sive-bedded, fine to medium sand near the top of the formation. The uppermost 20-110 cm (0.6-3.3 ft.) is leached and partially oxidized to a soft, slightly compressive, clayey and silty fine sand. The stratum from which the whale was removed is a medium gray, silty fine sand. The dominant minerals are quartz (56%), aragonite and calcite (41% collectively), and small amounts of glauconite, heavy minerals, and mica. The associated fauna contains a diverse assemblage of bi­ valves, gastropods, and echinoderms. Barnacles, bryozoans, corals, endolithic sponges, and inarticulate brachiopods are common to rare. Foraminifers and ostracodes (Hazel, 1971) are abundant and moderately diverse. The dead right whale came to rest on the floor of a shallow shelf swept by gentle currents (Figure 2). GEOLOGIC SETTING Rice's Pit, which is about 13.2 m (43.5 ft.) deep, is located on the Hampton Flat and is excavated into the Yorktown and overlying Tabb Formations. The Tabb Formation is composed of a fining upward sequence with a basal thin pebble sand that grades upward into a well-sorted, light gray, quartzose sand and into a surficial heterogeneous mixture of clay, silt, and fine sand. It is about 1.7 m (5.6 ft.) thick on the eastern side of the pit. The Tabb was deposited in an ancestral Chesapeake Bay during the late Pleistocene. STRATIGRAPHIC POSITION TABLE 1 The age assignment and stratigraphic framework of the Yorktown Formation has changed with time. Lyell (1845), Rogers (1881), Clark and Miller (1912), Mansfield (1944), and others until the early 1970s assigned a Miocene age to the Yorktown. Subsequently, it was placed in the Pliocene on the basis of the foraminiferal (Akers, 1972; Berggren, 1973), ostra­ code (Hazel, 1971), and molluscan assemblages (Ward and Blackwelder, 1980; Campbell, 1993). The whale came from NUMBER 93 297 . w • > i M . v'jft O < * } * • * V •* +*"• + FIGURE 2.—Mandible of USNM 22553 during excavation in Rice's Pit. (F.C. Whitmore photo.) the Turritella alticostata Zone (Zone 2) of Mansfield (1944), the Orionina vaughani zone of Hazel (1971), the molluscan M- 5 Zone of Blackwelder (1981), and planktonic Zone N19 of Blow (1969) (Akers, 1972). The stratigraphic placement of the whale-bearing unit within the Yorktown is somewhat problematic because of the great lat­ eral variation in thickness and lithology of late Tertiary forma­ tions in southeastern Virginia. These changes resulted from syn- and postdepositional deformation in the terrace zone of the Chesapeake Bay impact structure on the York-James penin- TABLE 1.—Stratigraphic setting of Rice's Pit, Hampton, Virginia. sula (G.H. Johnson, pers. comm., 1997). Rice's Pit is in a gen­ tly tilted, erosionally truncated fault block in the outer terrace zone of the Chesapeake Bay impact structure. The whale-bear­ ing bed is probably equivalent to the middle upper Yorktown: the silty fine sand facies of Johnson (1969, 1976) and the Mor- garts Beach Member of Ward and Blackwelder (1980). The se­ quence of beds in Rice's Pit spans the boundary between the Rushmere and Morgarts Beach Members of the Yorktown For­ mation, according to Campbell (1993). Systematics Balaena ricei, new species TYPE SPECIMEN.—USNM 22553. Right maxilla, vomer, dor­ sal half of cranium, right periotic with auditory ossicles, right and left mandibles both lacking posterior ends, seven cervical vertebrae, three partial thoracic vertebrae, two lumbar verte­ brae, 16 caudal vertebrae, 10 chevron bones (haemal arches), sternum, right scapula, right humerus, right radius, right ulna, four metacarpals, and eight ribs. Collected by Frank C. Whit- 298 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY 20 cm FIGURE 3.—Skull, right posterolateral view. more, Jr., Nicholas Hotton III, Kurt Hausschildt, and party, September 1960. Donated by William M. Rice. ETYMOLOGY.—Named after William M. Rice, owner of the borrow pit from which the type specimen was collected. DIAGNOSIS.—A species of Balaena that differs from B. mys- ticetus in having humerus longer relative to length of forearm, in having transverse processes present on first 11 caudal verte­ brae, in having foramina piercing the transverse processes only in caudal vertebrae posterior to the fourth caudal, having an opening in the atlas shaped like a figure eight—the upper half for the spinal cord and the lower half for the odontoid process 20 cm FIGURE 4.—Cranium, posterior view. of the axis; differs from B. etrusca in having figure-eight- shaped opening in atlas, and in smaller size; differs from B. montalionis in shape of supraoccipital shield. DESCRIPTION (all measurement in centimeters unless other­ wise noted).—Cranium (Figures 3, 4): Distance from dorsal edge of foramen magnum to anterior border of supraoccipital, 67.0; breadth of supraoccipital at temporal fossa, 78.0; dorso­ ventral height of occipital condyle, 25.0; breadth of occipital condyle, 15.5; height of foramen magnum, 13.0; breadth of fo­ ramen magnum, 8.5. The lateral surfaces of the cranium are too broken to allow the delineation of the parietal bones, and the extent of the fron­ tal bone cannot be determined because only the base of the right supraorbital process has been preserved. The supraoccipital extends upward at approximately a 75° angle. Near the apex of the supraoccipital is a wide, laterally sloping medial ridge. Posteroventrally, this ridge is abruptly re­ placed by a 4 cm deep, U-shaped depression that has a promi­ nent lateral ridge on each side. These lateral ridges gradually diminish posteroventrally for 20 cm. At this point the supraoc­ cipital possesses no relief in the midsection. Another prominent ridge, typically found in modern balaenids, forms the lateral border of the supraoccipital along the temporal fossa. This ridge is created by upward curvature of the overhanging border of the occipital shield. Only the base of the squamosal bone is preserved. The 35 cm section present extends anterolaterally from the supraoccipital at a 60° degree angle. The entire ventral surface of the cranium was destroyed by heavy machinery when the operator was re­ moving fill material from the borrow pit. Fortunately, the right periotic and the associated malleus, incus, and stapes became disarticulated from the cranium, allowing close inspection of the periotic and these infrequently recovered auditory ossicles. Maxilla (Figures 3, 5): Length, 224.0; length along curve, 240.0; breadth at nasal region, 20.0; breadth at anterior end, 10.0; height at nasal region, 32.0; height at anterior end, 0.2. NUMBER 93 299 20 cm FIGURE 5.—Right maxilla, ventral view. 20 cm FIGURE 6.—Vomer. The maxilla extends forward from the anterior edge of the supraoccipital region in a well-rounded arc. The maxilla gradu­ ally loses its triangularity, flattening anteriorly into a thin and narrow but ventrally convex projection of bone. The external side is flat, having almost no lateral curvature. Four foramina lie on this side near the junction with the frontal. The dorsal surface of the posterior part of the maxilla exhib­ its a prominent ridge along the external side bordered internally by a flat shelf upon which the premaxilla rests (Figure 3). On the inner margin of this shelf the maxilla curves downward, forming the lateral wall of the vomerine trough. At about one- third the length of the maxilla, the ridge converges with and displaces the shelf. The ridge gradually diminishes and disap­ pears in the anterior mid-region of the maxilla. Therefore, the anterior two-thirds of the maxilla is characterized by a thin, ventrally curved and ventrally sloping surface. Combined with the 75° slope of the occipital shield, the ex­ aggerated curvature of the maxilla creates a very highly arched rostrum. The arch of the rostrum in conjunction with the long jaws forms an immense buccal cavity capable of engulfing huge volumes of marine organisms. Vomer (Figure 6): The vomer is a delicate V-shaped bone having a curvature comparable to that of the maxilla. At the in­ ternal vertex of the "V," the thin sides unite at about half the dorsal height, forming a thickened and sturdy interior. In cross section the vomer resembles an inverted "A," being solid from the crossbar to the apex. Periotic (Figures 7, 8): Viewed from the lateral aspect, the most prominent part of the periotic is the pars cochlearis. The length of the pars cochlearis is approximately twice its width. The surface between the fenestra rotunda and the pterygoid fossa is nearly flat with slight curvature near its edges and two shallow grooves on its face. 1 cm FIGURE 7.—Right periotic, ventral or tympanic view. 300 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY FIGURE 8 (left).—Right periotic, internal or cranial view. FIGURE 9 (below).—Right malleus: a, dorsal view; b, ventral view; c, medial view; d, lateral view; e, anterior view; / posterior view. (ant.p.=broken base of anterior process; he.f.=hemicircular facet; ho.f.=horizontal facet; p.mus.=pro- cessus muscularis.) 1 cm ant. p. ant. p. ant. p p. mus. ant. p. he. f. 1 cm NUMBER 93 301 f. ho. f. f. he. f. c. b f. he. f c. b. f. s. f. ho. f. 1 cm FIGURE 10.—Right incus: a, view showing facet for horizontal facet of malleus, and cms longum; b, view showing facet for hemicircular facet of malleus; c, view showing articular facets for malleus; d, view showing articular facet for stapes. (c.l.=crus longum; c.b.=crus breve (broken); f.he.f.=facet for hemicircular facet of malleus; f.ho.f.=facet for horizontal facet of malleus; f.s. =facet for stapes.) The groove for the facial nerve and the fallopian aqueduct forms a wide channel that begins at the base of the posterior process and curves forward below and just beyond the fenestra ovalis. The fenestra ovalis is blocked by the foot plate of the stapes, which cannot be removed without damage to it. The an­ terior process is nearly flat, with very slight grooves radiating from its base. Only the base of the posterior process remains. Malleus (Figure 9): Length, 19 mm; width, 16 mm. The delicate anterior process is broken, but its structure is still dis­ cernible through inspection of the cross section of the process at the point of fracture. It is apparent that the anterior process had an interior flat-bottomed groove and a rounded exterior. Posteriorly, the minutely convex hemicircular facet and the sig- FlGURE 11 (right).—Right stapes: a, medial view; b, ventral view, showing articulation for incus. i-M b Wi& 5 mm nificantly smaller horizontal facet, which join at nearly a right angle, form the contact for the incus. Incus (Figure 10): Length, 12 mm; width, 9 mm. The two facets of the incus that are adjacent to the malleus are both con­ cave, the smaller much more so than the other. The horseshoe- shaped larger facet meets the smaller facet at an acute angle ad­ jacent to the cleft of the larger facet. The crus breve is missing, but the crus longum is complete. The crus longum is thin and short. At its end is a small circular facet that serves as the point of contact between the incus and the stapes. Stapes (Figure 11): The facet on the head of the stapes and adjacent to the incus is larger and more nearly oval than that of the crus longum of the incus. The foot plate of the stapes can­ not be removed from the fenestra ovalis of the periotic. The midsection of the stapes was destroyed with the breaking of the stapes head from the main body of the stapes. Mandible: Length (partial), 310.0; length along curve (par­ tial), 340.0; height (anterior), 35.5; height (posterior), 10.0; width (posterior), 20; width (anterior), 11. The posterior sections of the mandibles are absent (Figures 12-14). Although the most diagnostic part of the mandible is missing, the remaining part can be informative. The following section refers to the right mandible only Throughout most of its preserved three meters, the mandible projects in a straight line, curving inward only in the most ante­ rior 60 cm. At the distal end, the mandible twists inward and slightly upward. The residual dental groove extends from ap- iU 20 cm FIGURE 12.—Right mandible, medial view. 302 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY - ^ -^,-nnr.:: 4 ^ J : ' 20 cm FIGURE 13.—Right mandible (anterior end). proximately the middle to the distal end of the preserved por­ tion of the mandible, immediately below and internal to the narrow, rounded dorsal edge. From a width of 1 cm at the pre­ served proximal end, this groove rapidly widens to 4 cm at the anterior extremity of the jaw. The depth of the dental groove correspondingly increases from 3 cm to 5 cm. At the most proximal point, the dental groove is positioned at the midline of the mandible and is a wide notch separating two terminal projections of bone. From the distal end of the mandible extending posteriorly along the internal face, 5 cm above the lower edge of the jaw is a broad swollen ridge, 1 cm high and 35 cm long. On the exter­ nal side near the dorsal surface of the mandible are five large foramina spaced 16 to 25 cm apart along the length of the jaw. The most anterior foramen is the largest. It is the posterior end of an anteriorly directed canal that is 25 cm long, 4 cm wide, and 2 cm deep. The foramen itself is round, 1 cm in diameter; it lies 110 cm from the distal end of the mandible (measuring along the external curvature). The height and width of the mandible increase greatly in the posterior direction. In cross section, the anterior end is small in comparison to the posterior. Cervical Vertebrae (Figures 15-18): Transverse breadth of atlas, 54.0; vertical height of atlas, 38.0; transverse breadth of articular surface of atlas, 39.0; transverse breadth of articular surface (right facet), 16.0; vertical height of articular surface (right facet), 27.0; depth of articular concavity, 9.0; transverse breadth of spinal canal (anteriorly), 10.5; transverse breadth of spinal canal (posteriorly), 20.0; vertical diameter of spinal ca­ nal (anteriorly), 9.5; vertical diameter of spinal canal (posteri­ orly), 10.0; ventral length of atlas, 12.5; ventral length of atlas with axis, 16.0; ventral length of cervical region, 26.0. The seven cervical vertebrae are fused together, forming one compact mass. The fusion of the cervicals and the pancake-like thinness of the five posterior cervicals, both of which allow a shortening and strengthening of the neck, are characteristic of all of the balaenids. The development of this condition may be an adaptation to accommodate the intensified amount of water resistance and skull weight caused by the extreme arching of the rostrum and the heaviness of the large mandibles. No other group of mysticetes possesses a cervical region in which all or most of the vertebrae are fused, although fusion of cervical vertebrae to a lesser extent is a common feature of many odontocetes (Eschricht and Reinhardt, 1866:107). All of the dorsal transverse processes on the cervical verte­ brae are broken. The left halves of the last five neural arches and the right half of the last neural arch in the cervical region also are broken. The remaining portions of neural arches on the 20 cm FIGURE 14.—Dorsal view of anterior end of right mandible. FIGURE 15.—Cervical vertebrae, anterior view. NUMBER 93 303 10 cm FIGURE 16.—Cervical vertebrae, left lateral view. last six cervicals are thin, wide, dorsolateral^ oriented projec­ tions. They unite in an osseous mass that forms the roof of the spinal canal, which in turn is surmounted by a spinal ridge con­ sisting of the dorsal processes of these neural arches and the os­ seous growth uniting them. Of the processes on the atlas, only the dorsal process is fused to the axis. This leaves a wide gap between the atlas and axis on the lateral borders of the neural canal. On the axis and third cervical vertebra, both ventral trans­ verse processes are present but broken. The base of the left ventral transverse process remains on the fourth cervical and that of the right ventral transverse process on the fifth cervical. Other ventral transverse processes may once have been present, but because the surfaces of several centra are broken, their traces are absent. The nearly flat dorsal surface of the fused centra, which is the floor of the spinal canal, is interrupted by transverse grooves that represent the lines of fusion between the verte­ brae. The spinal canal changes from a subcircular opening at the atlas to an oval that is nearly twice as wide as high at the third cervical. The second through the sixth cervical vertebrae are ventrally covered by a wide, smooth bone growth, which forms a low ridge down the midline and originates at the anterior edge of the axis (Figure 17). The lateral edges of this growth extend al­ most to the bases of the inferior transverse processes, where the individual vertebrae may be distinguished. This growth termi­ nates posterior to the atlas, where a distinct transverse groove separates the ventral surface of the atlas from that of the axis. The atlas is the largest of the seven cervical vertebrae. Its length alone constitutes almost half the length of the cervical region and, combined with the axis, forms the anterior two- thirds of the cervical region. The opening in the atlas for the spinal cord and for the odontoid process of the axis is in the shape of an inverted figure eight having a slightly constricted midsection (Figure 15). Just posterior to the dorsal margin of each articular facet of the atlas lies a bridged-over transverse arterial canal (Eschricht and Reinhardt, 1866:108). These ca­ nals are 2 cm wide and nearly 2 cm high, and they are bridged 10 cm FIGURE 17.—Cervical vertebrae, ventral view. 304 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY FIGURE 18.—Cervical vertebrae, posterior view. over for a transverse distance of 5 cm on the right canal and 6 cm on the left (Figure 16). A large posteriorly directed pit 3 cm in diameter lies at the base of the dorsal process of the atlas (Figure 15). This large cavity may have been the site of carti­ lage that did not ossify. The transverse processes of the atlas are thick and broad, but their length cannot be ascertained be­ cause the ends are broken. The dorsal transverse processes of the axis are much more delicate than those of the atlas (Figure 16). These axial pro­ cesses are almost as wide vertically but are about one-third as thick as those of the atlas. The length is indeterminable because both processes are broken. The ventral transverse processes of the axis are approximately 16 cm long, but their ends are worn and may not be complete. An upward curvature at the ends of the ventral transverse processes and a corresponding down­ ward curvature of the dorsal transverse processes form an in­ complete ring through which the vertebral artery passed (Esch- richt and Reinhardt, 1866:108). The third through seventh cervicals are almost identical in shape and dimension. All have extremely delicate transverse processes except at the point where they meet the osseous mass that is the roof of the spinal canal. The centra of these cervicals are thin and flat; each is about 1 cm thick and 16 cm high, al­ though the seventh is thicker than its predecessors. Thoracic Vertebrae: Three badly weathered thoracic verte­ brae are present. None of these vertebrae possess fused epiphy­ sial disks, and all are too eroded to allow accurate measure­ ments or diagnosis. Lumbar Vertebrae (Figures 19, 20): Only two lumbar ver­ tebrae have been preserved. Because their exact position in the vertebral column is uncertain, the more anterior vertebra has been designated lumbar-A and the more posterior is lumbar-B. Vertical diameter of centrum (anterior), (A) 21.5, (B) 23.5; transverse diameter of centrum (anterior), (A) 24.0, (B) 26.0; anteroposterior length of centrum, (A) 15.5, (B) 17.5; height to top of dorsal process, (A) 57+, (B) 55+; transverse diameter of neural canal (anterior), (A) 12.0, (B) 9.0; vertical diameter of neural canal (anterior), (A) 7.0, (B) 5.0; transverse length of transverse processes, (A) 33+, (B) 26+. The transverse processes of lumbar-A are longer than twice the anteroposterior length of the centrum, and they are directed slightly anteriorly. The neural spine of the vertebra is broken but is 27 cm in height and is directed posterodorsally at about a 70° angle. Neither vertebral epiphysis has fused to the centrum of lumbar-A. The dimensions of lumbar-B are similar to those of lumbar- A, but the processes and neural canal are smaller, although the centrum is larger in diameter. The anterior epiphysis may have fused to the centrum of lumbar-B, but this is uncertain because FIGURE 19.—Lumbar vertebra, anterior view. NUMBER 93 305 FIGURE 20 (left).—Lumbar vertebra, right lateral view. FIGURE 21 (right).—Fifth largest caudal vertebra, right lateral view. a distinct suture exists in the entire circumference between the two. Caudal Vertebrae (Figures 21, 22, Table 2): Measurements of the first caudal vertebra (see Table 2 for others): Vertical di­ ameter of centrum (anterior), 24.0; transverse diameter of cen­ trum (anterior), 26.0; anteroposterior length of centrum, 20.0; height to top of dorsal process, 55+; transverse diameter of neural canal, 8.5; vertical diameter of neural canal (anterior), 5.0; transverse length of transverse processes, 24+. The first caudal vertebra is herein defined as the most ante­ rior vertebra possessing a contact with a chevron bone. Sixteen caudal vertebrae are present and range in length from 20 cm to 6 cm. The most anterior caudal vertebrae resemble the poste­ rior lumbar vertebra, but these caudals have larger centra, shorter processes, and neural spines that are more vertically di­ rected than the posterodorsally slanted spine of the posterior lumbar. Posteriorly, the vertebral processes diminish rapidly in size so that on the last eight caudal vertebrae present, no trans­ verse processes exist at all, and the 11th caudal present is the last vertebra possessing a neural arch. The first foramina in the transverse processes appear in the sixth caudal vertebra. The foramen on the right transverse process of this vertebra mea­ sures more than twice the diameter of the foramen on the left 10 cm transverse process. On the ventral surface of the caudals, the articular facets for the haemal arches turn increasingly inward on the more posterior vertebrae. By the ninth caudal these fac­ ets join to make two foramina, each 3 cm in diameter. Between these ventral foramina and those of the transverse processes lies a broad, ventrally directed groove on the surface of the cen­ trum. A similar groove extends from the transverse process to the reduced neural arch, where another foramen has formed al­ lowing passage transversely into the neural canal through the sides of the neural arch. Posteriorly, the grooves on the centra gradually become bridged over by bone, and the foramina lengthen into tunnel-like structures. In the 13th caudal present, which is devoid of processes, the tunnel structures are intercon- TABLE 2.—Caudal vertebraeof Balaena ricei (measurements in cm). Measurement Vertical diameter of centrum (anterior) Transverse diameter of centrum (anterior) Anteroposterior length of centrum Height to top of dorasl process Transverse length of transverse length 2 24.5 27.0 18.0 54.0 22.5 3 25.0 27.5 22.0 53+ 20.5 4 26.0 28.0 21.5 52.0 17+ 5 26.0 28.0 19.5 49+ 12.5 6 26.0 29.0 20.0 45+ 10.0 Caudal i 7 8 26.0 28.0 19.5 43+ 6.0 27.0 29.5 20.0 41 + 5+ vertebral number 9 10 11 26.5 28.0 19.0 39.0 2+ 26.0 27.0 18.5 35.0 - 25.5 25.5 18.0 33.5 - 12 24.5 28.0 17.5 29+ - 13 19.0 21.0 12.5 21.0 - 14 19.0 18.0 10.5 17.0 - 15 11.5 12.5 8.0 12.0 - 16 9.0 9.5 6.0 9.0 - 306 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY FIGURE 22 (left).—Eighth largest caudal vertebra, right lateral view. FIGURE 23 (right).—Largest chevron bone, anterior view. 10 cm nected, completely encircling the centrum, and are exposed ex­ teriorly only through five moderately sized foramina. Haemal Arches (Figure 23): The chevrons decrease in height from 24 cm in the anteriormost to 10 cm in the posteri­ ormost. Only slight morphological variation exists between the individual chevrons. The intermediate chevrons gradually de­ crease in size posteriorly, and the smallest have not fused at their vertices. The largest chevrons possess a ventral process but the smallest consist of only an arch. Sternum: Anteroposterior length, 30.0; transverse breadth, 23.0; thickness, 5.0. Sterna of Balaenidae are characteristically variable (True, 1904). The sternum of Balaena ricei is shaped like an elongated, flat shield (Figure 24). The external surface is convex transversely, giving the appearance of having a low ridge along the midline. Longitudinally, the internal surface is nearly flat with a broad depression at the anterior end. Collec­ tively, these features produce an anteroposterior cross section that is relatively thin anteriorly and posteriorly but thick in the midsection. On the lateral edges of this thickened section is a rough surface indicative of the presence of cartilage. This sur­ face served as the point of articulation for the first pair of ribs. Ribs (Figure 25): Only one rib remains complete. This also is the only rib with an articulating tubercular process, and it probably belongs to the first pair. The tubercular process is 12 cm long with a flat end indicative of a point of articulation. The distance from the tubercular process to the distal end is 109 cm, 10 cm and the external curvature is about 170 cm. The distal end is nearly flat, 13 cm wide, and 4 cm thick. The most distal surface is flat to allow articulation with the sternum. Scapula: Height (glenoid fossa to middle of convex mar­ gin), 69.0; greatest anteroposterior length, 86.0; smallest an­ teroposterior length (at neck), 23.0; glenoid fossa (anteroposte­ rior length), 25.0; glenoid fossa (transverse breadth), 22.0; acromion length (ventral surface), 19.0; coracoid length (dorsal surface), 11 .0. The presence of the prominent coracoid process on the scap­ ula (Figures 26, 32) of Balaena ricei is similar to that of the scapula of B. mysticetus and unlike that of Eubalaena glacialis, which lacks the coracoid process (True, 1904). The 7 cm high coracoid process of B. ricei extends anteriorly for 10 cm with a slight upward and outward orientation. This process is stout, is ' 10 cm FIGURE 24.—Sternum, dorsal view. NUMBER 93 307 FIGURE 25 (left).—Right first rib, posterior view. FIGURE 26 (right).—Right scapula, lateral view. ovate in cross section, and has a width more than one-half its height. The acromion process is larger than the coracoid process, but proportionately it is not as sturdy. The acromion curves inter- noventrally for 21 cm from its origin above the coracoid pro­ cess to the acromion's anterior end. At this point the acromion is parallel to the coracoid process. The spina scapulae extends as a sharp ridge from the dorsal midsection of the acromion process to the external anterior edge of the scapula, meeting it at mid-height. The anterior edge of the scapula is 4 cm wide and shallowly convex at this point. The anteroposterior length of the scapula is 20% greater than its height. The breadth of the neck is about 27% of the maxi­ mum anteroposterior length of the scapula. The neck of the scapula of B. ricei is thus more robust than that of B. mystice- tus, which possesses a scapular neck having a breadth only about 2 1 % of the convex margin (Eschricht and Reinhardt, 1866). The posterior edge of the scapula is rounded and about 3 cm thick. The glenoid fossa is elliptical in circumference and shal­ lowly concave at the center where the head of the humerus ar­ ticulates with it. Humerus (Figures 27, 28, 32): Length, 48.5; greatest trans­ verse diameter (proximal), 21.5; anteroposterior diameter (proximal), 28.0; greatest transverse diameter (distal), 13.0; an­ teroposterior diameter (distal), 23.5; least circumference of shaft, 49.0. The humerus is similar in design and structure to the humeri of the other mysticetes. The head of the humerus is large and rounded, tilting outward and backward to allow the maximum amount of freedom from restraint at the point of contact with the glenoid fossa of the scapula. There is a shallow groove be­ tween the head and the shaft, indicating that fusion of the prox­ imal epiphysis was not yet complete in this specimen. The mid­ section of the shaft is straight and has only slight lateral curvature, whereas the anterior and posterior sides of the shaft are rounded. At its distal end, the humerus flares out, forming a broad point of contact for the radius and ulna. At this contact, the base of the humerus is V-shaped. The V forms a 130° angle, and at the vertex the humerus is 13 cm thick, transversely. An­ other V-shaped angle of articulation is situated at the posterior edge of the point of articulation between the ulna and the hu­ merus. This is the contact point for the olecranon process of the ulna where it abuts the posterior edge of the distal end of the humerus. This surface forms a 140° angle with the contact for the main proximal articular face of the ulna. Radius (Figures 29, 32): Length, 52.0; transverse diameter (proximal), 12.0; anteroposterior diameter (proximal), 19.0; transverse diameter (distal), 10.5; anteroposterior diameter (distal), 22.5; least anteroposterior diameter of shaft, 17.5. The radius is flatter and slightly longer than the humerus. The proximal end of the radius is thicker transversely and nar­ rower anteroposteriorly than the distal end, which is flat and widened anteroposteriorly to accommodate the carpals. The distal ends of the ulna and radius form a broader articular sur­ face for the carpals (45 cm) than the distal end of the humerus, which is only 26 cm wide. The anterior edge of the radius curves outward anteriorly, widening the bone considerably and flattening the edge into a comparatively thin border. The poste­ rior edge is concave, creating a 6 cm gap internally between the midsections of the radius and ulna when these bones are articu­ lated. Ulna (Figures 30-32): Length, 49.0; transverse diameter (proximal), 10.0; anteroposterior diameter (proximal), 16.0; transverse diameter (distal), 7.0; anteroposterior diameter (dis­ tal), 21.0; least anteroposterior diameter of shaft, 9.5. The ulna is lacking its distal epiphysis but is only slightly shorter than the radius, which is complete. At the proximal end of the ulna, the olecranon process also is lacking. This process 308 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY I :•?• •i I- ..' ' ' :^J %i*4 r 10 cm FIGURE 27.—Right humerus, lateral view. probably had not yet fused to the main body of the ulna, but it also may not have ossified before the animal died. When the olecranon process is present, the ulna is nearly immobilized in its contact with the humerus and radius. This locking fit gives the flipper rigidity against the resistant water pressures met during swimming. As in the radius, the distal end of the ulna is flared, providing a wide carpal area of contact and creating a broader external surface toward the extremity of the flipper. The anterior edge of the ulna extends from the distal flare in a nearly straight diagonal line to the base of the proximal contact against the radius. The posterior edge is curved in a well- rounded, anteriorly directed arc. Metacarpals (Figure 33): The only bones of the manus present are the four largest metacarpals. The metacarpals are basically similar in shape (Table 3), and size is the principal characteristic distinguishing them. Each metacarpal resembles 10 cm FIGURE 28.—Right humerus, medial view. a flattened cylinder with a flared proximal end. All ends have the rough porous appearance that cartilage-covered bone pos­ sesses. DISCUSSION.—This specimen closely resembles Balaena mysticetus but on a smaller scale. Some differences are notable, but, as many of the most advanced characteristics of B. mys­ ticetus (such as the lack of a coronoid process on the mandible) have not been preserved in this specimen, a complete compari­ son is prevented. The mysticete flipper illustrates another struc- TABLE 3.—Measurements of metacarpals (in cm). Metacarpal 2nd 3rd 4th 5th Length 10.5 13.5 9.5 7.5 Proximal w 9.0 10.0 10.0 8.5 idth Distal wi 7.0 8.0 7.0 5.0 dth Proximal thickness 5.5 6.0 5.0 4.0 NUMBER 93 309 • V FIGURE 29 (left).—Right radius, lateral view. FIGURE 30 (right).—Right ulna, medial view. 10 cm 10 cm ture that has undergone extreme modification. The lengthening of the radius and ulna in comparison with the humerus is evi­ denced in most of the genera of whalebone whales, including Balaena. The relationship of the length of the forearm to that of the humerus in B. mysticetus is approximately 3 to 2, whereas in B. ricei it is about 1 to 1. The humerus of B. mysticetus has been described as "very short and thick, almost as broad as long (as 4 to 5)" (Eschricht and Reinhardt, 1866:129). That of B. ricei is much more elongate and narrow, the thick proximal end being only a little more than one-half as wide as the bone is long, a width-to-length ratio of about 1 to 2. In the caudal re­ gion more distinctions may be noted between B. mysticetus and B. ricei. In B. mysticetus the anterior caudals have very dimin­ ished traverse processes, and foramina for the branches of the aorta may be present on the transverse processes of the second caudal (Eschricht and Reinhardt, 1866:124). In B. ricei, how­ ever, the transverse processes are not completely absent until the 12th caudal, and foramina on the transverse processes do not appear before the fifth caudal. A major point of distinction between B. mysticetus and B. ricei may be observed in compar­ ing the atlas vertebrae. The atlas of B. ricei possesses a figure- eight-shaped opening, the upper one-half for the spinal cord and the lower half for the odontoid process of the axis (Figure 15). From an anterior view, therefore, the odontoid process is clearly visible. The atlas of B. mysticetus possesses no opening for the odontoid process of the axis, and from the anterior view the only visible portion of the centrum of the axis is that which is incorporated in the floor of the neural canal. The same is true of B. etrusca Capellini, 1873, which is much larger than B. ri­ cei: its fused cervical vertebrae measure 260 mm anteroposteri­ orly as opposed to 166 mm for B. ricei. Balaena mysticetoides Emmons, 1858, was based upon a tympanic bulla from the Yorktown Formation of North Caro- 310 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY FIGURE 31 (left).—Right ulna, lateral view. FIGURE 32 (upper right).—Right flipper, lacking carpals and metacarpals. FIGURE 33 (lower right).—Third right metacarpal. 10 cm 10 cm Una (McLeod et al., 1993). The holotype specimen has been lost, but the figure presented by Emmons (1858, fig. 26) is of a nondiagnostic tympanic bulla of an indeterminate mysticete. Balaena svedenborgii Lilljeborg, 1867, possesses a distinctly different caudal section from that of B. ricei. According to Lill­ jeborg, foramina do not appear in the transverse processes until the 10th caudal (fifth in B. ricei), and the lower channel for the branches of the aorta at the point of attachment for the haemal arches does not become a closed canal until the 15th caudal (eighth in B. ricei; Figure 22). McLeod et al. (1993:52) synon­ ymized B. svedenborgii with B. mysticetus. Balaena primigenia Van Beneden, 1872, is based upon only a posterior sphenoid, several tympanic bullae, a complete rib, and a phalanx. This species thus has no diagnostic parts com­ parable with B. ricei. In 1852, Leidy described B. palaeatlantica and B. prisca. Balaena palaeatlantica, from the Miocene at City Point on the James River in Virginia, is based upon an 8V2 in. (21.6 cm) fragment of mandible, some vertebrae, and a zygomatic pro­ cess. (The term Miocene as used by Leidy in 1852 includes beds that we would now designate as Pliocene.) The transverse measurements of the mandibular fragment are 2 in. (5.2 cm) anteriorly and 2V* in. (7.0 cm) posteriorly. The height is VA in. (9.5 cm) anteriorly and 4/4 in. (11.4 cm) posteriorly. Balaena prisca is based upon a mandibular fragment 14 in. (35.5 cm) long. The vertical diameter is 3 in. (7.6 cm) and the transverse diameter, 2 in. (5.0 cm). Kellogg (1968) regarded this specimen as undiagnostic. These mandibular fragments represent species much smaller than B. ricei, and the holotype of B. prisca was considered by both Cope and Kellogg (Kellogg, 1968:125) to be from a cetothere. Balaena simpsoni Philippi, 1887, is based upon a badly bro­ ken occipital shield, which Philippi admitted could not be accu­ rately diagnosed. It represents an adult estimated to be 20-25 ft. (6-7.5 m) long. Donoso-Barros (1976) placed this species in the genus Caperea. Fordyce (1984) regarded it as undiagnostic. Balaena affinis Owen, 1844, B. definita Owen, 1844, and B. pampaea Ameghino, 1891, are based upon tympanic bullae NUMBER 93 311 TABLE 4.—Measurements of the supraoccipital shield of Balaena montalionis and B. ricei (in cm). * Measurements from Capellini (1904). + This measurement erroneously given by Capellini as 0.066 m. Measurement From superior margin of foramen magnum to apex of occipital shield Width at level of superior margin of foramen magnum Width 150 mm anterior to foramen magnum Width 300 mm anterior to foramen magnum Width 400 mm anterior to foramen magnum Balaena montalionis* 53.0 66.0t 52.5 43.0 30.5 Balaena ricei 67.0 78.0 71.0 67.5 56.6 and thus cannot be compared with B. ricei. Balaena definita is based upon a balaenopterid bulla (McLeod et al., 1993). Balaena montalionis Capellini, 1904, from the Pliocene of Tuscany is based upon a partial skull including a complete oc­ cipital shield, squamosals, left supraorbital process of the fron­ tal, nasals, partial maxilla, basioccipital, and basisphenoid (Capellini, 1904, pl. 1). Capellini (1904) stated that the holo­ type of B. montalionis is too incomplete to allow comparison with other fossil balaenids of Tuscany, but that it differs mark­ edly from B. mysticetus in the conformation of the supraoccipi­ tal. In this respect B. montalionis also differs from B. ricei (Ta­ ble 4), a difference sufficient to differentiate the two species. Conclusions The cosmopolitan occurrence of Balaena in Pliocene time was part of the appearance of the modern whale fauna, domi­ nated by large baleen whales, sperm whales, and the many gen­ era of the family Delphinidae (Fordyce, 1989; Whitmore, 1994). This faunal change has been attributed to changes in ocean currents resulting from the growth of the West Antarctic ice sheet, the cutting off of warm water flowing into the Atlan­ tic from the Mediterranean, and the rise of the Isthmus of Pan­ ama. Late Miocene and Pliocene fossils of Balaena are associated in Europe and North America with remains of Balaenula, a smaller genus now extinct (Van Beneden, 1872; Whitmore and Kaltenbach, MS). The relationship between these two genera is on tenuous ground because of the paucity of good fossil mate­ rial, as is the question of the number of species of Balaena that existed during the Pliocene. This number may be reduced when we know more about size range and intraspecific variation in Pliocene Balaena; but, considering the number of species of Balaenoptera in modern seas, five species of Pliocene Balaena may not be an unreasonable number. Literature Cited Akers, W.H. 1972. Planktonic Foraminifera and Biostratigraphy of Some Neogene For­ mations, Northern Florida and Atlantic Coastal Plain. Tulane Stud­ ies in Geology and Paleontology, 9: 139 pages. Ameghino, F. 1891. Caracteres diagnosticos de cincuenta especies nuevas de mamiferos fosiles argentinos. Revista Argentina de Historia Natural (Buenos Aires), l(3a): 129-167, figs. 26-75. Berggren, W.A. 1973. The Pliocene Time Scale: Calibration of Planktonic Foraminiferal and Calcareous Nannofossil Zones. Nature. 243:391-397. Blackwelder, B.W. 1981. Late Cenozoic Stages and Molluscan Zones of the United States Middle Atlantic Coastal Plain. Paleontological Society Memoir, 2: 1-34. Blow, W.H. 1969. Late Middle Eocene to Recent Planktonic Foraminiferal Biostratig­ raphy. In P. Bronnimann and H.H. Renz, editors, Proceedings of the First International Conference on Planktonic Microfossils, Geneva, 1967, pages 199—422, plates 1-54. Leiden: E.J. Brill. Campbell, L.D. 1993. Pliocene Molluscs from the Yorktown and Chowan River Forma­ tions in Virginia. Publications of the Virginia Division of Mineral Resources. 127: 259 pages, 43 plates. Capellini, G. 1873. Sulla Balaena etrusca. Memorie dell'Accademia delle Scienze dell 'Instituto di Bologna, series 3a, 3:313-331. 1904. Balene Fosili Toscane, II: Balaena montalionis. Memorie dell'Acca­ demia delle Scienze dell 'Instituto di Bologna, series 6, 1:47-54. Clark, W.B., and B.L. Miller 1912. The Physiography and Geology of the Coastal Plain Province of Virginia. Bulletin of the Virginia Geological Survey, 4: 274 pages. Donoso-Barros, R. 1976 ("1975"). Contribucion al conocimiento de los Cetaceos vivientes y fosiles del territorio de Chile. Guyana. Zoologia, 36:1-127. [Date on title page is 1975; actually published in 1976.] Emmons, E. 1858. Report of the North Carolina Geological Survey: Agriculture of the Eastern Counties, Together with a Description of the Fossils of the Marl Beds. 314 pages. Eschricht, D.F, and J. Reinhardt 1866. On the Greenland Right Whale, (Balaena mysticetus Linnaeus). [Pub­ lications of the] Ray Society, London, 1866: 150 pages, 106 plates. Fordyce, R.E. 1984. Evolution and Zoogeography of Cetaceans in Australia. In M. Ar­ cher and C. Clayton, editors, Vertebrate Zoogeography and Evolu­ tion in Australia, pages 929-948 . Perth, Western Australia: Hesperian Press. 312 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY 1989. Origins and Evolution of Antarctic Marine Mammals. In J.A. Crame, editor, Origins and Evolution of the Antarctic Biota. Geo­ logical Society (London), Special Publication, 47:269-281. Hazel, J.E. 1971. Ostracode Biostratigraphy of the Yorktown Formation (Upper Mi­ ocene and Lower Pliocene) of Virginia and North Carolina. U.S. Geological Survey Professional Paper, 704: 13 pages. Johnson, G.H. 1969. Guidebook to the Geology of the Lower York-James Peninsula and South Bank of the James River. Guidebook, 1: 33 pages. Williams­ burg, Virginia: Department of Geology, College of William and Mary. 1976. Geology of the Mulberry Island, Newport News North and Hamp­ ton Quadrangles. Virginia Division of Mineral Resources Report of Investigations, 41: 72 pages. Kellogg, R. 1968. Fossil Marine Mammals from the Miocene Calvert Formation of Maryland and Virginia, Part 5: Miocene Calvert Mysticetes De­ scribed by Cope. Bulletin of the United States National Museum, 247:103-132, plates 4 6 ^ 8 . Leidy, J. 1852 ("1851"). [Description of Balaena prisca and B. palaeatlantica.] Proceedings of the Academy of Natural Sciences of Philadelphia, 5(12):308-309. [Date on title page is 1851; published in 1852.] Lilljeborg, W. 1867. On Two Subfossil Whales Discovered in Sweden. 48 pages, 11 plates. Uppsala: W. Schultz. Lyell, C. 1845. On the Miocene Tertiary Strata of Maryland, Virginia, and of North and South Carolina. Quarterly Journal of the Geological Society of London, 1:413—427, appendix. Mansfield, W.C. 1944. Stratigraphy of the Miocene of Virginia and the Miocene and Pliocene of North Carolina. In Julia Gardner, Mollusca from the Mi­ ocene and Lower Pliocene of Virginia and North Carolina. U.S. Geological Survey Professional Paper, 199-A: 1-16. McLeod, S.A., F.C. Whitmore, Jr., and L.G. Barnes 1993. Evolutionary Relationships and Classification. In J.J. Bums, J.J. Mantague, and C.J. Cowles, editors, The Bowhead Whale. Society for Marine Mammalogy. Special Publication, 2:45-70. Owen, Richard 1844. Appendix to Professor Henslow's Paper, Consisting of a Description of the Fossil Tympanic Bones Referable to Four Species of Balaena. Proceedings of the Geological Society of London, 4:283-286. Philippi, R.A. 1887. Die Tertidren und Quartdren Versteinerungen Chiles. 266 pages, 58 plates. Leipzig: F.A. Brockhaus. Rogers, B. 1881. A Reprint of Annual Reports and Other Papers on the Geology of the Virginias. 832 pages. New York: D. Appleton & Co. True, F.W. 1904. Whalebone Whales of the Western North Atlantic Compared with Those Occurring in European Waters, and Some Observations on the Species of the North Pacific. Smithsonian Contributions to Knowledge, 33: 332 pages, 50 plates. U.S. Geological Survey 1962. Paleontology, Evolution [Finding of 50-foot Fossil Whale]. In Geo­ logical Survey Research 1962. U.S. Geological Survey Professional Paper, 450A:474. 1965. Significant Finds of Vertebrate Fossils in Virginia and Florida. In Geological Survey Research 1965. U.S. Geological Survey Profes­ sional Paper, 525A:71. Van Beneden, P.J. 1872. Les baleines fossiles d'Anvers. Bulletin de I'Academie Royale des Sciences, des Lettres, et des Beaux-Arts de Belgique (Bruxelles), se­ ries 2, 34:6-20. Ward, L.W., and B.W. Blackwelder 1980. Stratigraphic Revision of Upper Miocene and Lower Pliocene Beds of the Chesapeake Group, Middle Atlantic Coastal Plain. U.S. Geo­ logical Survey Bulletin, 1482-D: 61 pages. Whitmore, F.C, Jr. 1994. Neogene Climate Change and the Emergence of the Modem Whale Fauna of the North Atlantic Ocean. Proceedings of the San Diego Society of Natural History, 29:229-238. Whitmore, F.C, Jr., and J.A. Kaltenbach MS. Neogene Cetacea of the Lee Creek Phosphate Mine, North Carolina. Paleontology of the Late Oligocene Ashley and Chandler Bridge Formations of South Carolina, 3: Eomysticetidae, a New Family of Primitive Mysticetes (Mammalia: Cetacea) Albert E. Sanders and Lawrence G. Barnes ABSTRACT A new family of relatively large, archaic fossil mysticetes, the Eomysticetidae, is based upon a new genus and new species, Eomysticetus whitmorei, from the late Oligocene Chandler Bridge Formation in the vicinity of Charleston, South Carolina, USA. With the exception of the primitive mysticete Micromysticetus rothauseni Sanders and Barnes (2002), all previously named Ceta­ cea from these deposits are odontocetes. Eomysticetus whitmorei, known by much of an associated skeleton, is a baleen-bearing mysticete having rostral features that are characteristically associ­ ated with baleen-whale feeding adaptations. The rostrum is rela­ tively broad and flat, the palate has nutrient foramina associated with baleen development, and the dentaries are elongate, oval in cross section, and edentulous. This whale shares many cranial fea­ tures with Archaeoceti, e.g., the narrow, elongate intertemporal region, narrow supraorbital processes of the frontals, elongate zygomatic processes of the squamosals, and small cochlear portion and narrow anterior process of the periotic. The numbers and structure of the vertebrae and ribs are intermediate between archaeocetes and cetotheriid mysticetes, and the relative length of the humerus compared with that of the distal limb bones (radius and ulna) is intermediate between those of archaeocetes and Neo­ gene mysticetes. A second, more highly evolved species, Eomysticetus carolinen­ sis, described herein, is represented by a partial skull also from the Chandler Bridge Formation. The osteology of Eomysticetidae strongly reinforces the ancestral-descendant relationship of Archaeoceti to Mysticeti and helps to substantiate the theory of the monophyly of the Cetacea. Eomysticetus whitmorei is the most archaic baleen-bearing mysticete yet described and survived into late Oligocene time as a relict form. In its degree of cranial tele­ scoping it is more primitive than the contemporaneous toothed mysticetes of the family Aetiocetidae and contemporaneous baleen-bearing members of the Cetotheriidae. The presence of the second species, E. carolinensis, in the same formation demon- Albert E. Sanders, The Charleston Museum, 360 Meeting Street, Charleston, South Carolina 29403. Lawrence G. Barnes, Natural His­ tory Museum of Los Angeles County, 900 Exposition Boulevard, Los Angeles, California 90007. strates the newly recognized evolutionary diversity of the Oli­ gocene Cetacea and the fact that multiple lineages of various groups evolved simultaneously. The family Eomysticetidae is the presumed sister taxon to all of the more-derived baleen-bearing Mysticeti. Introduction The Tertiary marine deposits of the South Carolina coastal plain have furnished important fossil cetacean remains since 1845, when Robert W. Gibbes described the archaeocete Doru- don serratus from Eocene beds in the vicinity of the Santee River. Subsequent discoveries have included two odontocetes, Agorophius pygmaeus (Muller, 1849) and Xenorophus sloanii Kellogg, 1923, both of which were described from holotypes found in the Ashley Formation near Charleston. As demon­ strated by Whitmore and Sanders (1976), the Ashley Formation is of late Oligocene age, and not Eocene as had been supposed by previous authors. In recent years, studies emanating from the Charleston Museum have disclosed a previously unrecog­ nized formation overlying the Ashley. This rock unit, the Chandler Bridge Formation of Sanders et al. (1982), also is of late Oligocene age (ca. 28 million years ago (Ma)) and has pro­ duced a wealth of cetacean material that is providing critical new information about the evolution and systematics of Oli­ gocene Cetacea (Whitmore and Sanders, 1976; Sanders, 1980; Sanders et al., 1982; Weems and Sanders, 1986; Sanders and Barnes, 1989; Barnes and Sanders, 1990). In October 1975, a Charleston Museum party under San­ ders's direction excavated the remains of the first baleen whale to be found in the Oligocene beds around Charleston. The spec­ imen was recovered from the Chandler Bridge Formation in Dorchester County, South Carolina, approximately 20 miles (32.2 km) north of Charleston. Subsequently, another baleen- bearing mysticete was found in this formation, and these two specimens are the subject of this paper. Previously described Oligocene mysticetes from the North Atlantic and Tethys re- 313 314 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY gions had consisted of two taxa, Cetotheriopsis lintianus (von Meyer, 1849) from Linz, Austria, and Cetotheriopsis tobieni Rothausen, 1971, from the North Sea Basin near Dusseldorf, Germany. Those taxa and the new form Micromysticetus rot­ hauseni are discussed elsewhere in this volume (Sanders and Barnes, 2002). The first South Carolina specimen is a partial skeleton repre­ senting a new genus and new species with cranial characters that are in many ways intermediate between those of previ­ ously described members of the suborders Archaeoceti and Mysticeti (Sanders and Barnes, 1989; Barnes and Sanders, 1990). The cranium and dentaries of this specimen are nearly complete and were associated with vertebrae, ribs, and the right forelimb. The second specimen is a partial skull representing another species in the genus. Two of the most intense subjects of current debate about ce­ tacean evolution are their place of origin and ancestry and whether they are monophyletic, diphyletic, or triphyletic. The descriptions of the new fossil cetaceans in this study shed light on the second of these questions and reinforce the theory that the Cetacea are monophyletic. New morphologic and taxo­ nomic interpretations and subordinal definitions are proposed, and comparisons are made with relevant previously named fos­ sil species. ACKNOWLEDGMENTS.—Initial thanks must go to Catherine Key, who discovered the holotype of Eomysticetus whitmorei and reported it to Sanders. She and Mary Worley contributed many hours of assistance with the excavation work, to which Chaz Due and Joel Padgett also made valuable contributions. We thank Leroy Blackwell for permission to excavate the holo­ type on his property and for furnishing a backhoe and operator to remove the Pleistocene overburden from the site. A consid­ erable portion of the preparation work was made possible by a grant from the Charleston Scientific and Cultural Educational Fund; the remainder was funded by the Charleston Museum. For their assistance with the preparation of the holotype we thank Barry Albright, Daniel Davis, Jonathan Geisler, Chester Linder, Aaron Stokes, Mark Taylor, and Robert E. Weems (USGS). Jonathan Durst-Glenn assisted in proofreading the manuscript. We are particularly grateful to Vance McCollum and Barry Albright, the collectors of the holotype of Eomysticetus caro­ linensis, for their generosity and interest, and we appreciate Mr. McCollum's assistance in the preparation of that specimen. Tanya Elston also aided with the preparation work. Bryan Stone produced all of the photographs except those in Figures 9, 10, 11, 12, and 17c, which were made by the first author with assistance from Aaron Stokes. Chris Olm prepared the drawing for Figure 3B, to which the first author added final de­ tails. The remainder of the line drawings were produced by the first author. Peter Laurie made the vertebral measurements for Figure 28, and Clyde White assisted with the tabulations. Mil­ ton Rhodes assisted with tabulations for Figure 29. The Charleston Museum and the Natural History Museum of Los Angeles County and its foundation provided facilities, salaries, and travel funds for the authors that made this study possible. Ewan Fordyce reviewed the manuscript and provided helpful comments. Clayton E. Ray and Frank C. Whitmore, Jr., pro­ vided access to fossil cetacean material in the NMNH and were helpful in numerous other ways. In our description of Micromysticetus rothauseni elsewhere in this volume, we have recorded our appreciation of Clayton Ray and his many contributions to vertebrate paleontology. We wish now to acknowledge another long-time friend and col­ league, Frank C. Whitmore, Jr., whom we have honored with the patronym Eomysticetus whitmorei. We can think of no pa- leocetologist more deserving of patronymic recognition, but it is at best a very inadequate means of expressing to him our sin­ cere thanks for his friendship and encouragement over the years. We have been the beneficiaries of his wise counseling and sound advice on many occasions, but as much as anything we have appreciated his ever-cheery disposition, even in the face of recent visual difficulties that would have defeated a less courageous person. Admired and respected by all who know him, he is at once an excellent scientist and a valued friend, and we consider it a privilege to have worked with him. Thank you, Frank, for everything. MATERIAL AND METHODS Terminology for cranial anatomy follows Kellogg (1927) and Fraser and Purves (1960). With some modifications re­ quired by the morphology of the specimen, cranial measure­ ments follow the methodology of Perrin (1975) and of Kellogg (1936, 1965). For the measurements of the mandibles, verte­ brae, and forelimb we have followed Kellogg (1936, 1965, 1968, 1969), also with required modifications. The reconstruc­ tion of the skull is based upon photographs of the specimen. Geological interpretations of the unit that produced the holo­ type specimen are in accordance with the original description of the Chandler Bridge Formation (Sanders et al., 1982) and with subsequent observations about that unit (Weems and Sanders, 1986). Authorships of modern taxa mentioned in the text follow Rice (1984), and synopses of living species, genera, and su- prageneric groups and their authorships may be found there and in Hershkovitz (1966). ABBREVIATIONS—Institutional abbreviations used in this pa­ per are as follows: ChM The Charleston Museum, Charleston, South Carolina NMNH National Museum of Natural History, Smithsonian Institution, Washington, D.C. USGS United States Geological Survey USNM collections of the National Museum of Natural History, includ­ ing collections of the former United States National Museum Anatomical abbreviations used in this paper are as follows: aon antorbital notch be basioccipital crest MBEF Boc ch Eo fm fps Fr gf hpt jn lc Max mea mf mrg Na Oc Pa Pal Pgl Pmx pop Pt pts Soc sop Sq sqf ssf tf Vo zps 193 basioccipital cranial hiatus exoccipital foramen magnum foramen pseudovale frontal glenoid fossa hamular process of pterygoid jugular notch lambdoidal crest maxilla eaxternal auditory meatus mandibular foramen mesorostral groove nasal occipital condyle parietal palatine postglenoid process premaxilla paroccipital process pterygoid fossa for pterygoid sinus supraoccipital supraorbital process squamosal squamosal fossa secondary squamosal fossa temporal fossa vomer zygomatic process of squamo Systematic Paleontology Class MAMMALIA Linnaeus, 1758 Order CETACEA Brisson, 1762 Suborder MYSTICETI Flower, 1864 Superfamily EOMYSTICETOIDEA, new superfamily INCLUDED SUPERFAMILIES.—Eomysticetoidea, new super- family; Balaenopteroidea (Gray, 1868) sensu Mitchell, 1989; Eschrichtoidea (Ellerman and Morrison-Scott, 1951) sensu Mitchell, 1989; Balaenoidea Brandt, 1873 (sensu Mitchell, 1989). INCLUDED FAMILY.—Eomysticetidae, new family, only. 315 fering from members of Aetiocetidae by the absence of teeth in adulthood; differing from Cetotheriidae, Balaenopteridae, Bal­ aenidae, Neobalaenidae, and Eschrichtiidae by having elongate anterolateral processes of nasal bones, extremely long zygo­ matic processes of squamosals, very thin blade-like anterior process of periotic and small posterior process with well-de­ fined facet for articulation with tympanic bulla, and humerus as long as radius and ulna. Differs from Balaenopteridae, Balae­ nidae, Neobalaenidae, and Eschrichtiidae by having naris near mid-length of rostrum, exceptionally elongate nasal bones, pa­ rietals exposed along intertemporal region between frontals and apex of supraoccipital, periotic with transversely com­ pressed anterior process and no dorsal process on cochlear por­ tion, and large coronoid process of dentary and posterior edge of coronooid elevated above anterior edge. The two known species of Eomysticetus share the following similarities: exoccipital thick and posteriorly flared; condyles broad and not protruding from skull; basioccipital slightly vaulted transversely; basioccipital crests short anteroposteri­ orly and wide transversely; large recess for posterior process of periotic; postglenoid process short, thin anteroposteriorly, and continuous with falciform process medially; glenoid fossa with oblique ridge from anteromedial to postlateral; dorsal curvature of zygoma; flattened ventral surface of glenoid fossa; large fo­ ramen pseudovale; wide jugular notch; narrow intercondylar notch; asymmetrical and interfrngering frontoparietal suture; sagittal crest present on apex of parietals in intertemporal re­ gion; sagittal crest on apex of supraoccipital at right of midline; secondary anteroposterior crest flanking each side of sagittal crest at apex of occiput; paroccipital process not extending lat­ erally as far as lateral edge of zygoma; floor of squamosal fossa with a pit-like depression herein termed the "secondary squa­ mosal fossa." TYPE SPECIES.—Eomysticetus whitmorei, new species. INCLUDED SPECIES.—Eomysticetus whitmorei, new species, and Eomysticetus carolinensis, new species. Late Oligocene, South Carolina, USA. ETYMOLOGY.—From acoc; (eos) (Greek), dawn, and u-igxccKO (mystako) (Greek), whisker or mustache, in reference to the ba­ leen plates in mysticete whales; and from cetus or cete (Latin, from Greek ketos), whale. Family EOMYSTICETIDAE, new family DIAGNOSIS.—The same as for the genus. TYPE AND ONLY INCLUDED GENUS.—Eomysticetus, new ge­ nus. Eomysticetus, new genus DIAGNOSIS.—A mysticete differing from all other baleen- bearing Mysticeti by having an elongate and narrow intertem­ poral region and floor of squamosal fossa with small pit-like depressions herein termed "secondary squamosal fossae"; dif- Eomysticetus whitmorei, new species FIGURES 3-17 DIAGNOSIS.—A species of Eomysticetus differing from E. carolinensis, n. sp., in having transversely thicker basioccipital crests; zygomatic process of squamosal thicker, more nearly parallel to sagittal plane, and extending beyond apex of su­ praoccipital; a facet on ventral surface of distolateral end of zy­ goma for articulation with jugal; lambdoidal crests more dor­ sally directed and not overhanging temporal fossae; parietals nearly vertical on either side of intertemporal region rather than sloping ventrolaterally; sagittal crest on parietals in intertempo- 316 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY FIGURE 1.—Map showing the type localities of Eomysticetus whitmorei, new genus and new species (A), and Eomysticetus carolinensis, new species (B), near Charleston, South Carolina. ("Excavation site" on map is loca­ tion of Charleston Museum excavation of the Chandler Bridge Formation (Sanders, 1980). Base map from Sand­ ers 1980, courtesy of National Geographic Society.) rai region blade-like posteriorly and not rounded; dorsal expo­ sure of parietal along midline of intertemporal region 50% longer; dorsomedial surface of zygomatic process of squamo­ sal convex rather than flat; deeper supracondylar fossa; articu­ lar surfaces of occipital condyles more protuberant posteriorly and more distant from occipital surface. HOLOTYPE.—ChM PV4253; skull, both periotics, both tym­ panic bullae, both dentaries; seven cervical vertebrae, seven thoracic vertebrae, two lumbar vertebrae, and one possible cau­ dal vertebra; parts of at least 17 ribs and a possible sternal rib; right scapula, humerus, radius, and ulna; collected by Albert E. Sanders and party, October 1975. TYPE LOCALITY (Figure 1).—South Carolina, Dorchester County, Greenhurst Subdivision, west bank of drainage ditch at junction with Chandler Bridge Creek, 0.16 km southwest of Jamison Road (County Road 377), approximately 32°52'54"N and 80°09'30"W (USGS, Stallsville 7.5' topographic quadran­ gle). Elevation of dorsal surface of cranium 5.7 m (18.58 ft.) above mean sea level. FORMATION AND AGE (Figure 2).—Bed 3 of the Chandler Bridge Formation, a late Oligocene (Chattian) marine unit laid down approximately 28 Ma (Sanders et al., 1982). The upper­ most portion of the Chandler Bridge Formation, bed 3, is a beach zone from which a large number of marine vertebrate re­ mains were recovered in a major excavation conducted by the ^ Ground level Wando Formation (Late Pleistocene ca. 0.6 mi FIGURE 2.—Stratigraphic section at the type locality of Eomysticetus whit­ morei, new genus and new species, from the Chandler Bridge Formation. (Not to scale.) NUMBER 93 317 Charleston Museum (Sanders, 1980). The site of that excava­ tion is the type locality for the Chandler Bridge Formation and is 2.1 km (1.3 mi.) southeast of the type locality of Eomystice­ tus whitmorei. The Chandler Bridge Formation unconformably overlies the late Oligocene Ashley Formation (ca. 30 Ma), a calcarenite that underlies the entire Charleston area. Together, these two forma­ tions have produced one of the largest and most diverse assem­ blages of cetacean remains recovered from Oligocene deposits. Squalodontoids and other primitive toothed cetaceans are by far the most numerous, whereas mysticetes are relatively rare. An early Chattian age for the Chandler Bridge Formation is in­ dicated by the presence in this formation of undescribed squal­ odonts of the same evolutionary grade as Eosqualodon langew­ ieschei Rothausen, 1968 (see Whitmore and Sanders, 1976; Sanders, 1980), found in Eochattian sands (Chattian A) at Do­ berg, Germany, that have been referred to nannoplankton zone NP24 (Martini and Muller, 1975) and are considered to be of early Chattian age (Curry et al., 1978:46). A detailed appraisal of the age of the Chandler Bridge Formation was given by Sanders et al. (1982). As the latter authors have noted, "Be­ cause the Chandler Bridge Formation is thin and permeable and thus leached of carbonate, attempts to recover a calcareous microfauna or microflora have been either unsuccessful or equivocal in that the few specimens obtained may represent material reworked from the underlying Ashley Member [now Ashley Formation]. . . into bed 1" (Sanders et al., 1982:H114). E. Martini, however, recently examined the nannoplankton in samples of the Ashley Formation from the type locality of the Chandler Bridge Formation and found the Ashley to be refer­ able to zone NP24 (pers. comm., E. Martini to K. Rothausen, Rothausen to A.E. Sanders, June 1990). Because the same evo­ lutionary grades and many of the same genera are represented in the cetaceans, phocids (Koretsky and Sanders, 2002), sea turtles, a new crocodilian (Erickson and Sawyer, 1996), and other vertebrate faunas of the Ashley and the Chandler Bridge Formations, we consider that these two formations belong to the same biostratigraphic interval (NP24) and that very little time (probably no more than about 2 My) elapsed between the deposition of these two units. The age of the Chandler Bridge Formation is placed at approximately 28 Ma and the underly­ ing Ashley Formation at about 30 Ma. Those estimates are sug­ gested by (1) the primitive aspect of the cetacean fauna of the two units, among which are three undescribed toothed mystice­ tes with archaeocete-like dentition (Barnes and Sanders, 1996), several relatives of the most-archaic known odontocete, Xeno­ rophus sloanii Kellogg (1923), from the Ashley Formation; and the archaic Eomysticetus whitmorei and E. carolinensis de­ scribed in this paper; but primarily by (2) biostratigraphic cor­ relation of the Ashley Formation with nannoplankton zone NP24. ETYMOLOGY.—The species is named in honor of Frank C. Whitmore, Jr., formerly of the USGS, who has contributed greatly to the study of fossil cetaceans and who has supported us in our efforts in innumerable ways. DESCRIPTION.—Skull: The skull of Eomysticetus whit­ morei is very long and has a flat, narrow rostrum; a narrow, tri­ angular braincase; large and widely placed zygomatic pro­ cesses of the squamosals; and an elongate and greatly con­ stricted intertemporal region (Figure 3, Table 1). We have esti­ mated the length of the skull at 1590 mm by articulating the complete right dentary with the right glenoid process and leav­ ing 5 mm clearance for connective tissue. The rostrum is composed mostly of maxilla on the dorsal surface, as in other mysticetes. The ventral surface of the ros­ trum lacks development of nutrient grooves, as is typical of other mysticetes. The mesorostral gutter is open dorsally for only a short distance immediately anterior to the narial open­ ing. The premaxillae converge at a point 180 mm anterior to the nasals, and at this location they are a mere 10 mm apart. From that point anteriorly they are closely approximated on the anterior portion of the rostrum, effectively roofing over the me­ sorostral groove in that region. Anterior to the naris the pre­ maxillae expand anteriorly and are convex transversely, but ad­ jacent to the narial region the premaxillae are narrower and more convex along their midlines. On the right side, adjacent to the narial opening, a piece of the right maxilla attaches to the premaxilla. It laps both dorsally and ventrally to the lateral margin of the premaxilla, and its dorsal surface is convex. Pos­ teriorly, the premaxillae diverge posteriorly at the nares and curve around the lateral margins of the nasal bones, producing a slightly convex surface that slopes ventrolaterally away from the nasals. Adjacent to the anterolateral corner of the naris, the premaxilla is elevated, as in Aetiocetidae, and then slopes an­ teroventrally (Figure 4). The premaxillae widen posteriorly, ex­ tending into the interorbital area and terminating a few milli­ meters anterior to the posterior tip of the nasals. Breakage of the posterior ends of the premaxillae and nasals prohibits an exact measurement of the latter distance. The posterior termi­ nation of each premaxilla appears to have tapered between the ascending process of the frontal and the posterior end of the na­ sal and probably terminated at the same point as the nasals. The nasal bone is 300 mm in length, exceeding that of any other described mysticete. It extends posteriorly to a point be­ tween the temporal crests and constitutes approximately 18.7% of the estimated total length of the skull. At its midline, the an­ terior margin of the nasal is virtually straight across trans­ versely, then is slightly concave laterally. Farther laterally, to­ ward the margin of the premaxilla, the anterior edge of the nasal puts forth a small process that is directed farther anteri­ orly and is sutured into a recess along the medial side of the premaxilla (Figure 3). The right and left nasals are joined at the midline by an irreg­ ular, meandering suture. They are arched transversely, and at the apex of the narial opening they are slightly upturned. From the edge of the narial opening the nasals slope gently down­ ward and backward into a shallow but quite noticeable swale 318 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY 10 cm A FIGURE 3.—Eomysticetus whitmorei, new genus and new species: A, holotype, ChM PV4253, skull, dorsal view; B, reconstruction of skull based upon holotype, dorsal view. (Solid lines in areas of missing bone (see Figure 3A) are hypothetical configurations. Abbreviations are explained in "Material and Methods.") that extends posteriorly from the anterior margins of the nasals as far as the apex of the supraoccipital (Figure 4). The nasals are widest anteriorly, tapering posteriorly as they overlap the frontals, terminating posteriorly slightly behind the point of the temporal crests. The supraorbital processes of the frontals are incomplete on the holotype. Their posterior margins are preserved, but the an­ terior and lateral margins are missing on both sides. The right supraorbital process of the frontal is more complete than the left one and seems to have been relatively narrow in compari­ son with most mysticetes, although that appearance might be an artifact of preservation. The posterior margin is concave, and the anterior margin is mostly incomplete, so that the rela­ tionship with the maxilla is not clear. The distal margin of each NUMBER 93 319 TABLE 1.—Measurements (in mm) of the holotype skulls of Eomysticetus whitmorei, new genus and new species (ChM PV4253), and Eomysticetus carolinensis, new species (ChM PV4845). Parentheses denote estimated mea­ surements. Character Condylobasal length Length of rostrum Preserved portion of skull, broken end of left premaxilla to posterior surface of left occipital condyle Anterior end of left premaxilla to apex of supraoccipital Greatest length of left premaxilla E. whitmorei E. carolinensis 'Estimated with complete right dentary articulated supraorbital process is incomplete but is slightly upturned. A well-defined temporal crest extends transversely across the posterodorsal surface of the supraorbital process and reaches nearly to the midline of the skull, converging with the posterior margin of the supraorbital process and continuing laterally onto the dorsal surface of the process. This crest is highest at its me­ dial part. The intertemporal region is greatly elongate and narrow and is composed dorsally of a relatively short exposure of the fron­ tals and the anteroposteriorly more elongate parietal bones. The frontals and parietals are joined by irregularly and randomly in- terdigitating sutures (Figure 5). The intertemporal constriction ascends toward the apex of the occipital shield, and the poste­ rior half of the parietals form a short blade-like sagittal crest (1590)' (1147) 1035 820+ 640+ Greatest width of right maxilla (preserved portion) Anterior end of right premaxilla to anterior end of right nasal bone Anterior end of right nasal bone to apex of supraoccipital Deepest point of swale between tip of nasals and apex of supraoccipital Greatest length of right nasal bone at midline Anterior width of right nasal bone Combined width of nasal bones, anteriorly Combined width of nasal bones, posteriorly Length of anterolateral extension of left nasal bone from plane of anterior end of nasals Height of anterior ends of nasal bones above plane of premaxillae at nasal opening Greatest width of external nares Transverse distance between outside margins of premaxillae at level of anterior ends of nasal bones Length of intertemporal region, temporal crest to plane of apex of supraoccipital Transverse distance between outside margins of preserved portions of supraorbital processes Greatest width across zygomatic processes of squamosals Maximum length of right zygomatic process, extremity of postglenoid process to anterior end of zygoma Length of temporal fossa Width of temporal fossa Dorsal margin of foramen magnum to apex of supraoccipital Transverse diameter of foramen magnum Vertical diameter of foramen magnum Greatest distance between lateral margins of occipital condyles Vertical diameter of right occipital condyle Greatest transverse diameter of right occipital condyle Distance from inner margin of right occipital condyle to outer edge of right exoccipital Distance between lateral margins of exoccipitals Vertical distance from basisphenoid to apex of supraoccipital Greatest distance between lateral margins of basioccipital crests Internal length of braincase Internal height of braincase, left cerebral cavity Internal height of braincase, right cerebral cavity Greatest internal height of braincase at midline Greatest internal width of braincase 56 350+ 469 37 (300) 35 71 (40) (26) (23) 49 95+ 192 373 480 261 357 170 229 (32) 46.7 142 76 58 163 (350) (168) 132+ (120) 80 81 87 157 _ _ - - - - _ - - - _ - - - (510) 233 - 183 236 43 44.2 134.5 84 53 147.5 (342) 131 (100) - - 87 - (Figures 6, 7). The sagittal crest diverges at the juncture with the apex of the supraoccipital, and the remaining bone surface indicates that it was confluent with the lateral margins of the occipital shield. Anteriorly, the supraoccipital is triangular and relatively small and has a prominent sagittal crest beginning approximately at the middle and extending toward the apex. The anterior region of the occipital shield is elevated, but immediately posterior to the sagittal crest it is concavely depressed. Mediolateral to this depression there is a thickened tuberosity on the supraoccipital next to the inner wall of each lambdoidal crest. At that point, the supraoccipital begins to flare outward ventrolaterally More posteriorly, the surface of the occipital shield is flat to convex dorsal to the foramen magnum, and a condyloid fossa is 320 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY FIGURE 4.—Eomysticetus whitmorei, new genus and new species: A, holotype, ChM PV4253, skull, left lateral view; B, reconstruction of skull based upon holotype, left lateral view. (Dashed lines represent hypothetical con­ figurations. Abbreviations are explained in "Material and Methods.") present above each condyle. The occipital condyles are large, are located relatively low on the occipital shield, and are broad transversely. Lateral to the condyles, the surface of each exoc­ cipital is unusually convex. As in most fossil and extant baleen whales, the exoccipital is mostly thick. In its lateral portion, however, the paraoccipital is thin anteroposteriorly and does not extend laterally farther than the middle of the postglenoid process. The lambdoidal crest forms the lateral margin of the occipital shield and descends posteriorly and ventrally toward the back of the squamosal fossa. It is continuous with a prominent and narrow ridge-like structure that extends onto the dorsal surface of the zygomatic process of the squamosal. This ridge walls in the posterior end of the squamosal fossa, overhangs the second­ ary squamosal fossa, and is higher than in later mysticetes. A flexure in this crest is the homologue of the squamosal promi­ nence of Micromysticetus rothauseni (Sanders and Barnes, 2002). Between the braincase and the zygomatic arch, the squamo­ sal fossa is deep, anteroposteriorly elongate, and wide. The posterior end of the fossa ascends steeply to meet the lateral part of the lambdoidal crest. This oblique part of the crest is prominent and spans to the posterior part of the zygomatic pro­ cess of the squamosal. At the posterior end of the squamosal fossa, adjacent to the base of the zygomatic process of the squamosal, is a small, circular, pit-like fossa, herein named the secondary squamosal fossa (Figures 3, 5). The lateral wall of the braincase is nearly vertical, is slightly concave near the middle, and is not bowed laterally as in most of the highly evolved taxa of Mysticeti. The zygomatic process of the squamosal is very large rela­ tive to the size of the cranium, is elongate, and diverges only slightly from the midline of the cranium. It arches anterodor­ sally in a broad curve and bends ventrally at its anterior ex­ tremity. The medial face of the process is composed of dense bone and is divided into two planes, one upper and one lower. Its lateral surface is uniformly and smoothly convex and flares slightly ventrolaterally at the ventral margin, especially so at the anterior end of the process. In the ventral tip of the process there is a recess for articulation of the jugal bone (Figure 8). Lateral to this recess, the margin of the zygomatic process flares ventrolaterally. The arch is slightly expanded dorsoven­ trally at its anterior end and also arches slightly anteroposteri­ orly to create the large, anterodorsally inclined glenoid fossa. At the posterior end of the lateral surface of the zygomatic process there is a prominent, transversely oriented stemomas­ toid fossa. This fossa is inclined anterodorsally and is located immediately lateral to the small paroccipital process. The latter NUMBER 93 321 H FIGURE 5.—Eomysticetus whitmorei. new genus and new species, holotype, ChM PV4253, skull: enlarged view of braincase showing frontoparietal suturing and secondary squamosal fossa. (Abbreviations are explained in "Material and Methods.") projects slightly ventrally and posteriorly from the occipital shield. The recess for the external acoustic meatus is deep and is clearly defined between the postglenoid process of the squamo­ sal and the paroccipital process. The mastoid process (posterior process of the bulla) is short and occupies all of the recess. For a mysticete, Eomysticetus whitmorei has a relatively small cranial hiatus, the recess between the squamosal and the basioccipital in which the periotic lies. Lateral to the cranial hi­ atus and immediately posterior to the large, obliquely oriented foramen pseudovale is a rather thick falciform process of the squamosal that is bridged to the postglenoid process of the squamosal by a thin wall of bone. This wall of bone is a struc­ ture unique to Eomysticetus and marks the lateral border of a large peribullary sinus that extends anteriorly from the cranial hiatus, but it is not possible to determine its anterior extent be­ cause of breakage of bone in the area. On the lateral wall of the cranial hiatus, dorsal to the medial end of the mastoid process, is an epitympanic recess. As in Micromysticetus (Sanders and Barnes, 2002) and other primitive mysticetes, the ventral sur­ face of the basioccipital is arched transversely, a form that is considered to be a primitive character. The basioccipital crests are thick and knob-like. The glenoid fossa is broad transversely with a nearly square articular surface that is delimited anteriorly by a transverse margin extending laterally from the anterior edge of the squa­ mosal fossa. The postglenoid process is relatively short and thin and is thickest laterally, where its ventral margin is con­ vex; it then becomes more thin medially, where it terminates at the margin of the peribullary fossa. ^w; r-* i i 5 cm W%0 s*'\ FIGURE 6.—Eomysticetus whitmorei. new genus and new species, holotype, ChM PV4253, braincase, left lateral view. (Abbreviations are explained in "Material and Methods.") 322 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY , 10 cm f a . i A FIGURE 7.—Eomysticetus whitmorei, new genus and new species: A, holotype, ChM PV4253, cranium, ventral view; B, reconstruction of holotype cranium, ventral view. (Dashed lines represent hypothetical configurations. Abbreviations are explained in "Material and Methods.") Periotic: Both of the periotics of Eomysticetus whitmorei axe preserved and are relatively small for a mysticete (Figure 9). The left one is virtually complete, but the right one is missing approximately one-half of the pars cochlearis. The holotype left periotic measures 78 mm from the anteriormost point of the an­ terior process to the posteriormost extent of the posterior pro­ cess. The pars cochlearis is 38 mm in anteroposterior length, 25 mm in dorsoventral diameter, and 31 mm in transverse diame­ ter. At the anterior end of the cochlear portion the bone is deeply incised by the fossa for the tensor tympani muscle. The anterior process is extremely compressed transversely and broadly expanded dorsoventrally (Figure 10). The poste­ rior process is much shorter and is laterally expanded dorsally but more laterally compressed ventrally. The cochlear portion is relatively small, does not project very far medially from the body of the bone, and is not as globose as that of Micromystice­ tus rothauseni (Figure 10; see also Sanders and Barnes, 2002, fig. 12). A large dorsal process, such as is typical of the more highly evolved mysticetes, is not present, but the bone around the internal acoustic meatus is slightly convex, especially lat­ eral to the meatus. In its overall proportions, the periotic resem­ bles those of the basilosaurid Archaeoceti more than the in­ flated periotics of the derived Mysticeti (Figure 11). In the primitive character state (e.g., Basilosauridae such as NUMBER 93 323 Sq '-%& 5 cm Pa ssf- sqf zps FIGURE 8.—Eomysticetus whitmorei, new genus and new species, holotype, ChM PV4253, enlarged view of intertemporal region of cranium, dorsal view. (Abbreviations are explained in "Material and Methods.") Zygorhiza), the internal acoustic meatus on the cochlear por­ tion of the periotic is a single large aperture, and included within it are the foramina for the facial nerve (VII), the vestibu­ lar nerve (=foramen singulare) (VIII), and the cochlear nerve (=tractus spiralis foraminosus) (VIII). This primitive condition also is widespread among the odontocetes. In E. whitmorei, however, the internal acoustic meatus is di­ vided into two parts separated by a thick, dorsally projecting transverse septum of bone that isolates the large foramen for the facial nerve from the foramina for the vestibular nerve and the cochlear nerve (Figure 10c). This situation also is present in such primitive Cetotheriidae as Herpetocetus spp., and in the more derived Cetotheriidae. In the Balaenopteridae the same bony septum is very prominent and extends very far cranially, and its extensive development is correlated with the extreme dorsal enlargement of the dorsal tuberosity and the bone around it. The presence of the septum dividing the originally single in­ ternal acoustic meatus is an autapomorphy of Eomysticetidae and of most other Mysticeti beyond it. The loss of a distinct sinuous dorsal crest on the periotic, originally traversing from the center of the cochlear portion to the anterior process, is related to the inflation of the cerebral surface of the periotic and development of a dorsal tuberosity. More extreme cerebral or dorsal extension of the dorsal tu­ berosity, associated with dorsal extension of the transverse sep­ tum dividing the internal acoustic meatus and also of other bone surrounding the internal acoustic meatus, is an autapo­ morphy of the Balaenopteridae. The septum dividing the internal acoustic meatus in E. whit­ morei is curiously missing in Micromysticetus rothauseni. That taxon has the primitive character state of the single, large, ovoid internal acoustic meatus (Sanders and Barnes, 2002, fig. 12). Micromysticetus rothauseni also has a well-defined sinu­ ous dorsal crest that traverses the pars cochlearis, also a primi­ tive character. Only a weak suggestion of that crest is present in E. whitmorei. Most likely, these features in Micromysticetus are character reversals or relict conditions. Because Micromystice­ tus is otherwise more highly derived than Eomysticetus whit­ morei (and we have classified it among the Cetotheriidae), and doubtless was a baleen-bearing mysticete, it is very unlikely that Micromysticetus represents some different lineage of Mys­ ticeti that evolved separately from archaeocete-like primitive mysticetes and was convergent with the Cetotheriidae. Tympanic Bulla: Both tympanic bullae are present. The bulla is proportionally small for a mysticete. The involucrum is tapered anteriorly and is not composed of thick bone. As is typ­ ical of mysticetes, the sigmoid process is large and convex. In overall shape, the bulla is ovoid and very much like that of ar­ chaeocetes of the family Basilosauridae. Dentaries (Table 2): The virtually complete right dentary is relatively long and slender with a slight lateral curvature, a dis- 324 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY TABLE 2.—Measurements (in mm) of the holotype dentaries of Eomysticetus whitmorei, new genus and new spe­ cies, ChM PV4253. Parentheses indicate estimated measurements. Character Total length (straight line) as preserved Greatest vertical diameter 100 mm behind anterior end of ramus, as preserved Greatest transverse diameter 100 mm behind anterior end of ramus, as preserved Greatest transverse diameter 500 mm behind anterior end of ramus (394 mm anterior to posterior face of condyle), as preserved Greatest vertical diameter 500 mm behind anterior end of ramus (394 mm anterior to posterior face of condyle), as preserved Greatest vertical diameter 500 mm behind anterior end of ramus Greatest transverse diameter 500 mm behind anterior end of ramus Greatest vertical diameter 1000 mm behind anterior end of ramus Greatest transverse diameter 1000 mm behind anterior end of ramus Greatest vertical diameter through coronoid process Horizontal distance from center of coronoid process to posterior face of condyle Horizontal distance across base of coronoid process Horizontal distance from posteriormost extent of condyle to beginning of anterior edge of coro­ noid Horizontal distance from posteriormost extent of condyle to end of posterior edge of coronoid Greatest distance from posteriormost extent of condyle to internal margin of orifice for mandibu­ lar canal Greatest vertical diameter of posterior end including condyle Greatest transverse diameter of condyle Greatest vertical diameter of condyle Left 894 90 50 91 45 - - - - (200) (216) 300 390 100 (210) (185) 46 101 Right 1513 63+ 32.5 - - 72.5 53 83+ 54 - 221 - 389 - (225) - 49 - A 2 cm :Mm- 'A v D 2 cm FIGURE 9.—Eomysticetus whitmorei, new genus and new species, holotype, ChM PV4253: A, left periotic; B, right periotic; C, right tympanic bulla in ventral view; D, left tympanic bulla in cerebral or dorsal view. NUMBER 93 325 l.a.m. s . f c vm-v.n. vni-c.n 2 cm 2 cm FIGURE 10.—Eomysticetus whitmorei, new genus and new species, holotype, ChM PV4253, left periotic: A, ven­ tral view; B, cerebral or dorsal view; C, dorsolateral view. (Dashed lines represent hypothetical configurations; a.c.=aqueducrus cochlearis; c.n.=foramen for cochlear nerve (VIII); f.i.=fossa incudis; f.n.=foramen for facial nerve (VII); fe.o.=fenestra ovalis; fe.r.=fenestra rotunda; f.t.t.=fossa for tensor tympani muscle; i.a.m.=internal auditory meatus; n.f.=notch for exit of facial nerve; pa. co.=pars cochlearis; po.=promontorium; pr.a.=anterior process; pr.p.=posterior process; s.f.=subarcuate fossa; v.n.=foramen for vestibular nerve (VIII).) tinct downward bend anteriorly, a large and laterally deflected coronoid process, and a small posteriorly directed condyle (Figure 12). Approximately 60% (894 mm) of the left dentary is preserved (Figure 13). Most of the posterior end is present in both dentaries, but the anterior end is missing in the left one. The posterior end of the left dentary is complete except for the lower angle, which is missing in both dentaries. The coronoid process is preserved only on the right side and is relatively large and has a high apex. There is a large mandibular foramen, and, as is typical of mysticetes, the dentary is elongate and lacks alveoli. On the external (labial) surface of the nearly complete right dentary is a series of 10 mental foramina begin- • '*' c 3 cm FIGURE 11.—Left periotics: A, Zygorhiza kochii (Reichenbach), ChM PV5065; B, Eomysticetus whitmorei, new genus and new species, holotype, ChM PV4253; C, Micromysticetus rothauseni Sanders and Bames, holotype, ChM PV4844. 326 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY 2 cm FIGURE 12.—Eomysticetus whitmorei, new genus and new species, holotype ChM PV4253, right dentary: A, labial view; B, lingual view; c, dorsal view; D, distal end showing groove for symphyseal ligament. (cm.=man- dibular condyle; cor.=coronoid process; f.g.=gingival foramen; f.m.=mental foramen; m.f.=mandibular fora­ men.) ning at 95 mm anterior to the anteriormost edge of the coronoid and appearing thereafter at 155, 280, 410, 490, 580, 660, 750, 870, and 940 mm from that point. The posteriormost foramen opens posteriorly. Though badly broken and eroded, a remnant of the gingival groove is preserved on the dorsal margin of the dentary. The groove for the symphyseal ligament begins about 20 mm posterior to the anterior end of the dentary and is ap­ proximately 43 mm in anteroposterior length (Figure 12D). It is situated in the ventral one-third of the dentary height and bends slightly dorsally at its anterior end. The ligamentary groove is quite distinct and is indicative of the typical ligamental man­ dibular symphysis of the more highly evolved mysticetes, such as cetotheriids and balaenopterids. For approximately 600 mm of its length anterior to the coro­ noid, the right dentary is almost entirely straight (Figure 12c). From there forward, the anterior part tapers in uniform dors­ oventral diameter for the remaining length of the bone. Ap­ proximately 600 mm from the anterior end, the horizontal ra­ mus begins to bend noticeably downward and is of almost uniform dorsoventral diameter from that point to its anterior terminus. Unlike the mandibles of most Neogene mysticetes, which are bowed laterally between the level of the coronoid process and the anterior end, the lateral curvature of the dentary in E. whitmorei is restricted to the anterior half and forms a much flatter arc than in such forms as Pelocetus calvertensis Kellogg (1965, fig. 6), Mesocetus siphunculus Cope, 1895 (see NUMBER 93 327 > = — — : . 3 10 cm FIGURE 13.—Eomysticetus whitmorei, new genus and new species, holotype, ChM PV4253, left dentary: A, dor­ sal view; B, labial view; c, lingual view. (For abbreviations, see Figure 12.) Kellogg, 1968, fig. 49), and Parietobalaena palmeri Kellogg (1968, fig. 89), all from Miocene deposits of Virginia. The flat­ tened curvature of the mandible of E. whitmorei is most nearly like that of Diorocetus hiatus Kellogg (1968, fig. 59), in which the curvature is reduced to conform to the narrow, strongly ta­ pered rostrum in that species, which also is from Miocene de­ posits of Virginia. The length and slight degree of convexity in the mandible of Eomysticetus demonstrates the presence of a long, narrow rostrum in this new Oligocene mysticete, as indi­ cated also by the narrowness of two large anterior fragments of the left and right maxillae. Posteriorly and in dorsal aspect, the external wall of the man­ dible follows a broad, laterally directed (concave) curve be­ tween the level of the anterior end of the coronoid and the pos­ terior edge of the condyle (Figure 12c). To our knowledge, that configuration of the posterior region of the dentary is not known in any other mysticete. In most previously described taxa, the lateral surface follows a convex line and is then sharply recurved in the area between the apex of the coronoid and the condyle, thus placing the posterior edge of the coronoid further medial than the anterior edge. In E. whitmorei the poste­ rior edge is on the same long, concave curve as the anterior edge and is not offset medially. Along the base of the coronoid process, the lateral surface of the ramus sweeps dorsally in a broad curve that terminates at the anterior edge of the mandibu­ lar condyle, which lies approximately 60 mm above the plane of the dorsal margin (Figures 12A, 13B). Extended posteriorly, the axis of the dorsal margin bisects the condyle approximately at the center of its vertical diameter. The condyle is thus seated extremely high on the posterior end of the mandible. A similar situation exists in the dentaries of Mesocetus siphunculus (see Kellogg, 1968, fig. 49) and Diorocetus hiatus (see Kellogg, 1968, fig. 59), but in neither of these forms is the condyle ele­ vated nearly so high as in A. whitmorei. In all three taxa, how­ ever, the condyle is located much higher than in Parietobal­ aena palmeri and Pelocetus calvertensis, the anterior face being only slightly above the plane of the dorsal edge in the lat­ ter forms. Of the four Miocene cetotheriids mentioned herein, the mandible of Diorocetus hiatus most nearly resembles that of Eomysticetus whitmorei in the moderate degree of convexity of the ramus and in the dorsoventral height of the anterior face of the condyle. The mandibular foramen is relatively large and opens posteriorly. Its anteriormost margin is aligned approxi­ mately with the apex of the coronoid process, and the anterior margin is a broad curve. The lateral wall within the foramen is thin and dense. The coronoid process is exceptionally large for a mysticete and more nearly resembles that of the archaeocete Zygorhiza kochii (Reichenbach, 1847) (see Kellogg, 1936, fig. 31a). It has a relatively broad base anteroposteriorly and a high, rounded apex. The apex is present on the right dentary ofE. whitmorei but is missing in the left one (Figures 12, 13). It is thick anteri­ orly and thinner posteriorly. As seen in Figure 12, the dorsal half of the coronoid is curved laterally in a gentle arc so that it is angled dorsolateral^ as in many Neogene mysticetes, but not to such an extreme as in the latter forms (e.g., Parietobalaena 328 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY TABLE 3.—Measurements (in mm) of the cervical vertebrae of Eomysticetus whitmorei, new genus and new spe­ cies, ChM PV4253. Parentheses indicate estimated measurements. Character Anteroposterior diameter of centrum Transverse diameter of centrum, anteriorly Vertical diameter of centrum, anteriorly Tip of neural spine to ventral face of centrum, anteriorly Greatest vertical diameter of neural canal, anteriorly Greatest transverse diameter of neural canal, anteriorly Greatest distance between outer ends of diapophyses Greatest distance between outer ends of parapophyses Least anteroposterior diameter of right pedicle of neural arch Greatest transverse diameter of centrum, posteriorly Greatest vertical diameter of centrum, posteriorly Anteroposterior diameter into transverse diameter, centrum (anteriorly) Anteroposterior diameter into vertical diameter, centrum (anteriorly) Atlas 57 137 73 (128) 73 46 - - 54 (127) (65) 2.40 1.32 Axis 66* 127 (48) - - 50 - - 15 112 83 1.92 0.76 C.3 39 102 84 - - - - - - 105 92 2.62 2.15 C.4 33 97 - - - - - - - 99 90 2.94 - C.5 36 (100) (95) - - - - (186) - (96) 92 2.78 2.64 C.6 40 98 (83) - (43) (41) (156) (182) 19 94 (88) 2.45 (2.08) C.7 40 95 83 (170) (45) (89) (196) (124) 20 97 88 2.38 2.08 "•Includes odontoid process. palmeri Kellogg (1968, fig. 89)). In E. whitmorei, there also is a proportionately shorter distance between the anterior edge of the condyle and the terminus of the posterior edge of the coro­ noid, which also is situated well above the dorsal margin of the ramus. Cervical Vertebrae: All seven of the cervical vertebrae are preserved (Table 3). None is ankylosed with any of the others in the series; however, the posterior face of the axis vertebra is deeply concave and closely encloses the anterior face of the third cervical, apparently presaging the eventual ankylosis of these two vertebrae in some Neogene mysticetes. The anterior face of the third cervical vertebra is slightly concave trans­ versely, except for its upper portion, which is thickened antero­ posteriorly so that it projects anteriorly to fit into a concavity along the dorsal margin of the posterior face of the axis verte­ bra. In all of the other cervical vertebrae following the axis, the anterior face of the centrum is flat and the posterior face is con­ cave. The epiphyses are firmly ankylosed to the centra in all of the cervical vertebrae, indicating the physical maturity of the specimen. The atlas is missing both of its transverse processes, but the size of the broken areas at their bases on the centrum suggests that the processes were of modest size and apparently were not large enough to have enclosed a transverse foramen (Figure 14B). They were situated relatively high on the centrum, di­ rectly upon the juncture of the dorsal and lateral surfaces. The transverse diameter of the anterior face is broadest at that point (137 mm) and narrowest across the ventral margin (86 mm). The facets for articulation with the occipital condyles of the cranium are deeply concave and somewhat angular along their lateral margins. Vertically, they are 86 mm in diameter and ex­ tend above the level of the dorsal surface of the centrum to a point just below the roof of the neural canal, their dorsal mar­ gins forming the anterior edges of the pedicles of the neural arch. The dorsal margins of the facets extend anteriorly beyond the level of the ventral margins. Twelve millimeters behind the anterior margin of the neural arch, each of the pedicles is pierced by a large (13 mm on right, 15 mm on left), laterally di­ rected arterial foramen, or canal, that emerges into a shallow groove on the dorsal surface of the centrum. The spine is miss­ ing from the neural arch, but it may not have been much more than an elevated, knob-like continuation of the low, median ridge present on the anterior portion of the arch. The lower por­ tion of the neural canal is deeply concave, the ventral margin being only 44 mm above the ventral surface of the centrum. Anteriorly, its lateral margins are formed entirely by the lateral margins of the articular facets; the transverse diameter of the canal is 42 mm dorsally and 17.5 mm ventrally. The margins of the posterior face of the atlas are not well preserved and pro­ vide only an approximate measurement of the transverse diam­ eter, which was at least 127 mm. The facets for articulation with the axis vertebra are much smaller than the anterior facets, their dorsal margins lying well below the dorsal surface of the centrum. The right one is the best preserved of the two and is greater in vertical diameter (~60 mm) than in transverse diame­ ter (-47 mm). The ventral margin of the posterior face is pre­ served sufficiently to determine that a hyapophysis is not present on the atlas of E. whitmorei. The axis vertebra is missing the neural arch (Figure 14c). The stumps of the pedicles are directed posteriorly at an angle and are flattened anterolaterally. The transverse processes are short and blunt, are imperforate, and are bent posteriorly. The ends are upturned and thus do not project below the level of the ventral surface of the centrum. The anterior face of the axis slopes posteriorly at a slight angle. The anterior facets for artic­ ulation with the atlas are concave transversely and are sepa­ rated above the odontoid process by an interval of approxi­ mately 52 mm across the midline of the centrum. The odontoid process is broadly rounded ventrally; dorsally, it is flattened on either side of a median ridge that extends posteriorly onto the dorsal surface of the centrum. Between the pedicles of the neu­ ral arch, two foramina are situated on each side of the bony ridge. The posterior face is deeply excavated both transversely and dorsoventrally, but more so in the former direction and es­ pecially along the dorsal margin. The neural arch and most parts of the elements of the trans- NUMBER 93 329 3 cm 2 cm B FIGURE 14.—Eomysticetus whitmorei, new genus and new species, holotype, ChM PV4253: A, atlas vertebra in anterior view; B, same, right lateral view; C, axis vertebra, anterior face. (od.=odontoid process; tr.=trans verse process.) verse processes are missing on the third, fourth (Figure 15), and fifth cervical vertebrae. The left lower transverse process is present on the third cervical, and both of the lower processes are preserved on the fourth cervical. On the dorsal surface of the centrum of each of these three vertebrae there is a pair of foramina at the midline. On the third through the seventh cervi­ cal vertebrae, the profile of the anterior face of the centrum is approximately that of a flattened circle, the greatest diameter occurring transversely on a level coinciding with the dorsal margin of the base of the parapophyses. The profile of the pos­ terior face is of the same general conformation, with the excep­ tion of the third cervical. In that vertebra, the ventral margins of the posterior face assume a triangular shape, sloping sharply from the widest point of the anterior face to a broadly rounded angle at the edge of the ventral surface of the centrum. This an­ gle projects below the ventral margin of the parapophyses, which extend below the ventral margin of the anterior face. On the sixth cervical vertebra, the neural spine and the left half of the neural arch are missing (Figure 16A). From the pre­ served right portion it can be seen that the anterior profile of the neural canal was that of a triangle with rounded angles. The ventral articular facet of the postzygapophysis has the form of a truncated ellipse, is 23 x20 mm in dimension, and projects pos­ teriorly 5 mm beyond the centrum. The diapophysis is much more slender than the parapophysis, the base of which is larger in this cervical than in any of the others, extending approxi- 330 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY d.a. A 3 cm d.a. FIGURE 15.—Eomysticetus whitmorei, new genus and new species, holotype, ChM PV4253, cervical vertebrae, anterior views: A, third cervical; B, fourth cervical. (d.a.=diapophysis; p.a.=parapophysis.) mately 17 mm below the ventral surface of the centrum. The vertical diameter of the lateral foramen is 46.6 mm, greater than that of any of the other cervical vertebrae. The articulating portions of the diapophyses and the parapophyses are missing. On the seventh cervical vertebra, the neural arch is complete and the neural canal is triangular in anterior profile (Figure 16B) . The pedicles of the arch are considerably larger than those of the other cervical vertebrae and are noticeably asym­ metrical in dorsoventral diameter, the right one being 33 mm and the left one 28 mm. There also is a striking difference in the dimensions of the ventral facets of the postzygapophyses, the right one being approximately 40 x 16 mm and the left one approximately 31 x 11 mm. Both are much longer anteroposte­ riorly and more narrow transversely than corresponding facets of the sixth cervical. The upper and lower transverse processes are more complete on the right side than on the left, the entire right parapophysis being missing. The articulating portions of the right processes are not preserved. In contrast to the sixth cervical, the diapophysis is much larger than the parapophysis, which is greatly reduced from the size of its preceding counter­ part. The increased dorsoventral width of the pedicles has de­ creased the size of the lateral foramen; the vertical diameter of the right one is only 36 mm, compared with 46.6 mm on the sixth cervical. At the greatest transverse width of the posterior face of the centrum there are demifacets for the articulation of the first rib. These features are best observed when this verte­ bra is placed in normal articulating position with the first tho­ racic vertebra, which has a corresponding demifacet opposite each of those on the last cervical. Thoracic Vertebrae: Seven thoracic vertebrae are pre­ served, among them the first three and the fifth, seventh, eighth, and the possible 15th in the thoracic series. Lacking di­ rect evidence of the total number of thoracic vertebrae nor­ mally present in E. whitmorei, we assume that there were no fewer than 12, based upon Kellogg's (1968:175) comments about the number of dorsal (thoracic) vertebrae in fossil mys­ ticetes from the Miocene Calvert Formation in Maryland and Virginia. Eomysticetus whitmorei appears to have had at least 15 pairs of ribs, however, and thus would have had 15 thoracic vertebrae. As shown in Table 4, the anteroposterior diameter of the centrum in the preserved thoracic vertebral series increases from the first through the last, and the transverse diameters are consistently greater than the vertical diameters. Ratios of lengths and widths of the thoracic vertebral centra are dia­ grammed in Figure 29 and are interpreted in the "Discussion" section below. The anteroposterior length of the centrum of the first thoracic vertebra (Figure 17A) is only 10 mm greater than that of the seventh cervical vertebra. The spinous process is preserved on the first thoracic vertebra, but the right and left sides of its neu­ ral arch are missing. The spinous process slopes anteriorly at an angle of approximately 60° from the plane of the roof of the neural arch, and from that plane to its tip the process is 81 mm in vertical thickness. On the lateral margins of the centrum, the anterior demifacet for the capitulum of the first rib is situated below the level of the posterior demifacet for the capitulum of the second rib. The second thoracic vertebra (Figure 17B) is essentially complete, with only its left transverse process, the margins of its neural arch, and the posterior edge of its spinous process showing appreciable degrees of erosion. The profile of the neu­ ral canal continues the triangular shape that is present on the seventh cervical. The apex of the neural spine is intact, the ver­ tical height of this process being 137 mm, 56 mm greater than that of the first cervical. The pedicles are markedly different in their transverse dimensions, the right one being 33 mm and the left one 29 mm. They are much narrower anteroposteriorly, the right one being 22.5 mm and the left one 20 mm in greatest di­ ameter. Arising from the base of the pedicle, the transverse pro­ cess projects dorsolateral^ at an angle of approximately 35° from the center of the anterior face of the centrum and anteri­ orly to a point about 40 mm beyond the centrum, placing the articular facet for the tuberculum of the second rib on a level with the roof of the neural arch and well anterior to the anterior face of the centrum. The somewhat elliptical prezygapophysial NUMBER 93 331 3 cm X :*&*% d.a. p.a. i * '- 't/''''- '• - "^' * k'^,. B FIGURE 16.—Eomysticetus whitmorei, new genus and new species, holotype, ChM PV4253, cervical vertebrae, anterior views: A, sixth cervical; B, seventh cervical. (For abbreviations, see Figure 15.) facet is situated on the dorsal surface of the diapophysis, form­ ing a shallow depression approximately 37 mm in length. The postzygapophysis extends 9 mm beyond the posterior face of the centrum. Its ventral articular facet is elliptical in form and measures 26 x 16 mm. On the lateral margins of the centrum, the anterior facet for the capitulum of the second rib is located slightly ventral to the posterior facet for the capitulum of the third rib. The third thoracic vertebra (Figure 17c) is similar in form and preservation to the second one, both having a neural spine 332 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY FIGURE 17.—Eomysticetus whitmorei, new genus and new species, holotype, ChM PV4253, thoracic vertebrae: A,B, first and second thoracics, respectively, in anterior view; c, second and third thoracics (left to right) in lateral view. (cr.=anterior demifacet for capitulum of rib; ct.=facet for tuberculum of rib; df.=posterior demifacet for capitulum of rib; mp.=metapophysis; ns.=neural spine; pz.=postzygapophysis; tr.=transverse process.) that is comparatively slender anteroposteriorly. The dorsal tip of the spinous process is missing only a millimeter or two, its vertical dimension being approximately 143 mm with that al­ lowance. Posteriorly, its base has two facets to receive the ante­ rior end of the base of the neural spine of the fourth thoracic. The neural canal is triangular in anterior view, but its apex is not as acute as that of the second thoracic, from which it differs also by being 3 mm less in vertical diameter. The transverse processes project dorsolaterally, but in this vertebra the axis of each process passes through the ventral margin of the centrum, rather than through the center as in the second thoracic. The right process extends anteriorly approximately 38 mm beyond the anterior face of the centrum. It is the best preserved of the two, the extremity of the left one having been eroded away. NUMBER 93 333 ' ' & • • * % • * ' . C FIGURE 18.—Eomysticetus whitmorei, new genus and new species, holotype, ChM PV4253, thoracic vertebrae, anterior views: A, fifth thoracic; B, seventh thoracic; c, eighth thoracic. The articular facet for the tuberculum of the right third rib is well preserved and measures 32 x 26 mm. The center of this facet is approximately 6 mm above the level of the roof of the neural canal. On the centrum the anterior facet for the capitu­ lum of the third rib is positioned slightly more ventrally than the posterior facet for the capitulum of the fourth rib. As in the second thoracic, the transverse diameter of the right pedicle of the neural arch exceeds that of the left, the right being 21.5 mm and the left approximately 18 mm, with allowance for erosion of the external surface of the left pedicle. The prezygapophys- ial facets are not well preserved along their outer edges, but the left one is sufficiently intact to determine its length as 43 mm. The postzygapophysis extends approximately 4 mm beyond the posterior face of the centrum. Its ventral facet is too badly eroded to yield a measurement. The vertebra that we have identified as the fifth thoracic (Figure 1 8A) qualifies for that position by virtue of the antero­ posterior length of its centrum (76 mm) compared with that of the other thoracics (see Table 4). In this specimen, the right side of the neural arch and approximately one-half of the verti­ cal diameter of the spinous process are preserved. The anterior profile of the neural canal is not so triangular as it is on the sec­ ond and third thoracic vertebrae, the vertical diameter being 4 mm less than that of the third thoracic (see Table 4). The right 334 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY TABLE 4.—Measurements (in mm) of the thoracic vertebrae of Eomysticetus whitmorei, new genus and new spe­ cies, ChM PV4253. (Parentheses indicate estimated measurements.) Character Anteroposterior length of centrum Transverse diameter of centrum, anteriorly Vertical diameter of centrum, anteriorly Minimum anteroposterior length of pedicle of neural arch Transverse diameter of neural canal, anteriorly Vertical diameter of neural canal, anteriorly Distance between ends of transverse processes Dorsal edge of metapophysis to ventral face of centrum, anteriorly Tip of neural spine to ventral face of centrum, anteriorly Transverse diameter of centrum, posteriorly Vertical diameter of centrum, posteriorly Anteroposterior length into transverse diameter of centrum, anteriorly Anteroposterior length into vertical diameter of centrum, anteriorly T.l 49 96 82 19 - - - - - 92 82 1.97 1.68 T.2 56 90 78 20 63 40 190 27 176 94 82 1.55 1.39 T.3 63 91 81 27 72 37 (176) 30 (177) 94 83 1.52 1.35 T.5 76 93 76 - (64) 33 - - - - (84) 1.22 1.00 T.7 81 96 (79) 42 67 32 (162) 31 (190) (106) (84) 1.19 (0.98) T.8 84 102.6 81 49 (59) 32 - - - 102 (85) 1.22 0.96 T15(?) 93 (101) 87 - - - - - - 110 94 1.09 0.94 transverse process projects dorsolaterally, but not at as great an angle as those of the second and third thoracic vertebrae. Un­ like the latter vertebrae, the pedicles are not angled outward with the transverse process but instead arise almost directly vertical to the plane of the dorsal surface of the centrum. The end of the transverse process is badly eroded, but it is clear that the articular facet for the tuberculum of the fifth rib is well above the roof of the neural canal. The anterior end of the prezygapophysis is broken off, its preserved portion extending approximately 15 mm beyond the anterior face of the centrum. The posterior end of the postzygapophysis also is missing, but apparently it did not project much beyond the posterior face of the centrum. Its ventral facet is canted ventrolaterally and ap­ pears to have been more rounded than its counterparts on the second and third thoracic vertebrae. The seventh thoracic vertebra (Figure 18B) continues a trend of progressive increase in the anteroposterior diameter of the centrum in the thoracic series (see Table 4). Concomitantly, it continues a gradational trend in the decrease in the vertical di­ ameter of the neural canal. Much of the spinous process is pre­ served, but it lacks the dorsal tip and portions of the anterior and posterior edges. The left transverse process is badly eroded but extends at least 9 mm beyond the anterior face of the cen­ trum. The right transverse process is entirely missing. The postzygapophysis extends only slightly beyond the posterior face of the centrum. The eighth thoracic vertebra (Figure 18c) has most of its spinous process but lacks the tip and sections of the anterior and posterior margins. The left transverse process is preserved but is badly eroded and is incomplete anteriorly. Both the right transverse process and its pedicle are absent. At this position in the thoracic series, the transverse processes are still situated high on the flanks of the neural arch, the articular facet for the tuberculum of the rib being about level with the roof of the neural canal. As seen in Table 4, this vertebra reflects three progressive anterior-to-posterior trends in certain dimensions of the thoracic vertebrae: (1) a progressive increase in the an­ teroposterior length of the centrum; (2) a corresponding in­ crease in the anteroposterior dimension of the pedicles of the neural arch; and (3) a reduction in the height of the neural ca­ nal. Commonly seen in cetacean vertebral columns, these trends are nothing more than accommodations for the disparate sizes of the cervical and lumbar series. On the probable 15th thoracic vertebra (Figure 19) the trans­ verse processes are located at a midlateral position on the cen­ trum, clearly indicating that it is one of the last in the thoracic FIGURE 19.—Eomysticetus whitmorei, new genus and new species, holotype, ChM PV4253, probable fifteenth thoracic vertebra: A, anterior view; B, dorsal view. (tr.=transverse process.) NUMBER 93 335 TABLE 5.—Measurements (in mm) of the lumbar vertebrae and probable first caudal vertebra of Eomysticetus whitmorei, new genus and new species, ChM PV4253. Parentheses indicate estimated measurements. Character Anteroposterior length of centrum Transverse diameter of centrum, anteriorly Vertical diameter of centrum, anteriorly Minimum anteroposterior length of pedicle of neural arch Transverse diameter of neural canal, anteriorly Vertical diameter of neural canal, anteriorly Distance across ends of transverse processes Dorsal edge of metapophysis to ventral face of centrum, anteriorly Tip of neural spine to ventral face of centrum, anteriorly Transverse diameter of centrum, posteriorly Vertical diameter of centrum posteriorly Anteroposterior length into transverse diameter of centrum, anteriorly Anteroposterior length into vertical diameter of centrum, anteriorly Lumbar A 126 118 (119) 76 31 - (330) - - (114) (108) 0.94 (0.94) Lumbar B (127) (128) (114) (73) (20) - - - - (127) (128) (1.01) (0.90) First caudal (?) 125 (115) (114) 61 21 36 (258) 195 (252) (123) (HO) 0.92 0.92 series. Furthermore, its very short and blunt transverse process is additional evidence that this vertebra is the last thoracic. This vertebra exemplifies the archaeocete-like characters that are evident in the thoracic vertebrae of E. whitmorei. The entire neural arch, including the spinous process, is missing, as is the left transverse process. Only the posterior half of the right transverse process is preserved, but that portion is sufficient to demonstrate that the transverse processes on this vertebra are short and are positioned midlaterally on the centrum, as in the last seven thoracic vertebrae of the archaeocete Zygorhiza kochii (see Kellogg, 1936:139-142). In the Miocene cetotheres Pelocetus calvertensis (see Kellogg, 1965, pl. 12), Diorocetus hiatus (see Kellogg, 1968, pl. 54), and Thinocetus arthritus (see Kellogg, 1969, pl. 6), the transverse processes of the last three or four thoracic vertebrae extend from the centrum, but they are much more elongate than in either Eomysticetus whit­ morei or Zygorhiza kochii. These latter species also differ from the aforementioned cetotheres by having the articular facet for the tuberculum of the ribs positioned above the level of the roof of the neural canal on the third through the seventh thoracics. In Pelocetus calvertensis and Thinocetus arthritus, two species of cetotheriids for which there is sufficient material for com­ parison, the facet for the tuberculum does not extend dorsal to that level and is usually situated ventral to it (Kellogg, 1965, pis. 5-9; 1969, pl. 6). Lumbar Vertebrae: Only two lumbar vertebrae were found with the holotype skeleton of Eomysticetus whitmorei. Both are missing the neural arch, and in the absence of other lumbar ver­ tebrae, it is difficult to place these vertebrae in their exact posi­ tions in the vertebral column. We therefore believe it best to re­ fer to these vertebrae only as lumbar "A" and lumbar "B." Lumbar vertebra "A" (Figure 20A) is the anteriormost of the two and seems clearly referable to the anterior one-third of the series, judging from the length and form of the right transverse process, the only one preserved on this specimen (see Table 5). The transverse process is elongate and flattened, with a slightly expanded and dorsally curved distal extremity, showing that it did not articulate with a rib. As in vertebrae of the Archaeoceti, the transverse processes of this vertebra are directed ventrolat­ erally, not horizontally as in typical Mysticeti. The centrum is nearly cylindrical, its height and width being nearly equal. Compared with more highly evolved Mysticeti, such as Ce­ totheriidae (e.g., Thinocetus arthritis), the centrum is relatively more elongate, and in this regard it is reminiscent of the lumbar vertebrae of basilosaurine archaeocetes. It is relatively shorter than in dorudontines, its length being somewhat intermediate between basilosaurines and dorudontines. The base of the pedi­ cle is elongate anteroposteriorly and narrow transversely, indi­ cating that the pedicle was typical of the condition in Mysticeti. Lumbar vertebra "B" (Figure 2 0 B ) is from a more posterior location in the vertebral column than the one that we have la­ beled as lumbar vertebra "A." Judging by the disparity in size and shape between it and lumbar vertebra "A," several verte­ brae undoubtedly separated the two. It also suffered from se­ vere weathering and is fractured and lacks its processes. It has a longer centrum, of greater diameter, and this is typical of the more posterior lumbar vertebrae in species of Mysticeti. Its neural canal also is of smaller diameter, in agreement with a narrower nerve cord posteriorly in the body. Also, as with lum­ bar vertebra "A," the centrum is relatively more elongate than is typical in species of Mysticeti. Possible Caudal Vertebra: The posteriormost vertebra of the holotype skeleton is possibly a caudal vertebra (Figure 20c). It is definitely from a position more posterior in the body than any of the aforementioned vertebrae. Its centrum is larger, its neural process shorter, its transverse processes shorter and more ventrally directed, and its neural canal narrower (Table 5). The parts of the ventral surface of the centrum that would have borne facets for articulation with chevron bones have been abraded away; therefore, we cannot definitely place this bone in the caudal series, but it could be either a posteriormost lumbar or an anterior caudal. In its overall morphology, this vertebra resembles posterior lumbar and anterior caudal verte­ brae of archaeocetes (see Kellogg, 1936, fig. 60 for Zygorhiza kochii, and pl. 5 for Basilosaurus cetoides). The dorsal process is longer than those in this part of the col­ umn in B. cetoides, but the metapophyses extend dorsolateral^ as in the same species. The distal end of the intact left trans- 336 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY A 5 cm « . jf< f;,y \ I • 5 cm 5 cm FIGURE 20.—Eomysticetus whitmorei, new genus and new species, holotype, ChM PV4253, lumbar and caudal (?) vertebrae in anterior view: A, lumbar "A"; B, lumbar "B"; C, probable anterior caudal vertebra. (For other abbreviations see Figure 17.) verse process bends ventrally at its extremity, not dorsally as in the lumbar vertebra "A" described above. On the dorsal surface of the transverse process is an oblique crest of bone, represent­ ing a muscle scar, that is typically seen on vertebrae in the ante­ rior part of the caudal series in species of modern Cetacea. Ribs: The ribs of the holotype as they were found in the ex­ cavation were not in anatomical position, and the field occur­ rence did not therefore indicate the sequence in life. We have determined the apparent positions of the ribs using compari­ sons with other fossil and extant Cetacea (see Figure 21). Be­ cause it is typically short and broad in cetaceans, the first rib is the only one that is definitely identifiable. All the other posi­ tions that we indicate are approximations and might be inaccu­ rate by one or two positions. The ribs that have both the tuber­ culum and the capitulum are identifiable as right or left by the posteroventral slope of the facet on the tuberculum, by the an- NUMBER 93 337 FIGURE 21.—Eomysticetus whitmorei, new genus and new species, holotype, ChM PV4253; ribs in anterior view. 338 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY terior twist of the neck and capitulum, and by the direction of the prominence and strut at the change in angle of the shaft of the rib. The posterior ribs, which lack the capitulum and termi­ nate proximally in only an irregular facet, are less certainly identified as right or left; however, except for the posteriormost one, the ribs from the right and left sides are probably identi­ fied correctly. The determination of the sequence of the ribs af­ ter the first rib on each side was based upon (1) progressive lengthening and thickening of the shaft toward the middle of the series, (2) progressive shortening of the neck and reduction in size of the capitulum posteriorly, and (3) loss of head and re­ duction in overall size of the posteriormost ribs. According to these generalizations, there are at least 15 dif­ ferent rib positions among the bones of the skeleton. This judg­ ment assumes that the smallest rib, which is not complete, is a posteriormost rib, and that the slightly mismatched pair that we have identified as the right and left seventh ribs are in fact a pair. This number is equal to the 15 ribs that are present in spe­ cies of Archaeoceti and is more than the typical 12 of cetotheri- ids (see Kellogg, 1968:175). If the mismatched ribs we identify as the right and left seventh ribs in fact represent different posi­ tions, then the holotype of Eomysticetus whitmorei may have had 16 ribs on each side. Only the first rib on the left side is present, and it is missing part of its head. It is short, broad, and flat, its proximal end is strongly curved, and the shaft is short with a definite rugose ar­ ticular facet distally. The probable right second rib is not complete. It is not as broad and flat as the first, and it is definitely longer than that rib, even though it is broken distally. The probable right and left third ribs are not complete. The one from the right side is the better preserved of the two and is longer. The left one seems to match it well, but its surface is badly eroded. Each of these ribs is not so curved as the previ­ ous one, and each has a prominent angle, a prominent tubercu­ lum, and a long head. The possible fourth rib, represented only in the right series, is less curved proximally, is more slender, and has a shorter neck between the tuberculum and capitulum compared with the pre­ vious rib. The probable fifth rib, present only in the right series, is more elongate and has a more cylindrical cross section and a shorter neck than the previous rib. Both the right and left probable sixth ribs are present, and they differ from the previous one in the following ways: they are more elongate, the head is smaller, the neck is shorter and more slender, and the tuberosity at the change in angle is lo­ cated farther distally on the shaft of the rib. We are cautious about the assignment of both ribs to the same location, because the left one has a slightly more slender head. The possible left seventh rib is slightly thicker than the previ­ ous one at the proximal end. The probable left eighth rib differs from the previous one by having a shorter neck. The probable ninth rib is represented on the right side only. It has a shorter neck, and the tuberosity at the change in angle is located farther distally on the shaft of the rib. The rib that we believe to be the 10th is represented on the right side only Compared with the previous rib, the neck is much shorter, the head has a large and flat articulation, and the tuberosity at the change in angle is located farther distally on the shaft of the rib. The possible 11th rib is represented on the right side only. Compared with the previous rib, it has a less prominent tuber­ osity at the change in angle of the shaft, a very short neck, and barely separate tuberculum and capitulum. The possible 12th rib also is represented on the right side only. Compared with the previous rib, the tuberosity at the change in angle of the shaft is less prominent. It is the first rib without a capitulum, and the tuberculum has a rounded, blunt articulation. The possible 13th rib also is represented on the right side only. Compared with the previous rib, it is more slender and has a less prominent tuberosity at the change in angle of the shaft. It also lacks a capitulum, and the tuberculum has a smaller, blunt articulation. The possible 14th rib is represented on the left side only. It is more slender than the previous rib and has a less prominent tu­ berosity at the change in angle of the shaft. Like the previous two ribs, it lacks a capitulum and the tuberculum has even a smaller articulation. The presumed 15th is the smallest of the preserved ribs. It ap­ pears to be from the right side, and it is missing the proximal end. By its curvature, it appears similar to the posteriormost ribs of extant Cetacea. Forelimb: The anterior limb in Eomysticetus whitmorei is represented by a partial right scapula, the right humerus, the right ulna, and the right radius (Figure 22). The scapula is pre­ served only in the area immediately above the glenoid cavity and for a distance of 200 mm along the posterior edge. Al­ though lacking its anterior external and internal edges, the gle­ noid cavity is subovate (60*90 mm), shallow, and somewhat flattened transversely. The articular head and glenoid cavity are distinctly elongate rather than ovoid, as in the Miocene ce­ totheres Pelocetus calvertensis, Diorocetus hiatus, and Thi­ nocetus arthritus, for example. Anteriorly, the articular head extends well forward of the base of the anterior edge and acro­ mion and curves sharply inward. The posterior edge of the scapula rises directly from the posterior end of the articular head and projects backward in a low arc for a distance of 122 mm along the plane from the head to the terminus of the arc. Midway along that distance the edge is 33 mm thick, but in the space of only 40 mm above the edge, the scapula thins to 9 mm in transverse diameter. A similar but more concave arc is de­ scribed by the posterior edge of the scapula of the late Oli­ gocene odontocete Sulakocetus dagestanicus Mchedlidze, 1976 (see Mchedlidze, 1976, fig. 1, table 16) from the northern Caucasus region of Dagestan. NUMBER 93 339 ^ i scapula humerus FIGURE 22 (left).—Eomysticetus whitmorei, new genus and new species, holotype, ChM PV4253, right forelimb in lateral view, with reconstruction of humerus at right. (Dashed lines represent hypothetical configurations.) The humerus is incomplete, lacking the lateral edges of the head, the region of the radial tuberosity, most of the posterior face, and approximately 80 mm of the proximal portion of the medial side. The head is smooth and is directed obliquely later­ ally and posteriorly, the greater portion of it being positioned posterior to the long axis of the bone, which passes through the convergence of the radial and ulnar facets. It is flattened along its external side and is decidedly less globose than the heads of the humeri of Pelocetus calvertensis, Diorocetus hiatus, and Thinocetus arthritus. At the distal end of the humerus, the lat­ eral portion of the juncture of the radial and ulnar facets is pre­ served, permitting an accurate measurement of the length of this bone at its greatest distal extent. As in all non-archaeocete Cetacea, the radial and ulnar facets converge to form an obtuse angle, approximately 125° in this individual. The shaft of the humerus is quite different from those of Neogene mysticetes, being considerably longer, more robust, and of an entirely dif­ ferent configuration. The preserved portion above the ulnar facet is rounded, and immediately above the proximal margin of that facet there is a shallow gutter, above which the bone be­ gins to widen transversely. Anteriorly, the lateral and medial sides of the shaft are greatly flattened distally and converge to form a narrow keel for a distance of approximately 90 mm along the preserved portion of the anterior face. The keel arises from the proximal border of the radial facet and angles upward along a tangent of approximately 55° from the axis of the radial facet, a line approximately paralleling the anterodistal angle from the lower end of the deltoid crest in archaeocete humeri. This feature can thus be termed the anterodistal angle of the hu­ merus in E. whitmorei. Although the region above it is missing in the holotype, the slope of this angle suggests the presence of a prominent deltoid crest in humeri of Eomysticetidae (Figure 22). At a point approximately 148 mm from the distal end of the humerus, the anterior face begins to broaden proximally to­ ward the head. 340 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY TABLE 6.—Measurements (in mm) of the holotype right humerus of Eomys­ ticetus whitmorei, new genus and new species, ChM PV4253. Parentheses indi­ cate estimated measurements. TABLE 7.—Measurements (in mm) of the holotype right ulna of Eomysticetus whitmorei, new genus and new species, ChM PV4253. Greatest length Anteroposterior diameter of head Transverse diameter of head Anteroposterior diameter of shaft in valley above ulnar facet Anteroposterior diameter of shaft, 120 mm above distal tip Transverse diameter of shaft at same level Transverse diameter of anterior face of shaft at same level Transverse diameter of posterior face of shaft at same level Anteroposterior diameter of external portion of radial facet Anteroposterior diameter of distal end at level of anterior margin of radial facet Transverse diameter of distal end 329 110+ 93+ 95 103 63 16 47 54 77 (84) The humerus of E. whitmorei displays a remarkable combi­ nation of archaeocete and mysticete characters but is most sim­ ilar to humeri of Archaeoceti (e.g., Basilosaurus cetoides and Zygorhiza kochii (see Kellogg, 1936:65-66, 161-162)) in three major respects: (1) the greatest anteroposterior diameter of the shaft exceeds that of the distal end; (2) there is a pronounced anterodistal angle characteristic of archaeocetes but unknown in previously described mysticete humeri; and (3) the extreme length of the shaft, compared with humeri in Neogene mystice­ tes (Table 6). The most distinctive non-archaeocete character is the presence of separate, flattened radial and ulnar facets, in­ stead of the rounded trochlea and capitulum that are typical of humeri of archaeocetes and other mammals. That is the differ­ ence between the rotational elbow joint of archaeocetes and the nonrotational joint of the modernized Cetacea. The radius was badly shattered by compaction of sediments but has been restored closely enough to its proper form to yield a reasonably accurate measurement of its greatest length, which is 268 mm. Proximodistally, the anterior and posterior margins of the shaft do not appear to have been as strongly curved as those of the radii of archaeocetes and of Miocene ce- totheriids (see Kellogg, 1936, fig. 72; 1965, fig. 23; 1968, fig. 73; 1969, pl. 3). Although the distal end of the radius is badly crushed and incomplete, the preserved portion shows that it was expanded both transversely and anteroposteriorly. The dis­ tal epiphysis is missing. The proximal end is bent posteriorly to articulate with the radial facet of the humerus. The facet for ar­ ticulation with the ulna was not preserved, and the medial face immediately below the radial facet also is missing. Approxi­ mately 50 mm below the anterior edge of the proximal end, the transverse diameter of the shaft abruptly increases, marking the proximal termination of a broad, flat area that extends proxi­ modistally along the anterior face for a distance of about 65 mm, at which point it diminishes into the normal transverse di­ ameter and rounded form of this face of the shaft. This feature seems to be homologous with the "elliptical rugose area" noted by Kellogg (1936:162) in his description of the radius of Zy­ gorhiza kochii. Kellogg (1936:162) presumed it to be the area for the insertion of the pronator teres muscle, which may still have been slightly functional in Z. kochii, considering the prim- Length of preserved portion (from 13 mm above proximal portion of greater sigmoid cavity to distal end) Greatest diameter of greater sigmoid cavity Greatest transverse diameter of proximal portion of greater sigmoid cavity Greatest transverse diameter of distal portion of greater sigmoid cav­ ity Greatest length of anterior face of shaft below greater sigmoid cavity Anteroposterior diameter of shaft at 85 mm below greater sigmoid cavity Transverse diameter of shaft at same level 327 51 + 38.5 51 + 246 61 31 itive structure of its elbow joint. If present in Eomysticetus, it is doubtful that that muscle would have been functional at all in view of the more highly specialized, nonrotational elbow joint in this form. The ulna (Table 7) is missing the posterior portion of the ole­ cranon process and a large area of its internal side proximal to the shaft. Also missing are the distal sections of the anterior and posterior faces and the distal sections of the medial face. The shaft is transversely compressed and is broadly curved proximodistally. It is transversely expanded at both the distal and the proximal ends. The proximal portion of the greater sig­ moid cavity is slightly concave, the distal portion being broadly convex transversely. As with the radius, the distal epi­ physis is not present. Eomysticetus carolinensis, new species FIGURES 24-28 Cetothere "similar to C. tobieni."—Sanders and Barnes, 1991. DIAGNOSIS.—A species of Eomysticetus separated from E. whitmorei by the following characters: intertemporal region shorter; parietals one-third less in length along midline; apex of supraoccipital extending farther anteriorly (by approximately 38 mm); anterior portion of supraoccipital heavily corrugated; lambdoidal crests more laterally oriented and overhanging tem­ poral fossa; parietals sloping from intertemporal region more laterally and not so vertically directed; sagittal crest of parietals in intertemporal region rounded, not blade-like; squamosal fossa narrower; zygomatic process of squamosa thinner and more divergent from sagittal plane; distal end of zygoma ex­ tending to or slightly behind plane of apex of supraoccipital and lacking a prominent ventral facet for articulation with ju­ gal; dorsomedial side of zygoma not convex but flat to slightly concave; basioccipital crests trasversely thinner, more ventrally extended, and more knob-like; foramen ovale of larger diame­ ter and more elongate; anterior border of glenoid fossa broadly curving, not angular; glenoid fossa more square and deeper an­ teroposteriorly; postglenoid process thicker and deeper; supra­ condylar fossae shallower; intercondylar notch (between occip­ ital condyles) shallower; jugular notch bilobed (not preserved in E. whitmore). NUMBER 93 341 HOLOTYPE.—ChM PV4845, braincase lacking the entire ros­ trum and missing the left exoccipital and squamosal. Collected by Barry Albright and Vance McCollum during the summer of 1988. TYPE LOCALITY.—Found in a ditch in the Irongate subdivi­ sion, Dorchester County, South Carolina, approximately 33.8 km (21 mi.) north of Charleston and about 2.6 km (1.6 mi.) northwest of the type locality of Micromysticetus rothauseni Sanders and Barnes (2002). FORMATION AND AGE.—Chandler Bridge Formation, bed 2, lower Chattian correlative, late Oligocene, nannoplankton zone NP24 (by inference), ca. 28 Ma (Figure 23). ETYMOLOGY.—The specific name recognizes South Caro­ lina as the origin of the holotype. DESCRIPTION.—Lacking the left squamosal and exoccipital regions and all of the ventral portion of the postorbital region anterior to the basisphenoid, the holotype cranium of Eomys­ ticetus carolinensis is not as complete as the holotype of E. whitmorei, but the surfaces are well preserved and the basicra­ nium and most of the dorsal surface of the intertemporal region are present Figures 24-26). The lambdoidal crests are incompletely preserved in the api­ cal region and elsewhere are missing altogether, making it dif­ ficult to determine the original shape of the supraoccipital. Nonetheless, the curvature and angle at which the preserved remnant of the left lambdoidal crest extends posteriorly from the apex of the supraoccipital permit a reasonably close ap­ proximation of the shape of the occipital shield (Figure 24B) , which evidently is narrow and elongate, more so than in E. whitmorei. Also unlike the condition in E. whitmorei, the su­ praoccipital extends beyond the plane of the anterior tips of the zygomae. The dorsal surface of the supraoccipital in E. carolinensis is divided into two elevations, the posterior por­ tion being a shallow depression beginning approximately 40 mm anterior to the posterior margin of the supraoccipital and extending forward to the posterior margin of the parietals. At that point the supraoccipital abruptly ascends the parietals to a plateau formed by the parietals and slopes upward and for­ ward to the apex, a distance of approximately 130 mm. A con­ spicuous rounded eminence on each side of the midline marks the point of contact between the supraoccipital and the pari­ etals during an earlier stage of ontogeny when the bone of the supraoccipital was thinner and more plastic. A prominent me­ dial crest traverses virtually the entire length of the supraoc­ cipital, extending from a point near the apex to the posterior margin of the lower, excavated area, dividing that area into two sections. The crest achieves its maximum height on the plateau formed by the parietals, where it rises almost to the height of the adjacent lambdoidal crests. In this region the crest is flanked by two short crests, each of which is bordered on both sides by an elongate gully-like depression. The result­ ing pattern of alternating ridges and grooves produces a corru­ gated effect over most of the elevated portion of the supraoc­ cipital (Figure 24). ^L Ground level / • . t e n Mile Hill Beds (Late Middle Pleistocene) Pebble bed- FIGURE 23.—Stratigraphic section at type locality of Eomysticetus carolinen­ sis, new species, ChM PV4845. (Not to scale.) The holotype of E. carolinensis represents a mature individ­ ual, as indicated by the smooth articular surfaces of the occipi­ tal condyles. The dorsal margins of the condyles of E. carolin­ ensis lie only slightly below the level of the floors of the squamosal fossae (Figure 27). The vertical diameter of the fo­ ramen magnum exceeds the transverse diameter. The exoccipi­ tal is broad and does not extend very far posteroventrally, and most of the paroccipital process is missing. The lateral wall of the braincase is concave, the squamosal inclining posteriorly to the posterior margin of the lambdoidal crest along the inner flank of the squamosal fossa. The floor of the squamosal fossa is interrupted by the secondary squamosal fossa, a deep, circu­ lar, pit-like depression approximately 18 mm in diameter at its deepest point. The posterior wall of this pit receives a narrow gutter that steeply ascends to the floor of the squamosal fossa and extends to its posteriormost margin. Anterior to the pit, the floor of the fossa is broadly concave. The zygomatic process of the squamosal diverges more abruptly from the sagittal plane, is rather long, is tapered anteriorly, and is not so strongly arched as that of E. whitmorei. It is angled sharply outward from the long axis of the skull and extends forward to a point slightly behind the plane of the apex of the supraoccipital. A relatively small, shallow, and anterodorsally inclined stemo­ mastoid fossa occupies the posterior face of the squamosal near the suture with the exoccipital. On the ventral side of the cranium the basioccipital is well preserved. It is dorsally arched transversely and has large, 342 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY r\ FIGURE 24 (left).—Eomysticetus carolinensis, new species, holotype, ChM PV4845: A, cranium, dorsal view; B, reconstruc­ tion of cranium based upon holotype, dorsal view. (Dashed lines and solid lines in areas of missing bone (see Figure 24A) repre­ sent hypothetical configurations. Abbreviations are explained in "Material and Methods.") somewhat rounded basioccipital crests that are sharply diver­ gent posteriorly and descend prominently from the basicra­ nium. Extending forward from the anterior slope of the basioc­ cipital crest and along the lateral margin of the basisphenoid there are large, elongate, and obliquely oriented interdigitations that form the suture for the hamular process of the pterygoid. The glenoid fossa is large and almost rectangular in form, its transverse width (102.7 mm) nearly equaling its anteroposte­ rior length (106 mm). Slightly posterior to the cranial hiatus, the innermost margin of this fossa forms an acute angle. The glenoid process is elongate transversely and is very narrow an­ teroposteriorly. There is a correspondingly long external audi­ tory meatus. The most noticeable features of the holotype cranium of E. carolinensis are the elongate supraoccipital shield and the long intertemporal region. The preserved portion of this region is 135 mm in length anterior to the apex of the supraoccipital and consists primarily of the parietals, which interdigitate with a narrow posterior extension of the frontals over a broad area an­ teroposteriorly. Dorsally, a finger of the frontals extends back­ ward to within 60 mm of the apex of the supraoccipital. Later­ ally, the parietals project forward for a distance of at least 110 mm anterior to the apex, overlapping the posterior end of the frontals for a distance of at least 50 mm. Although only a 78 mm portion of the frontals is preserved, the degree to which the left side of this section has begun to flare outward anteriorly in­ dicates that a forward extension of approximately 50 mm along a line of intensifying outward curvature would bring it into NUMBER 93 343 FIGURE 25 (left).—Eomysticetus carolinensis. new species, holo­ type, ChM PV4845: A, cranium, ventral view; B, reconstruction of skull based upon holotype, ventral view. (Dashed lines and solid lines in areas of missing bone (see Figure 24A) represent hypotheti­ cal configurations. Abbreviations are explained in "Material and Methods.") contact with the posterior margin of the supraorbital processes of the frontal. We estimate that the distance between the apex of the supraoccipital and the level of the posterior edges of the supraorbital process was approximately 165 mm. The distance from the posterior margin of the temporal fossa to the apex of the supraoccipital is about 195 mm; thus, the anteroposterior length of the temporal fossa in this specimen can be conserva­ tively estimated as approximately 360 mm. Although no portion of the rostrum was preserved, we be­ lieve that the nasals of E. carolinensis must have been elon­ gated as in E. whitmorei, and it is not unreasonable to think that the rostrum was long and narrow like that of E. whitmorei. Discussion COMPARISONS WITH OTHER CETACEAN TAXA General evolutionary trends within the suborder Mysticeti are toward progressively large body size, relatively large heads, short necks, grooves in the throat region, and telescoping of the cranial elements with an emphasis on movement of occipital bones anteriorly over the braincase. In its forward progress through the various grades of telescoping (Miller, 1923), the acute anterior margin of the supraoccipital shield seems to have wedged between the parietals and gradually forced them apart, producing well-developed lambdoidal crests that usually over­ hang the temporal fossae in Oligocene and primitive mystice­ tes. In most later mysticetes of Neogene time the dorsal portion of the parietals has been reduced to a small area in the anterior region of the skull roof. In Eomysticetus whitmorei and E. car­ olinensis the parietals occupy a prominent portion of the inter­ temporal region of the skull. The braincase of E. whitmorei shares more characters with archaeocetes than does any other described mysticete skull. The same could probably be said for E. carolinensis were more of the skull preserved. Comparison of the skulls of the late Eocene archaeocete Zygorhiza kochii, the late Oligocene Eo­ mysticetus whitmorei, the late Oligocene "Mauicetus" lopho- 344 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY 5 cm * Soc Eoc FIGURE 26 (left).—Eomysticetus carolinensis, new species, holotype, ChM PV4845: A, cranium, right lateral view; B, reconstruction of skull based upon holotype, right lateral view. (Dashed lines represent hypothetical configurations. Abbrevia­ tions are explained in "Material and Methods.") 10 cm cephalus Marples, 1956, the early Miocene Aglaocetus moreni (Lydekker, 1894), the late Miocene Cetotherium rathkii (Brandt, 1873), and the late Oligocene toothed mysticete Aetio- cetus cotylalveus Emlong, 1966, shows a general progression from the non-telescoped archaeocete skull represented by Zy­ gorhiza kochii through the telescoping of the cranial elements in the grades represented by Eomysticetus, "Mauicetus lopho­ cephalus," Aglaocetus, and Cetotherium. The most striking evolutionary changes in the crania of those four taxa are the shortening of the intertemporal region and a corresponding re­ duction in the length of the nasal bones. Although more primi­ tive in rostral characters and presence of teeth, Aetiocetus coty­ lalveus actually has a more highly telescoped cranium than does either Eomysticetus or Mauicetus, and it really does not fit into the progression of telescoping in the mysticetes. That ob­ servation lends strong support to Kellogg's (1969:1) doubt that Aetiocetus was an antecedent of the baleen-bearing whales and to the suggestions of Barnes (1987, 1989) and Barnes et al. (1995) that it is unlikely that any of the aetiocetid mysticetes were involved with the ancestry of modern mysticetes. True baleen-bearing Mysticeti were contemporaneous with the Aeti- ocetidae and in fact occur even earlier in the fossil record than any known aetiocetid. Those facts, coupled with the presence of autapomorphies in aetiocetids that are not shared with the earliest baleen-bearing Mysticeti (for example the compara­ tively shorter and wider intertemporal region of the Aetio- cetidae), confirm that all known Aetiocetidae were relict taxa. Likewise, Eomysticetus whitmorei and E. carolinensis were in their time also relict taxa, for they have cranial characters that are more primitive than some earlier-occurring baleen-bearing mysticetes. It seems most probable that the Aetiocetidae and the Mammalodontidae are side branches of Mysticeti in which baleen was never developed, the retention of teeth into adult­ hood perhaps being a paedomorphic character that became firmly entrenched. Whether the same can be suggested for the third family of toothed mysticetes, the Llanocetidae, is a ques­ tion that must await knowledge of the cranial characters of spe­ cies in that group. Among currently known fossil and living mysticetes, the ex­ tremely long nasal bones in Eomysticetus whitmorei are ap­ proached in length by those of "Mauicetus" lophocephalus (see Crowley and Barnes, 1996). A comparison of the measure­ ments of the nasals of E. whitmorei with those of the holotype of " M " lophocephalus (see Marples, 1956:568) can be decep- zps FIGURE 27.—Eomysticetus carolinensis, new species, holotype, ChM PV4845, cranium, posterior view. (Abbreviations are explained in "Material and Meth­ ods.") tive if the differences in sizes between the respective skulls are not taken into account. The nasal bones in E. whitmorei are ap­ proximately 300 mm long, whereas those of the holotype of "M." lophocephalus are given (by Marples) as 350+ mm in length. The distance between the lateral margins of the exoc­ cipitals is 495 mm in " M " lophocephalus but only 340 mm in E whitmorei, demonstrating that the holotype skull of " M " lo­ phocephalus is approximately 30% larger. Consequently, if the skull of Eomysticetus whitmorei were as large as the New Zealand specimen, the nasals would measure about 393 mm, exceeding those of the latter. Compared directly, the length of the nasals is 88% of the distance between the lateral margins of the supraoccipitals of Eomysticetus whitmorei, but only 71% of the same distance in the holotype skull of " M " lophocephalus. Unfortunately, we find that comparisons of the nasal bones of these two specimens based upon the original description of "M" lophocephalus seem uncertain at best because of the dis­ crepancy between the figured length of the nasals in Marples's (1956:567, fig. lc) reconstruction and their apparent length in photographs A and B of his plate 1. In the reconstruction, the nasals and the premaxillae are shown to terminate at the ante­ rior edge of the frontals, whereas in both of the two published photographs (A, the prepared specimen, and B, the skull in place) the nasals appear to extend posteriorly onto the frontals as in Eomysticetus and other, more typical baleen whales. Not having seen the holotype of " M " lophocephalus, we cannot venture a qualified opinion as to which of the two alternatives is correct. Apparently that question will never be settled, be­ cause the holotype cranium of " M " lophocephalus has been lost (E. Fordyce, pers. comm., July 1994). Among other ceta­ ceans, the length of the nasal in Eomysticetus whitmorei appar- 345 ently is matched by that of Remingtonocetus harudiensis (Ku­ mar and Sahni, 1986:331-332, figs. 3, 4). The anterior portion of the nasals are missing in the figured paratype of the latter taxon, but their length was estimated as 310 mm (Kumar and Sahni, 1986:333, table 1). The presence of a temporal crest on the dorsal surface of the supraorbital process is one of the defining features of mystice­ tes. In Odontoceti, the temporal line is at the posterior margin of the process. The fact that the temporal crest in E. whitmorei merges with the middle part of the posterior margin of the su­ praorbital process indicates that this is a very primitive position for the structure. The intertemporal region is elongate and nar­ row, exceedingly so in comparison with most other fossil and extant Cetacea. The structure of the intertemporal region of E. whitmorei is a primitive condition for a cetacean, especially for a baleen whale. The prominent sagittal crest on the upper part of the occipital shield also is present in some species of the family Cetotheriidae (e.g., Micromysticetus rothauseni Sanders and Barnes, 2002). The condition of the paroccipital process— being thin laterally and only extending laterally as far as the middle part of the postglenoid process—is primitive among Cetacea and also is present in extant gray whales (Eschrichtius robustus). The pit-like secondary squamosal fossa, described above, is known to occur only in E. whitmorei and E. carolin­ ensis. For a mysticete, Eomysticetus whitmorei has a relatively small cranial hiatus, another primitive character. The periotic of E. whitmorei closely resembles that of Mi­ cromysticetus rothauseni (Sanders and Barnes, 2002, figs. 12, 13) in having a large, flattened, hatchet-shaped anterior process and a short, stubby posterior process. That form is very similar to the configuration of the periotic of the archaeocete Zy­ gorhiza kochii. The dentary of Eomysticetus whitmorei is unique among known mysticetes, the posterior end being mas­ sive in size compared with Neogene forms. In its shape, dors­ oventral height, anteroposterior length, and elevation of the end of the posterior edge above the anterior edge, the coronoid pro­ cess of E. whitmorei is more nearly like that of an archaeocete than a mysticete and thus appears to represent a transitional stage between the configuration of the coronoid in Eocene ar­ chaeocetes, such as Zygorhiza kochii, and that of the Miocene cetotheres, such as Parietobalaena palmeri. In lateral aspect, the dentary displays the parallel dorsal/ventral profiles charac­ teristic of Mysticeti. The vertebral column ofE. whitmorei also exhibits features intermediate between archaeocetes and mysticetes. Interest­ ingly, the vertebrae are more like those of the archaeocete Zy­ gorhiza kochii than their counterparts in other baleen whales. In the third through the seventh thoracic vertebrae the articular facet for the tuberculum is situated above the level of the roof of the neural canal, as in Z. kochii. The transverse processes of the lumbar and caudal vertebrae bend ventrolaterally as in Ar­ chaeoceti, rather than extending horizontally as in more highly evolved Mysticeti. In size and relative length/width ratios of their centra, the thoracic, lumbar, and caudal vertebrae ofE. 346 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY T Thoracic v e r t e b r a e L Lumbar v e r t e b r a e C C a u d a l v e r t e b r a e O-'-CVJ O * - O T - C M > - w c j * i n i D s c o o ) r - ' - ' - T - N n ' s t i n i D N ( o o ) r . ' - ' " N c o ^ i q ( O N i o o ) T - i T - i - ^ . Y h h h ' H h h h ' h ' h h J J J J j J J j j J J d O O d O O O O O O O O FIGURE 28.—Curves of mean values of vertebral width/length ratios of five specimens of extant Kogia breviceps and four specimens of extant Kogia simus. whitmorei are intermediate between those of dorudontine ar­ chaeocetes and cetotheriid mysticetes (Figure 29). Unpublished studies initiated by Sanders have shown that the two living species of the odontocete genus Kogia (K. breviceps and K. simus) have distinctly different vertebral length/width ratios (Figure 28), making it possible to identify isolated verte­ brae of these two taxa. We have used this method in compari­ sons of the vertebrae of Eomysticetus whitmorei with those of selected archaeocetes and mysticetes to see how those forms might be sorted out on the basis of vertebral ratios. In Figure 29 we compare the curves of ratios obtained by dividing the trans­ verse width of the anterior end of the centrum by the anteropos­ terior length of the centrum in the thoracic vertebrae of E. whit­ morei, three Eocene archaeocetes (Zygorhiza kochii (Reichenbach, 1847), Dorudon stromeri Kellogg, 1936, Basilo­ saurus cetoides (Harlan, 1845)), and three Miocene cetotheres (Pelocetus calvertensis Kellogg, 1965, Thinocetus arthritus Kellogg, 1969, and Halicetus ignotus Kellogg, 1969). The tho­ racic vertebrae were selected because only two lumbar verte­ brae and one caudal vertebra are available for Eomysticetus. As illustrated by the curve for the thoracics of Basilosaurus ce­ toides, lower ratios reflect elongate centra; thus, the low range of the extremely elongate vertebrae of that taxon does not per­ mit plotting of the entire thoracic series within the vertical lim­ its of Figure 29, the ratio for the 15th thoracic vertebra being 0.49. As seen in Figure 29, the anterior portion of the curve for the thoracic vertebrae of Eomysticetus whitmorei falls between those of the archaeocetes Zygorhiza kochii and Basilosaurus cetoides below it, and those of the three cetotheriids above it. From the fifth thoracic vertebra posteriorly through the sev­ enth, eighth, and 15th (?), the Eomysticetus whitmorei curve most closely follows the curves of the two specimens of Zy­ gorhiza kochii (USNM 4679, USNM 11962) for which mea­ surements of thoracic vertebra were available (Kellogg, 1936: 143). Noting that the anterior portions of the curves of the three cetotheres form a discrete group separate from the curves of Eomysticetus whitmorei, Zygorhiza kochii, and Basilosaurus cetoides, we suggest that vertebral width/length ratios may be useful as indicators of broad phenetic relationships. In Figure 29 the first six thoracic vertebrae furnish the best indices for comparison of the taxa shown. Of the eight individ­ uals represented, the curves of three of them (Pelocetus, Hal­ icetus, and one of the Zygorhiza specimens) converge within the range of 1.34 and 1.37 at the seventh thoracic vertebra. At the eighth thoracic all but three of the curves (Basilosaurus ce­ toides, Thinocetus arthritis, and Dorudon stromeri) converge between 1.20 and 1.26. Thereafter, the curves diverge again as they approach the lumbar vertebrae. The position of the eighth thoracic as the focal point for five of the eight curves shown in­ dicates that it is in this vertebra that dramatic increases in the anteroposterior length of the centrum begin to take place. In the series of 15 thoracic vertebrae that is characteristic of the ar­ chaeocetes reviewed by Kellogg (1936), the eighth one is at the direct numerical center of the series. Anterior to that point, the first seven thoracics tend to reflect the shorter anteroposterior lengths of the cervical vertebrae, whereas those posterior to the eighth thoracic reflect the proportionately greater lengths of the lumbar centra. Recognition of the eighth thoracic vertebra as the commencement point of anteroposterior lengthening of the centra in the thoracic vertebrae of ancestral cetaceans (i.e., the Archaeoceti) suggests that the reduction in the number of tho- NUMBER 93 347 2.70 2.60 2.50 2.40 2.30 2.20 2.10 2.00 1.90 1.80 1.70 1.60 1.50 1.40 1.30 1.20 1.10 1.00 0.90 CO -* H (£> co 0) O i - CM CO to FIGURE 29.—Curves of width/length ratios of holotype thoracic vertebrae of Eomysticetus whitmorei, new genus and new species (late Oligocene); the late Eocene archaeocetes Basilosaurus cetoides, Dorudon stromeri, and Zygorhiza kochii; and the middle Miocene cetotheres Halicetus ignotus, Pelocetus calvertensis, and Thinocetus arthritus. (Lower values reflect more elongate vertebrae. Dashed lines and widely separated symbols indicate absence of specimens at those points in the column. See "Discussion" for sources of measurements used in these curves.) racic vertebrae from 15 in archaeocetes to 12 in cetotheres took place in the posterior region of the thoracic series. The reason for that reduction is probably rooted in the adaptive economics of cetacean evolution. Among modern mysticete taxa, Balaenoptera musculus and B. physalus have 15 thoracic vertebrae, B. borealis has 13-14, B. acutorostrata has 11, and Eschrichtius robustus, Megaptera novaeangliae, and Eubalaena glacialis all have 14 thoracic vertebrae (Kellogg, 1968:175). Kellogg (1968:175) also noted that "skeletons of adult Recent mysticetes are not only larger, with one exception, but are also comprised of more vertebrae than the Calvert Miocene cetotheres. This increase in the num­ ber of vertebrae occurs notably in the caudal series." In view of the numerous affinities that Eomysticetus whitmorei shares with archaeocetes, including certain morphological features of the thoracic vertebrae, it is not entirely surprising to find that this animal apparently carried a higher complement of thoracic vertebrae than the Miocene cetotheres that it preceded, and that they would be closer in number to the 15 thoracic vertebrae present in both Basilosaurus cetoides and Zygorhiza kochii 348 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY (Kellogg, 1936:46, 143). The presence of 15 thoracic vertebrae in this archaic Oligocene mysticete seems almost certainly to be a primitive character, quite unrelated to the increased num­ ber of thoracic vertebrae in extant mysticetes as discussed by Kellogg (1968:175). That premise is supported by the fact that the heads of the ribs of Eomysticetus have capitular articulations with the centra in the first through the ninth thoracic vertebrae, whereas capitu­ lar articulation among the extant balaenopterids has been re­ duced to the first three or four pairs of anterior ribs (Kellogg, 1968:175). In the late Eocene dorudontine archaeocete Zy­ gorhiza kochii, 10 of the anterior ribs have both the capitular and tubercular articulations (Kellogg, 1936:167), and in the Calvert cetotheres the heads of only seven or eight of the ante­ rior ribs have the two articular surfaces (Kellogg, 1968:175). The ribs of Eomysticetus whitmorei are of a normal cetacean type. They are neither pachyostotic nor osteosclerotic. They bow strongly laterally, indicating a full-bodied animal. The an­ teriormost rib is short and broad as is typical of derived Ceta­ cea. The most striking feature of the forelimb in Eomysticetus whitmorei is the large size of the humerus in comparison with the radius and the ulna, a condition more characteristic of Ar­ chaeoceti than of either the Mysticeti or the Odontoceti, the hu­ merus in archaeocetes being considerably longer and more ro­ bust than the radius and the ulna. In all known Neogene Cetacea, the humerus is much shorter than the radius and the ulna, reflecting evolutionary trends toward the shortening of the humerus as the forelimb becomes more highly specialized as a flipper-like structure. As seen in Table 8, the relative pro­ portions of the humerus and the radius and ulna in Eomystice­ tus whitmorei are intermediate between those of the middle Eocene archaeocete Basilosaurus cetoides (USNM 4675; Kellogg, 1936) and the middle Miocene cetothere Thinocetus arthritus (USNM 23794; Kellogg, 1969). Archaic limb propor­ tions apparently were characteristic of many late Oligocene od­ ontocetes as well. The humerus is longer than the radius and ulna (130|100,102) in the primitive odontocete Sulakocetus dagestanicus Mchedlidze, 1976 (see Mchedlidze, 1976:50-51) and greatly exceeds the length of the lower arm bones in the squalodontid Kelloggia barbarus Mchedlidze, 1976 (see Mchedlidze, 1976, pl. 26) from Azerbaijan. More-advanced trends are seen in two other late Oligocene odontocetes from the Caucasus region. The humerus and radius-ulna complex are about equal in length in Ferecetotherium kelloggi Mchedlidze, 1970 (Mchedlidze, 1976:15, pl. 5), a sperm whale (see Barnes, 1985), and the humerus is slightly shorter in Oligodelphis azer- bajdzanicus Mchedlidze and Aslanova, 1968 (Mchedlidze, 1976, pl. 11), a probable kentriodontid dolphin (see Barnes, 1985). In both Mysticeti and Odontoceti, therefore, late Oli­ gocene time appears to have been a period of transition be­ tween the archaeocete type of forelimb and the various kinds of more highly adapted limb structure of Neogene Cetacea, the TABLE 8.—Greatest proximodistal lengths of humerus, radius, and ulna in the late Eocene archaeocete Basilosaurus cetoides (USNM 4675; Kellogg, 1936:64, 67, 68, tables 13-15), the late Oligocene mysticete Eomysticetus whitmorei (holotype, ChM 4253), and the middle Miocene mysticete Thinoce­ tus arthritus (USNM 23794; Kellogg, 1969:9, 10, tables 9-11), all from North America. Specimen Basilosaurus cetoides (USNM 4675) Eomysticetus whitmorei (holotype, ChM PV4253) Thinocetus arthritus (USNM 23794) Humerus Radius Ulna 490± 329 260 250 268 385 334 327 398 shortened humerus having evolved convergently in the two ex­ tant groups. Eomysticetus whitmorei exemplifies a more primitive grade of telescoping than the late Oligocene species of Mauicetus of New Zealand. Apparently, it was one of the last members of a line that preceded Mauicetus, and it probably demonstrates the general appearance of the skull in the earliest baleen-bearing mysticetes. The progenitors of the line that it represents would seem quite evidently to have been forms that were in evolution­ ary transition from Archaeoceti to Mysticeti. The Archaeoceti is the most archaic suborder of Cetacea and includes three families, ranging from the most primitive Proto- cetidae through the more highly evolved Remingtonocetidae and Basilosauridae. Typical Eocene forms are characterized principally by nontelescoped skulls, the nares being situated anteriorly on the anterior part of the rostrum, and by a distinc­ tive dental formula and morphology. The Protocetidae are the oldest and most primitive archaeocetes and include early and middle Eocene animals from North Africa, India, and Pakistan in the Tethys region (e.g., Gingerich et al., 1983, 1990, 1995) and two new protocetids from Georgia (Hulbert et al., 1998) and South Carolina (Geisler et al., 1996). The South Carolina animal is currently under study at The Charleston Museum. All were less than 4 m long, had the normal mammalian dental for­ mula, had fully movable (but reduced) hind limbs, and had the petrosal still located within the braincase. The Basilosauridae include the medium-sized, generalized Dorudontinae and the more divergent, large Basilosaurinae. Species in this family have been discovered around the world, but no unquestionable basilosaurid represented by a skull has been recorded in rocks younger than late Eocene age. Like the Protocetidae, their skulls were not telescoped, but their molars and premolars were more highly evolved because none were three-rooted, there were accessory denticles on the anterior and posterior edges, and because M3 was lost. Generalized dorudontines may have been the ancestors of the mysticetes and the odontocetes (Barnes and Mitchell, 1978; Fordyce, 1980; Barnes and San­ ders, 1996). Although most of the major steps in cetacean evolution are represented in the fossil record, previously described material has not included skulls evincing transitional grades between ar­ chaeocetes and mysticetes or between archaeocetes and odon­ tocetes. Characters derived from archaeocetes are evident in NUMBER 93 349 early members of both of the modern suborders (e.g., Barnes and Mitchell, 1978), but the long absence of skulls of advanced archaeocetes showing trends toward specific odontocete or mysticete characters has led some authors to dismiss the Ar­ chaeoceti as being the possible ancestors of the living Cetacea. Kleinenberg and Yablokov (Kleinenberg, 1958, 1959; Klein­ enberg and Yablokov, 1958; Yablokov, 1964) have argued that modern baleen and toothed whales must have had separate ori­ gins because they are such dissimilar animals now and because they both differ so greatly from archaeocetes. Rice (1966, 1984) and some authors of other general works have accepted such arguments and have classified the archaeocetes, odonto­ cetes, and mysticetes as separate mammalian orders. Van Valen (1968) rebutted the arguments for triphyly by Yablokov, Kleinenberg, and others, and his conclusion that archaeocetes should be accepted as the ancestors of modern whales has been supported by statements by Fordyce (1980), Gaskin (1982), and Rothausen (1985). Proponents of the theory of polyphyletic origin of cetaceans have unknowingly rested much of their case upon unsound stratigraphic ground. Yablokov and others have argued that the "earliest" odontocetes from Charleston, South Carolina, i.e., Agorophius pygmaeus (Muller, 1849) and Xenorophus sloanii Kellogg, 1923, which had figured prominently in paleontologi­ cal analyses, were late Eocene in age, contemporaries of the ar­ chaeocetes, and therefore could not possibly be considered as their descendants; however, the deposits that produced these early odontocetes have now been reinterpreted as late Oli­ gocene in age (Whitmore and Sanders, 1976), which makes them millions of years younger than ancestral Eocene archaeo­ cetes and places them in an intermediate position both chrono­ logically and anatomically. Unfortunately, few summary arti­ cles have incorporated that information. One of the major shortcomings of Yablokov and Kleinen- berg's triphyletic argument is that the extremely different char­ acters of modern mysticetes and odontocetes show only that they have diverged in their evolutionary history, not necessarily that they had separate origins. Additionally, many of those au­ thors' arguments are simply based upon incorrect or incom­ plete data. Because archaeocetes are known only as fossils, skeletal structures are the only presently available characters that are useful when comparing them with living odontocetes and mysticetes. The early, primitive odontocetes and mystice­ tes have few of the skeletal features that some mammalogists, working only with living taxa, often consider to be important and/or diagnostic for the living groups. When considering only the cetaceans of the Oligocene, however, we find odontocetes and mysticetes that are remarkably similar to one another as well as to archaeocetes, and that, in fact, have many osteologi­ cal characters that are intermediate between archaeocetes and primitive members of the other two suborders. Of these, Eomysticetus whitmorei provides the most dramatic example of shared archaeocete and mysticete characters and should put to rest any further claims of polyphyly in cetacean phylogeny. THE PRIMITIVE PHYLOGENETIC POSITION OF Eomysticetus whitmorei Eomysticetus whitmorei was demonstrably a baleen-bearing mysticete. There are no dental alveoli on its palate nor on its dentaries, the palatal surface of its maxilla has sulci that indi­ cate the presence in life of blood vessels that would have nour­ ished baleen, the horizontal rami of the dentaries are elongate and parallel-sided, and a line of nutrient foramina along the dorsal (gingival) border of each dentary marks the former row of mandibular dental alveoli and their associated nutrient fo­ ramina. All of these features are typical of baleen-bearing mys­ ticetes that use a bulk-feeding mode. Eomysticetus whitmorei represents a previously undocu­ mented stage in the evolutionary history of baleen-bearing mysticetes, and at present it and E. carolinensis are the only named members of a previously unreported clade of Mysticeti recognized herein as the family Eomysticetidae. Other primi­ tive baleen-bearing mysticetes of a similar, but slightly more derived grade of evolution have been reported, but as yet most of them are either unnamed and/or uncertainly assigned to a family. Some or all of these may, upon future study, be deter­ mined to belong as well to the family Eomysticetidae. Autapomorphies that define Eomysticetus whitmorei and the family Eomysticetidae (Figure 30) include (1) the extremely narrow intertemporal region (not to be confused with the an­ teroposteriorly elongate intertemporal, a primitive character re­ tained from Archaeoceti); (2) the narrow, elongate rostrum; (3) the elongate nasal bones; (4) the small, pit-like secondary squa­ mosal fossae; (5) the elongate zygomatic processes of the squa­ mosals diverging anterolaterally from the sagittal plane of the skull (E. carolinensis); (6) the spatulate anterior end of the zy­ gomatic process of the squamosal (E. whitmorei); (7) the blade­ like anterior process of the periotic that is compressed trans­ versely and expanded dorsoventrally (E. whitmorei and proba­ bly E. carolinensis as well); and (8) the length of the humerus, which equals the length of the radius and ulna. Otherwise, Eomysticetus whitmorei has the most primitive cranial morphology of any named primitive baleen-bearing mysticete. The morphologic sequence arranged from E. whit­ morei to Eomysticetus carolinensis to Micromysticetus rothaus­ eni to Cetotheriidae to Balaenopteridae, in general demon­ strates development or enhancement of the following derived character states among Mysticeti: anteroposterior shortening of the frontal and parietal in the intertemporal region, anteroposte­ rior shortening and concomitant transverse widening of the oc­ cipital shield, more horizontal extension of the lambdoidal crests, loss of the secondary squamosal fossa, increase in size of the squamosal prominence, shortening of the zygomatic pro­ cess of the squamosal, increase in lateral displacement of the anterior end of the zygomatic process of the squamosal, widen­ ing of the intercondylar notch, and more ventrolateral flaring of the lateral wall of the braincase. 350 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY FIGURE 30.—Cladogram showing relationships of Eomysticetus whitmorei to other Cetacea. In part from Barnes, 1990; Bames and McLeod, 1984; McLeod et al., 1989, 1992. The nominal families Llanocetidae and Mammal- odontidae are omitted because their characters are poorly known. Characters at dichotomies are as follows: Node l=Order Cetacea: 1, Facial region of skull elongated; 2, Intertemporal region narrow and elon­ gated; 3, Sagittal and lambdoidal crests high; 4, Incisive (anterior palatine) foramina lost; 5, Tym­ panic bulla involuted and inflated; 6, Supraorbital process of frontal enlarged; 7, Basioccipital crest enlarged; 8, Falciform process of squamosal present lateral to anterior process of periotic; 9, Periotic and tympanic bulla separated on their medial sides from the basioccipital; 10, Tympanic bulla sepa­ rated posteriorly from the exoccipital; 11, Hypoglossal foramen in basioccipital located either at the apex of or inside the jugular notch; 12, Peribullary sinus and pterygoid sinus present as diverticula of middle ear sinus; 13, Incisors and canines not transverse, but aligned anteroposteriorly with the cheek-tooth row; 14, Central cusp of both upper and lower cheek teeth transversely compressed; 15, Mandibular foramen enlarged; 16, Scapula with supraspinarus fossa reduced and with acromion and coracoid processes parallel and directed anteriorly; 17, Postzygapophyses on thoracic and lumbar ver­ tebrae reduced. Point 2=Archaeoceti: Apparently paraphyletic; defined primarily by primitive characters. Node 3 = Mysticeti plus Odontoceti: 1, Bones of cranium telescoped; 2, Multiple maxillary foramina derived from infraorbital foramen; 3, Elongate (laterally) supraorbital process of frontal; 4, Vomer exposed on basicranium and extending posteriorly to cover basisphenoid/basioccipital suture; 5, Zygomatic process of the squamosal contacting postorbital process of frontal and/or connected to it by a ligament; 6, Posterior part of dentary thin and dense in area of mandibular foramen (the "pan bone"); 7, Monophylodonty; 8, Elbow joint nonrotational, with anteroposterior position of radius and ulna and formation of discrete flattened articular facets on distal end of humerus (one for radius, two for ulna); 9, Olecranon fossa of humerus lost; 10, Hyperphalangy. Point 4=Odontoceti: 1, Premaxillary foramen present (as an aperture of infraorbital foramen complex) with anteromedial, posteromedial and posterolateral sulci emanating from it; 2, Spiracular plate present on premaxilla anterolateral to nares (for premaxillary sac, a diverticulum of the narial pas­ sage); 3, Ascending process of maxilla expanding posteriorly over supraorbital process of frontal; 4, Antorbital notch present; 5, Pterygoid sinus extending anteriorly and around lateral side of narial pas­ sage; 6, Middle sinus present; 7, Lachrymal and jugal bones fused. Node 5=Mysticeti: 1, Flat (tabular) supraorbital process of frontal with knob-like postorbital process; 2, Antorbital process of maxilla present (antorbital process extending posterolaterally from dorsal surface of rostrum along anterior margin of supraorbital process); 3, Maxilla extending posteriorly ventral to supraorbital process of frontal; 4, Lateral margin of maxilla becoming thin; 5, Temporal crest moved to dorsal surface of supraorbital process of frontal; 6, Basioccipital and basisphenoid becoming thickened dorsoventrally; 7, Mandibular symphysis loose, with elongate ligamental groove at symphysis; 8, Atlas with single transverse process. NUMBER 93 351 Point 6=Family Aetiocetidae: 1, Intertemporal region shortened anteroposteriorly and widened trans­ versely; 2, Protuberance present on premaxilla at anterolateral corner of nasal bone; 3, Squamosal fossa shortened anteroposteriorly; 4, Elongate notch present in posterior border of palatine at poste­ rior end of palate. Node 7 = Family Eomysticetidae plus all other Mysticeti: 1, Functional teeth lost in adult; 2, Baleen present in postfetal stage; 3, Gingival foramina present and enlarged along dorsal margin of horizon­ tal ramus of dentary; 4, Sulci present on palatal surface of maxillae marking paths of nutrient vessels that nourish baleen plates; 5, Vomer forming a prominent ventral keel along midline of palate. Point %=Eomysticetus (=Family Eomysticetidae): 1, Intertemporal region extremely narrowed; 2, Nasal bones elongated; 3, Zygomatic process of squamosal elongated; 4, Zygomatic process of squa­ mosal diverging anterolaterally from sagittal plane of skull; 5, Anterior process of periotic com­ pressed transversely and expanded dorsoventrally ("blade-like"). Node 9=All Mysticeti beyond Family Eomysticetidae: 1, Nasal bones shortened anteroposteriorly; 2, Supraorbital process of frontal widened anteroposteriorly; 3, Intertemporal region shortened antero­ posteriorly so that zygomatic process of squamosal contacts postorbital process of frontal; 4, Anterior process of periotic thickened transversely; 5, Posterior process of periotic fused to posterior (mastoid) process of tympanic bulla; 6, Humerus shorter than radius or ulna. Node 10=Family Cetotheriidae plus Family Balaenopteridae (=Superfamily Balaenopteroidea): 1, Transversely aligned gap present between posterior margin of ascending process of maxilla and ante­ rior margin of supraorbital process of frontal; 2, Ascending process of premaxilla tapered between posterior ends of nasal and maxilla; 3, Zygomatic process of squamosal relatively shortened and blunt; 4, Postglenoid process globose. Point ll=Cetotheriidae: 1. Ascending process of maxilla tapered posteriorly between frontal and ascending process of premaxilla. Point 12=Family Balaenopteridae: 1, Cleft present along alisphenoid/squamosal suture in lateral wall of braincase; 2, Dorsal surface of supraorbital process of frontal depressed ventrally relative to inter­ temporal region; 3, Rostral bones (maxillae, premaxillae, nasals) extending posteriorly toward occip­ ital shield anteroposteriorly compressing frontals and parietals; 4, Mandibular canal and mandibular foramen reduced in diameter; 5, Bone of mandible porous and inflated; 6, Numerous throat grooves present; 7. Four digits in manus. Node 13=Family Eschrichtiidae plus Superfamily Balaenoidea: 1, Rostrum arched dorsally (at least 10°) (from Barnes and McLeod, 1984); 2, Nasal bones wide and "blocky"; 3, Bones around narial region of skull elevated; 4, Horizontal ramus of dentary torsioned in anterior part; 5, Mandibular condyle enlarged and nearly spherical; 6, Horizontal ramus of dentary expanded and arched dors­ oventrally; 7, Coronoid process of dentary reduced. Point 14=Family Eschrichtiidae: 1, Premaxillary foramen present on either or both premaxillae lateral to narial opening (convergence with Odontoceti); 2, Premaxillae wide posteriorly and excluding max­ illae from exposure on cranial vertex; 3, Nasal bones large and forming highest part of skull; 4, Tubercles for muscle attachment present on both sides of occipital shield; 5, Baleen plates thick anteroposteriorly; 6, Baleen plates cream in color; 7, Dorsal fin lost. Node 15=Superfamily Balaenoidea: 1, Zygomatic process of squamosal and glenoid fossa positioned far ventrally on the skull; 2, Pterygoids extending far posteriorly beneath basicranium; 3, Baleen plates numerous and thin; 4, Cervical vertebrae fused. Point 16=Family Neobalaenidae: 1, Body size reduced; 2, Rostrum arched about 17 degrees; 3, Occip­ ital shield extended anteriorly; 4, Zygomatic process of squamosal short; 5, Nasal bones small; 6, Horizontal ramus of dentary deep dorsoventrally; 7, Humerus shortened. Point 17=Family Balaenidae: 1, Head greatly enlarged, approximately one-third of body length; 2, Rostrum transversely compressed; 3, Rostrum greatly arched dorsally, premaxillae forming highest part of skull; 4, Baleen plates greatly elongated; 5, Supraorbital process of frontal sloping and greatly elongate (extended laterally); 6, Zygomatic process of squamosal flaring anterolaterally; 7, Involu­ crum of tympanic bulla flattened dorsoventrally; 8, Dorsal fin lost. OTHER PRIMITIVE BALEEN-BEARJNG MYSTICETES A primitive baleen-bearing mysticete, represented by a re­ markably complete skull and skeleton, was preliminarily re­ ported by Okazaki (1995) from a late Oligocene deposit on the island of Kyushu in Japan. This unnamed whale is quite similar to Eomysticetus whitmorei but is more highly derived. The two species share similar overall size, a narrow and elongate inter­ temporal region capped by a narrow sagittal crest, elongate and narrow nasal bones, a secondary fossa within the squamosal fossa, a relatively small but transversely expanded postglenoid process, a relatively small occipital shield, elongate and antero­ laterally flaring zygomatic processes of the squamosals, a tu­ bercle-like posterior end of the basioccipital crest, an antero­ posteriorly thick exoccipital, elongate and relatively flat rostrum, relatively straight horizontal ramus of the dentary, and a large and dorsally lobate coronoid process of the dentary. The 352 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY unnamed Oligocene mysticete from Kyushu differs from Eomysticetus whitmorei by having the following derived char­ acter states (derived polarity determined in relation to the con­ dition in Archaeoceti): a wider rostrum, an anteroposteriorly wider supraorbital process of the frontal, an anteroposteriorly shorter intertemporal region, larger and wider occipital shield that flares laterally over the temporal fossa rather than flaring dorsally, transversely narrower but more anteroposteriorly ex­ tensive squamosal fossa, and a relatively deeper horizontal ra­ mus of the dentary. The unnamed mysticete from Kyushu also differs from E. whitmorei by having more convex occipital condyles. Although this might be a primitive character state, because it is shared with the Archaeoceti, we interpret it as a derived character state in comparison with E. whitmorei and, therefore, as a reversal. The elongate intertemporal region of E. whitmorei is a primitive character state that is shared with ar­ chaeocetes, and for this reason (as with the archaeocetes) the zygomatic process of the squamosal makes no contact with the postorbital process of the frontal. In the Kyushu specimen, with its anteroposteriorly shorter intertemporal region, the zygo­ matic process of the squamosal contacts the postorbital process of the frontal (see Okazaki, 1995, fig. lb), and this is the de­ rived character state. That derived condition exists as well in all other (more derived) mysticetes, except in "Mauicetus''' lopho­ cephalus, as illustrated by Marples (1956, fig. lc; see also Barnes and McLeod, 1984, fig. 3c). The gap between the ante­ rior tip of the zygomatic process and the squamosal and the postorbital process of the frontal is smaller in " M " lophoceph­ alus than it is in E. whitmorei. In that regard, among the Mys­ ticeti E. whitmorei is the form most like the Archaeoceti, and " M " lophocephalus approaches the condition in various de­ rived groups of mysticetes, such as the Cetotheriidae, Bal­ aenopteridae, Eschrichtiidae, Neobalanidae, and Balaenidae. Late Oligocene Mauicetus is a problematic taxon. It is usu­ ally classified in the family Cetotheriidae, although Barnes and McLeod (1984:16) noted that it does not have autapomorphic characters that distinguish the type genus of the family Ce­ totheriidae, the late Miocene genus Cetotherium. Of the nomi­ nal species in the genus Mauicetus, only " M " lophocephalus is known by a relatively complete skull, and the illustrations of this skull (e.g., Marples, 1956, fig. lc; Barnes and McLeod, 1984, fig. 3c) are based upon interpretations of its morphology prior to its being completely removed from sediment. If we re­ strict our interpretation of the skull morphology to the speci­ men illustrated by Marples (1956), it is clear that " M " lopho­ cephalus is a relatively primitive baleen-bearing mysticete, although it is not nearly so primitive as Eomysticetus whit­ morei. In some characters (e.g., zygomatic process of squamo­ sal not contacting postorbital process of the frontal, smaller su­ praoccipital shield) " M " lophocephalus is more primitive than the unnamed baleen-bearing mysticete reported by Okazaki (1995) from Kyushu. It does not have the elongate and flaring zygomatic process of the squamosal that the unnamed mys­ ticete from Kyushu shares with E. whitmorei, and it is not so closely related to either of these taxa as they are to each other. Whether " M " lophocephalus is ultimately determined to be­ long to the Cetotheriidae or to some other clade, it is not inclu­ sive within the clade containing Eomysticetus whitmorei and the unnamed mysticete from Kyushu. Crowley and Barnes (1996) reported yet another type of un­ named primitive baleen-bearing mysticete of late Oligocene age from the Olympic Peninsula of Washington. This whale is not closely related to " M " lophocephalus, nor does it belong to the family Eomysticetidae. Although that specimen, Eomys­ ticetus whitmorei, and the unnamed mysticete from Kyushu all have a sagittal crest and an extensive exposure of the parietals in the intertemporal region, as well as a large and dorsally lo- bate coronoid process of the dentary, these are shared primitive characters and do not necessarily indicate close relationship. The late Oligocene mysticete from Washington has autapomor­ phic characters—such as a short and medially curved zygo­ matic process of the squamosal, a relatively anteroposteriorly shorter and wider intertemporal region, and a shorter rostrum— that serve to distinguish it from Eomysticetus whitmorei. From late Oligocene deposits in southern Baja California Sur, Mexico, have been reported other primitive baleen-bearing mysticetes (Barnes, 1998) that appear to share characters with Eomysticetus whitmorei: elongate nasal bones, elongate and di­ vergent zygomatic process of the squamosal, and a small su­ praoccipital shield whose margins flare dorsally rather than lat­ erally. These specimens have not yet been completely cleaned from their enclosing matrix, and only preliminary studies of them have been made. The specimens from Baja California Sur, along with the un­ named mysticete reported from Kyushu by Okazaki (1995), demonstrate that primitive baleen-bearing mysticetes of the Eomysticetus grade, and possibly belonging to the family Eomysticetidae, were widespread in late Oligocene time. With apparent records of this group in the western North Atlantic, the eastern North Pacific, and the western North Pacific, the group might have been essentially cosmopolitan in late Oli­ gocene time. CLASSIFICATION Within the classification of mysticetes we recognize the fam­ ilies Aetiocetidae, Eomysticetidae, Cetotheriidae, Balaenidae, Neobalaenidae, Balaenopteridae, and Eschrichtiidae. Barnes (1984) and Barnes et al. (1985) proposed classifications of Ce­ tacea that embraced both fossil and extant taxa, and Mitchell (1989) offered a new arrangement of Archaeoceti and Mys­ ticeti accompanying his description of Llanocetus denticrena- tus, an enigmatic, primitive late Eocene toothed mysticete from Antarctica. Our establishment of the new taxon Eomysticetus whitmorei necessitates further reappraisal of mysticete system­ atics. In the revised classification of Mysticeti presented below, subfamilies are used where they have become commonly rec­ ognized. In instances where the family group name is not used NUMBER 93 353 at the same rank as originally proposed, the original author is listed in parentheses followed by the author(s) who used the emended rank. Our classification differs from Mitchell's (1989) arrangement principally in our avoidance of the use of infraorders to group primitive, toothed mysticetes (i.e., Llano- cetus, Aetiocetus, Mammalodon) about which too little is known to warrant such definitive hierarchical assignments, es­ pecially in the case of Llanocetus denticrenatus, which Mitch­ ell (1989) described from only a fragment of the right dentary and a cranial endocast. The two archaeocete-like teeth preserved in the mandibular fragment of Llanocetus denticrenatus and the wide diastema between them differ considerably from dental morphology and tooth spacing in Aetiocetus cotylalveus Emlong, 1966, and Mammalodon colliveri Pritchard, 1939, but these differences do not necessarily indicate the absence of shared similarities in cranial morphology. We therefore propose the new superfamily Aetiocetoidea as a group to accommodate all of the few known Paleogene toothed mysticetes, and we accordingly refer the Llanocetidae of Mitchell (1989) to that superfamily pending knowledge of the cranial anatomy of llanocetids. In the following classification, a | in front of a taxonomic group indicates that it is extinct. Order Cetacea Brisson, 1762 Suborder Mysticeti Flower, 1864 fSuperfamily Aetiocetoidea, new superfamily tFamily Aetiocetidae Emlong, 1966 tSubfamily Chonecetinae Barnes, Kimura, Furusawa, and Sawa- mura, 1995 tFamily Llanocetidae Mitchell, 1989 tFamily Mammalodontidae Mitchell, 1989 tSuperfamily Eomysticetoidea, new superfamily tFamily Eomysticetidae, new family Superfamily Balaenopteroidea (Gray, 1868) Mitchell, 1989 tFamily Cetotheriidae (Brandt, 1872) Miller, 1923 tSubfamily Cetotheriopsinae Brandt, 1872 tSubfamily Cetotheriinae Brandt, 1872 Family Balaenopteridae Gray, 1864 Subfamily Megapterinae (Gray, 1866) Gray, 1868 Subfamily Balaenopterinae (Gray, 1864) Brandt, 1872 Superfamily Eschrichtioidea (Ellerman and Morrison-Scott, 1951) Mitchell, 1989 Family Eschrichtiidae Ellerman and Morrison-Scott, 1951 Superfamily Balaenoidea (Brandt, 1873) Mitchell, 1989 Family Neobalaenidae Gray, 1874 Family Balaenidae Gray, 1825 Absent from the foregoing arrangement and from most of the recent classifications of the Mysticeti are several taxa once thought to be mysticetes or to be mysticete ancestors. Com­ menting on the primitive late Oligocene cetacean Squalodon pygmaeus (Muller, 1849) from South Carolina, Gervais (1871:138) proposed that S. pygmaeus actually belonged among the rorquals. In erecting the new genus Agorophius for S. pygmaeus, Cope (1895:139) noted that "the form of the skull in this genus approaches distinctly that of Cetotherium ... and the permanent loss of the teeth would probably render it neces­ sary to refer it to a Mystacocete." Subsequent authors (e.g., True, 1907, 1908; Abel, 1913; Miller, 1923; Kellogg, 1928) correctly placed Agorophius pygmaeus in the Odontoceti. Pat- riocetus ehrlichi (Van Beneden, 1865), from the late Oligocene sands at Linz, Austria, was figured incorrectly by Abel (1913; see also Kellogg, 1928), who suggested that it was a primitive toothed mysticete. Miller (1923:44) recognized its true affini­ ties with the Odontoceti, and Rothausen (1968) illustrated the correct form of the rostrum and referred Patriocetus to the Squalodontidae. Archaeodelphis patrius Allen, 1921, known by a partial cra­ nium without locality data but thought to be of late Eocene age (Allen, 1921), was assigned to the odontocete family Ago­ rophiidae by Miller (1923:40). Kellogg (1928) placed Archaeo­ delphis in Odontoceti incertae sedis, but suggested it as a possi­ ble mysticete ancestor, as did Dechaseaux (1961:881). Rothausen (1968:97-98) included Archaeodelphis in the Ago­ rophiidae, but Whitmore and Sanders (1976) returned it to in­ certae sedis. We have considered the morphology of the holo­ type (MCZ 15749) of Archaeodelphis patrius in some detail and agree with Fordyce (1981:1042) that it is clearly an odon­ tocete and had nothing whatsoever to do with the ancestry of the mysticetes. Following Emlong (1966) in regarding Aetiocetus cotyla­ lveus as a member of the Archaeoceti, Mchedlidze (1976) sug­ gested that Ferecetotherium kelloggi Mchedlidze, 1970, and Mirocetus rjabinini Mchedlidze, 1970, belonged with Aetioce­ tus in the family Aetiocetidae, which he concluded was the ar­ chaeocete family that gave rise to the Mysticeti. Many authors (e.g., Barnes and Mitchell, 1978; Fordyce, 1980; Barnes, 1985, 1987, 1989; Barnes et al., 1995; Evans, 1987; Sanders and Barnes, 1989), however, agree with Van Valen's (1968) inter­ pretation of Aetiocetus cotylalveus as a primitive toothed mys­ ticete, thus leaving in question the other two taxa assigned to the Aetiocetidae by Mchedlidze. Barnes (1985) proposed that Ferecetotherium kelloggi is actually a primitive sperm whale, its slender, conical teeth constituting a homodont dentition quite unlike the dental characteristics of Archaeoceti. As for Mirocetus rjabinini, independent examinations of its holotype by Sanders and by Rothausen produced concurrent opinions: that the maxillae ascend well onto the frontals and probably overspread most of the supraorbital processes, and that Miroce­ tus rjabinini therefore belongs in the Odontoceti (Rothausen to Sanders, pers. comm., 1976, 1990). In summary, our opinions regarding the subordinal positions of the genera discussed above are as follows: Suborder Mysticeti Aetiocetus Llanocetus Mammalodon Suborder Odontoceti Agorophius Archaeodelphis Ferecetotherium Mirocetus Patriocetus 354 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY Conclusions Eomysticetus whitmorei is an archaic, baleen-bearing mys­ ticete of late Oligocene age. The length of the skull and the size of the vertebrae indicate that it was of moderate size (~8 m long) in life, and although its rostrum and mandibles are eden­ tulous as in other mysticetes, it has a braincase very much like that of an archaeocete. The stage of its cranial telescoping is unlike that of any other mysticete yet described. The palatal surface of its maxilla has sulci that indicate the presence in life of blood vessels that would have nourished baleen, but the pal­ atal surface was not highly vascularized, as is typical of highly evolved mysticetes, so the baleen probably was not very long. The coronoid process of the dentary is very large, and the tem­ poral musculature was extensive. In many characters, E. whitmorei is morphologically transi­ tional between archaeocetes and mysticetes, which demon­ strates clearly the direct ancestral-descendant relationship be­ tween the two groups and helps to refute the arguments for cetacean polyphyly. Eomysticetus whitmorei is a relict taxon, however, because baleen-bearing mysticetes with a higher degree of cranial tele­ scoping are known earlier in the Oligocene and also were con­ temporaneous with A. whitmorei in Chandler Bridge (lower Chattian) time. Its morphology demonstrates the stage of cra­ nial telescoping that was probably achieved in the early Oli­ gocene by other lineages of mysticetes. The species was con­ temporaneous with, and perhaps sympatric with, at least two other species of mysticetes (cetotheriopsine cetotheriids; Sand­ ers and Barnes, 2002), a diversity of primitive odontocetes, and some new and as-yet-undescribed toothed mysticetes from South Carolina (Barnes and Sanders, 1996). It also is broadly contemporaneous with relict tooth-bearing aetiocetid mystice­ tes and with several other lineages of primitive mysticetes that existed elsewhere in the world in late Oligocene time. Some of these other mysticetes had teeth, some had baleen, some had highly evolved crania, and some had primitive crania. The skull of Eomysticetus whitmorei exhibits the most primi­ tive cranial features known in a fossil baleen-bearing mysticete whale. Its prominent archaeocete-like characters—such as the location of the nasal opening far anterior to the vertex of the skull—provide new insights into the origin of baleen whales and suggest that a presently unknown line of Archaeoceti branched directly toward Mysticeti and that Odontoceti evolved from Archaeoceti along an entirely separate evolution­ ary pathway. Literature Cited Abel, O. 1913. Die Vorfahren der Bartenwale. Denkschriften der Akademie der Wissenschaften, Wien, Mathematisch-Naturwissenchaflliche Klasse, 90:155-224, 12 plates. Allen, G.M. 1921. A New Fossil Cetacean. Bulletin of the Museum of Comparative Zo­ ology at Harvard College, 65(1): 1—14, 3 figures, 1 plate. Bames, L.G. 1984. Search for the First Whale: Retracing the Ancestry of Cetaceans. Oceans, 17(2):20-23. 1985. Review: G.A. Mchedlidze, General Features of the Paleobiological Evolution of Cetacea, 1984. [English translation.] Marine Mammal Science, l(l):90-93. 1987. Aetiocetus and Chonecetus. Primitive Oligocene Toothed Mystice­ tes and the Origin of Baleen Whales. Journal of Vertebrate Paleon­ tology, supplement, 7(3): 10A. 1989. Aetiocetus and Chonecetus (Mammalia: Cetacea); Primitive Oli­ gocene Toothed Mysticetes and the Origin of Baleen Whales. In Fifth International Theriological Congress, Rome, 22-29 August 1989, Abstracts of Papers and Posters, 1:1:479. 1990. The Fossil Record and Evolutionary Relationships of the Genus Tursiops. In S. Leatherwood and R.R. Reeves, editors, The Bottle­ nose Dolphin, pages 3-26. New York: Academic Press. 1998. The Sequence of Fossil Marine Mammal Assemblages in Mexico. In O. Carranza-Castaneda and D.A. Cordoba-Mendez, editors, Avances en Investigacion: Paleontologia de Vertebrados. Instituto de Investigaciones en Ciencias de la Tier, Universidad Autonoma del Estado de Hidalgo, Publicacion Especial, 1:26-79 Barnes, L.G., D.P. Domning, and C E . Ray 1985. Status of Studies on Fossil Marine Mammals. Marine Mammal Sci­ ence, l(l):90-93. Bames, 1995 Bames, 1984 Bames, 1978 Bames, 1990 1996. Brandt, 1872 1873 Brisson 1762 L.G., M. Kimura, H. Furusawa, and H. Sawamura Classification and Distribution of Oligocene Aetiocetidae (Mamma­ lia; Cetacea; Mysticeti) from Western North America and Japan. The Island Arc, 3(4):392-431. L.G., and S.A. McLeod The Fossil Record and Phyletic Relationships of Gray Whales. In M.L. Jones, S. Swartz, and S. Leatherwood, editors, The Gray Whale: Eschrichtius robustus, pages 3-32. New York: Academic Press. L.G., and E.D. Mitchell Cetacea. In V.J. Maglio and H.B.S. Cooke, editors, Evolution of Af­ rican Mammals, pages 582-602. Cambridge, Massachusetts: Har­ vard University Press. L.G., and A.E. Sanders An Archaic Oligocene Mysticete from South Carolina. Journal of Vertebrate Paleontology, supplement, 10(3): 14A. The Transition from Archaeocetes to Mysticetes: Late Oligocene Toothed Mysticetes from near Charleston, South Carolina. In J.E. Repetski, editor, Abstracts of Papers, Sixth North American Paleon­ tological Convention, Smithsonian Institution, Washington, D.C, June 9-June 12, 1996. Paleontological Society, Special Publication, 8:24. J.F. von Uber eine neue Classification der Bartenwale (Balaenoidea) mit beriicksichtigung der untergegangenen Gattungen derselben. Bulle­ tin de I'Academie Imperiale des Sciences de St. Petersbourg, series 3, 17:113-124. Untersuchungen uber die fossilen und subfossilen Cetaceen Euro- pas. Memoires de I'Academie Imperiale des Sciences de St. Peters­ bourg, series 7, 20(1): viii + 372 pages, 34 plates. , M.J. Regnum animate in classes IX distributum sive synopsis methodica. NUMBER 93 355 Editio actior, 296 pages. Leiden: Theodorus Haak. Cope, E.D. 1895. Fourth Contribution to the Marine Fauna of the Miocene Period of the United States. Proceedings of the American Philosophical Soci­ ety, 34:135-155, 1 plate. Crowley, B.E., and L.G. Barnes 1996. A New Late Oligocene Mysticete from Washington State. In J.E. Repetski, editor. Abstracts of Papers, Sixth North American Paleon­ tological Convention, Smithsonian Institution, Washington, D.C, June 9-June 12, 1996. Paleontological Society, Special Publication, 8:90. Curry, D., C.G. Adams, M.C Boulter, F.C. Dilley, F.E. Eames, B.M. Funnell, and M.K. Wells 1978. A Correlation of Tertiary Rocks in the British Isles. Geological So­ ciety of London, Special Report, 12:1-72, 5 figures. Dechaseaux, C. 1961. Encephales de Cetaces fossiles. In J. Piveteau, editor, Trade de Paleontologie, 6(1):881-886. Paris: Masson. Ellerman, J.R., and J.C.S. Morrison-Scott 1951. Checklist of Palaearctic and Indian Mammals, 1938-1946. 810 pages. London: British Museum (Natural History). Emlong, D.R. 1966. A New Archaic Cetacean from the Oligocene of Northwest Oregon. Bulletin of the Museum of Natural History, University of Oregon, 3:1-51, 15 figures. Erickson, B.R., and G.S. Sawyer 1996. The Estuarine Crocodile Gavialosuchus carolinansis n. sp. (Cro- codylia: Eusuchia) from the Late Oligocene of South Carolina, North America. Monograph of the Science Museum of Minnesota, Paleontology, 3: 47 pages. Evans, P.G.H. 1987. The Natural History of Whales and Dolphins, xvi + 343 pages. New York: Facts on File Publications. Flower, W.H. 1864. Notes on the Skeletons of Whales in the Principal Museums of Hol­ land and Belgium, with Description of Two Species Apparently New to Science. Proceedings of the Zoological Society of London, 1864:384-420. Fordyce, R.E. 1980. Systematics of the Odontocete Whale Agorophius pygmaeus and the Family Agorophiidae (Mammalia: Cetacea). Journal of Paleontol­ ogy, 55(5): 1028-1045. 1981. A Review of Australian Fossil Cetacea. Memoirs of the National Museum of Victoria, 43:43-58, 2 plates. Fraser, F.C, and P.E. Purves 1960. Hearing in Cetaceans: Evolution of the Accessory Air Sacs and the Structure and Function of the Outer and Middle Ear in Recent Ceta­ ceans. Bulletin of the British Museum (Natural History), Zoology, 7(1): 1-140, frontispiece, plates 1-53. Gaskin, D.E. 1982. The Ecology of Whales and Dolphins, xii + 459 pages, 99 figures. London: Heinemann Educational Books. Geisler, J., A.E. Sanders, and Z. Luo 1996. A New Protocetid Cetacean from the Eocene of South Carolina, U.S.A.; Phylogenetic, and Biogeographic Implications. In J.E. Re­ petski, editor. Abstracts of Papers, Sixth North American Paleonto­ logical Convention, Smithsonian Institution, Washington, D C , 9-12 June 1996. Paleontological Society, Special Publication, 8:139. Gervais, P. 1871. Remarques sur F anatomie des cetaces de la division des balenides tirees de l'examen des pieces relatives a ces animaux qui sont con- servees au museum. Nouvelles Archives du Museum d'Histoire Na­ turelle, Paris, 7:65-146, 7 plates. Gingerich, P.D., M. Arif, and W.D. Clyde 1995. New Archaeocetes (Mammalia, Cetacea) from the Middle Eocene Domanda Formation of the Sulaiman Range, Punjab (Pakistan). Contributions from the Museum of Paleontology, University oj Michigan, 29(11 ):291-330. Gingerich, P.D., and D.E. Russell 1990. Dentition of Early Eocene Pakicetus (Mammalia, Cetacea). Contri­ butions from the Museum of Paleontology, University of Michigan, 28(1): 1-20. Gingerich, P.D., N.A. Wells, D.E. Russell, and S.M.I. Shah 1983. Origin of Whales in Epicontinental Remnant Seas: New Evidence from the Early Eocene of Pakistan. Science, 220:403^406, 2 figures. Gray, J.E. 1825. Outline of an Attempt at the Disposition of the Mammalia into Tribes and Families with a List of the Genera Apparently Appertain­ ing to Each Tribe. Philosophical Annals, 26:337-344. 1864. Notes on the Whalebone-Whales, with a Synopsis of the Species. Annals and Magazine of Natural History, series 3, 14:345-353. 1866. Catalogue of the Seals and Whales in the British Museum. Second edition, 402 pages, 102 figures. London: British Museum. 1868. Synopsis of the Species of Whales and Dolphins in the Collection oj the British Museum. 10 pages, 37 plates. London: Bernard Quaritch. 1874. List of Seals, Whales, and Dolphins of New Zealand. Transactions and Proceedings of the New Zealand Institute, 1873, 6(Article 16):87-89 [Transactions for 1873, published in 1874.] Hershkovitz, P. 1966. Catalog of Living Whales. Bulletin of the United States National Museum, 246: viii + 259 pages. Hulbert, R.C, R.M. Petkewich, G.A. Bishop, D. Bukry, and D.P. Aleshire 1998. A New Protocetid (Mammalia: Cetacea: Archaeoceti) and Associ­ ated Biota from the Middle Eocene of Georgia. Journal of Paleon­ tology. 72:907-927. Kellogg, R. 1923. Description of an Apparently New Toothed Cetacean from South Carolina. Smithsonian Miscellaneous Collections, 76(7): 7 pages, 2 plates. 1927. Study of the Skull of a Fossil Sperm-Whale from the Temblor Mi­ ocene of Southern California. Carnegie Institution of Washington Publication, 346:1-23. 1928. The History of Whales—Their Adaptation to Life in the Water. Quarterly Review of Biology, 3:29-76, 174-208. 1936. A Review of the Archaeoceti. Carnegie Institution of Washington Publication, 482: xv + 366 pages, 88 figures, 37 plates. 1965. Fossil Marine Mammals from the Miocene Calvert Formation of Maryland and Virginia, Part 1: A New Whalebone Whale from the Miocene Calvert Formation. Bulletin of the United States National Museum. 241A-45, plates 1-21. 1968. Fossil Marine Mammals from the Miocene Calvert Formation of Maryland and Virginia, Part 6: A Hitherto Unrecognized Calvert Cetothere. Bulletin of the United States National Museum, 247: 133-161, plates 49-57. 1969. Cetothere Skeletons from the Miocene Choptank Formation of Maryland and Virginia, Part 1: The Skeleton of a Miocene Choptank Cetothere. Bulletin of the United States National Museum, 294:1 — 24, plates 1-15. Kleinenberg, S.E. 1958. K voprosu o proiskhozhdenii kitoobraznykh. [On the Problem of Origin of Cetaceans.] Doklady Akademia Nauk SSSR, 122(5):950- 952. [In Russian.] 1959. On the Origin of Cetacea. In H.R. Hewer and N.D. Riley, editors, Proceedings of the Fifteenth International Congress of Zoology, 1958, London, pages AA5-^A1. Kleinenberg, S.E., and A.V. Yablokov 1958. O morfologii verkhnikh dykhatel'nykn putei kitoobrazhynykh. [The Morphology of the Upper Respiratory Passages in the Cetaceans.] Zoologischeskii Zhurnal, 37:1091-1099. Koretsky, I.A., and A.E. Sanders 2002. Paleontology of the Late Oligocene Ashley and Chandler Bridge Formations of South Carolina, 1: Paleogene Pinniped Remains; the 356 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY Oldest Known Seal (Carnivora: Phocidae). In R. Emry, editor, Cen­ ozoic Mammals of Land and Sea: Tributes to the Career of Clayton E. Ray. Smithsonian Contributions to Paleobiology, 93:179-183. Kumar, K., and A. Sahni 1986. Remingtonocetus harudiensis. New Combination, a Middle Eocene Archaeocete (Mammalia, Cetacea) from Western Kutch, India. Journal of Vertebrate Paleontology, 6(4):326-349, 10 figures. Marples, B.J. 1956. Cetotheres from the Oligocene of New Zealand. Proceedings of the Zoological Society cf London, 126(4):565—580. Martini, E., and C Muller 1975. Calcareous Nannoplankton from the Type Chattian. Proceedings of the Sixth Congress on Neogene Mediterranean Stratigraphy, Brat­ islava, Yugoslavia, 1975, 1:37-41. Mchedlidze, G.A. 1970. Nekotorye obshchie cherty istorii kitoobraznykh, Chast' I. [Some Features of the Historical Development of Cetacea, Part L] 112 pages. Tbilisi, Georgia: Akademia Nauk Gruzinskoi S.S.R., Institut Paleobiology, "Metsniereba" Publishers. [In Russian.] 1976. Osnovnyye cherty paleobiologicheskoy istorii kitoobraznykh. [Basic Features of the Paleobiological History of Cetaceans.] 136 pages, 32 plates. Tbilisi, Georgia: Akademia Nauk Gruzinskoi S.S.R., Institut Paleobiology, "Metsniereba" Publishers. [In Russian, English sum­ mary.] McLeod, S.A., R.L. Brownell, and L.G. Barnes 1989. The Classification of Cetaceans: Are All Whales Related? In Fifth International Theriological Congress, Rome, 22-29 August 1989, Abstracts of Papers and Posters, 1:490. McLeod, S.A., F.C. Whitmore, Jr., and L.G. Bames 1992. Evolutionary Relationships and Classification. In J.J. Bums, J.J. Montague, and C.J. Cowles, editors, The Bowhead Whale. Society for Marine Mammalogy, Special Publication, 2:45-70. Meyer, H. von 1849. [Untitled.] Neues Jahrbuch fiir Mineralogie, Geologie, und Paldon­ tologie, 1849:547-550. Stuttgart Miller, G.S., Jr. 1923. The Telescoping of the Cetacean Skull. Smithsonian Miscellaneous Collections, 76(5): 1-71, 8 plates. Mitchell, E.D. 1989. A New Cetacean from the Late Eocene La Meseta Formation, Sey­ mour Island, Antarctic Peninsula. Canadian Journal of Fisheries and Aquatic Sciences, 46(12):2219-2235, 4 figures. Muller, J. 1849. Uber die fossilen Reste der Zeuglodonten von Nordamerica, mit Rucksicht auf die europdischen Reste aus dieser Familie. iv + 38 pages. Berlin: Verlag von G. Reimer. Okazaki, Y. 1995. A New Type of Primitive Baleen Whale (Cetacea; Mysticeti) from Kyushu, Japan. The Island Arc, 3(4):432-435. Owen, R. 1839. Observations on the Basilosaurus of Dr. Harlan (Zeuglodon ce­ toides, Owen). Transactions of the Geological Society of London, series 2, 6:69-79. Perrin, W.F. 1975. Variation of Spotted and Spinner Porpoise (Genus Stenella) in the Eastern Pacific and Hawaii. Bulletin of the Scripps Institute of Oceanography, University of California, 21:1-206. Pritchard, B.G. 1939. On the Discovery of a Fossil Whale in the Older Tertiaries of Torquay, Victoria. Victorian Naturalist, 55(9): 151—159. Reichenbach, H.G.L. 1847. In C.G. Carus, Resultate geologischger. anatomischer und zoologi- scher Untersuchungen iiber das unter dem Namen Hydrarchos von der A.C. Koch zuerst nach Europa gebrachle und in Dresden aus- gestellte grosse fossile Skellet, pages 1-15, plates 1-7. Dresden and Leipzig. Rice, D.W. 1966. Cetaceans. In S. Anderson and J.K. Jones, Jr., editors, Recent Mam­ mals of the World: A Synopsis of Families, pages 291-324. New York: The Ronald Press Co. 1984. Cetaceans. In S. Anderson and J.K. Jones, Jr., editors, Orders and Families of Recent Mammals of the World, pages 447—490. New York: John Wiley and Sons. Rothausen, K. 1968. Die systematische Stellung der europaischen Squalodontidae (Od­ ontoceti, Mamm.). Paldontologische Zeitschrift, 42:83-104, 3 fig­ ures, 2 plates. 1971. Cetotheriopsis tobieni n. sp., der erste palaogene Bartenwal (Ce­ totheriidae, Mysticeti, Mamm.) nordlich des Tethysraumes. Abhand­ lungen des Hessischen Labdesamtes fur Bodenforschung, 6 0 : D I ­ MS, 3 figures, 3 plates. 1985. The Early Evolution of Cetacea. Fortschritte der Zoologie, 30: 143-147. Sanders, A.E. 1980. Excavation of Oligocene Marine Fossil Beds near Charleston, South Carolina. National Geographic Society Research Reports, 12:601- 621, 8 figures. Sanders, A.E., and L.G. Bames 1989. An Archaic Oligocene Mysticete from South Carolina, U.S.A. Ab­ stracts, Eighth Biennial Conference on the Biology of Marine Mam­ mals, Pacific Grove, California, 7-11 December 1989, page 58. 1991. Late Oligocene Cetothriopsis-hke Mysticetes (Mammalia, Cetacea), from near Charleston, South Carolina. Journal of Vertebrate Pale­ ontology, supplement, 11(3):54A. 2002. Paleontology of the Late Oligocene Ashley and Chandler Bridge Formations of South Carolina, 2: Micromysticetus rothauseni, a Primitive Cetotheriid Mysticete (Mammalia: Cetacea). In R. Emry, editor, Cenozoic Mammals of Land and Sea: Tributes to the Career of Clayton E. Ray. Smithsonian Contributions to Paleobiology, 93:271-292. Sanders, A.E., R.E. Weems, and EM. Lemon, Jr. 1982. Chandler Bridge Formation—A New Oligocene Stratigraphic Unit in the Lower Coastal Plain of South Carolina. U.S. Geological Sur­ vey Bulletin, 1529-H: 105-124, 4 figures. True, F.W. 1907. Remarks on the Type of the Fossil Cetacean Agorophius pygmaeus (Muller). Smithsonian Institution Publication, 1694: 8 pages, 1 plate. 1908. On the Classification of the Cetacea. Proceedings of the American Philosophical Society, 47( 189):385-391. Van Valen, L. 1968. Monophyly or Diphyly in the Origin of Whales. Evolution, 22(1): 37-41. Weems, R.E., and A.E. Sanders 1986. The Chandler Bridge Formation (Upper Oligocene) of the Charles­ ton, South Carolina, Region. In T L . Neathery, editor, Geological Society of America Centennial Field Guide. Southeastern Section, pages 323-326, 1 figure. Tulsa, Oklahoma: Society of Economic Paleontologists and Minerologists. Whitmore, F.C, Jr., and A.E. Sanders 1977 ("1976"). Review of the Oligocene Cetacea. Systematic Zoology, 25(4):304-320. [Date on title page is 1976; actually published in 1977.] Yablokov, A.V. 1964. Konvergentsiya iii parallelizm v razvitii kitoobraznykh. [Conver­ gence or Parallelism in the Evolution of Cetaceans.] Paleontolo- gicheskii Zhurnal, 1964, (1):97-106. [In Russian. Translated in In­ ternational Geology Review, 17(8): 1461-1468.] Plates for Joel Asaph Allen's Unpublished Monograph on the Mammalian Orders Cete and Sirenia and a Record of the Search for the Manuscript James G. Mead and Rosemary G Dagit A B S T R A C T In 1882 Joel Asaph Allen published his "Preliminary List of Works and Papers Relating to the Mammalian Orders Cete and Sirenia" as a Bulletin of the United States Geological and Geo­ graphical Survey of the Territories. That list, consisting of 1013 bibliographic records in chronological order from 1495 to 1840, was said to be one-third of the actual bibliography of a monograph on cetacean and sirenians on which Allen was working. Extensive archival searching has produced only the unpublished plates of the work. We have concluded that the monograph does not reside in any of the logical archival repositories. Introduction Joel Asaph Allen is perhaps best known for his work with the American Ornithological Union while curator at the American Museum of Natural History. Less well known, but by no means less significant, is his work on mammals. In 1882, "The Pre­ liminary List of Works and Papers Relating to the Mammalian Orders of Cete and Sirenia" was published in the Bulletin of the United States Geological and Geographical Survey of the Ter­ ritories, volume 6. This annotated bibliography contains 1013 titles extending from 1495 to 1840. Allen had received a copy of the galley proofs of the remaining titles, covering the years 1840 to approximately 1880. Because of a number of factors to be explained in this paper, that portion of the work was never published. The purpose of this research paper was to try to lo­ cate the missing portion of that work. INSTITUTIONAL ABBREVIATIONS.—The following institu­ tional abbreviations are used: AMNH American Museum of Natural History, New York James G. Mead, Division of Mammals, National Museum of Natural History, Smithsonian Institution, Washington, D.C. 20560-0108. Rose­ mary G. Dagit, 19989 Sischo Drive, Topanga, California 90290. GPO Government Printing Office, Washington, D.C. MCZ Museum of Comparative Zoology, Harvard University, Cam­ bridge, Massachusetts USGS United States Geological Survey USNM Collections of the National Museum of Natural History, in­ cluding those of the former United States National Museum, Washington, D.C. ACKNOWLEDGMENTS.—None of this would have been possi­ ble without the generous help of the archival community. In particular we want to express our thanks to the following: Carol Spawn, Academy of Natural Sciences, Philadelphia; Steve Cartlett, American Philosophical Society; Christine Rag- geri, College of Physicians, Philadelphia; Linda Stanley, His­ torical Society of Philadelphia/The Library Company; Head Librarian, MCZ; William Cox, Smithsonian Institution Ar­ chives; Mary Rabbitt, Historian, George Goodwin, Head Li­ brarian, and Cliff Nelson, Invertebrate Paleontologist, USGS; Michele Pacifico, USGS Records, and Edward Shamel, GPO Records, United States National Archives; James Dallet, Ar­ chivist, University of Pennsylvania Archives, Philadelphia; and all of the following at the AMNH: Pam Haas, Main Archivist; Mary Lecroy, Science Assistant , Orni thology; Marie Lawrence, Mammals Librarian; Charlotte Holton, Osborn Li­ brarian. History of Allen's Work Joel Asaph Allen was born in Springfield, Massachusetts, on 19 July 1838 (Allen, 1916). He entered the Wilbraham Acad­ emy as a student in 1858 and started his academic publishing career with a series of articles on the birds of New England, which appeared in the New England Farmer starting on 11 Au­ gust 1860. He entered the Lawrence Scientific School at Har­ vard in February 1862 under the tutelage of Louis Agassiz at the Museum of Comparative Zoology. Allen distinguished himself as a general naturalist, although his tendencies lay in the studies of birds and mammals, with which he was to be- 357 358 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY come an expert in later years. He went with Agassiz to Brazil in 1865 and, with few interruptions, was employed as an assis­ tant to Agassiz at the MCZ until he resigned that post to take up a curatorship at the American Museum of Natural History in New York in May of 1885. One of these interruptions was his assistantship to Spencer Fullerton Baird, then Assistant Secre­ tary of the Smithsonian Institution, to lead an expedition to the Yellowstone Territory in 1873. INVOLVEMENT WITH UNITED STATES GEOLOGICAL AND GEOGRAPHICAL SURVEY OF THE TERRITORIES Extended research into the Hayden Survey of the Territories and the history of the United States Geological Survey, as well as Allen's life, was necessary. In his autobiography Allen (1916:32) stated that he was a "special collaborator" with the Hayden Survey from 1876 to 1882. He was primarily head­ quartered at the MCZ. He also received payment "for services rendered as naturalist" for four months, February-May 1879. During his association with the Survey, he published a mono­ graph on North American Rodentia with Elliott Coues as coau­ thor in 1877. Allen corresponded frequently with Spencer Baird at the Smithsonian concerning specimen exchange and acquisition. In May 1876, he wrote asking for permission from Baird to give Hayden a chance to publish his work on North American mammals. This correspondence suggests that the Smithsonian had some interest in the work but does not indi­ cate how Allen became associated with the Hayden publica­ tions. January of 1877 found Allen "hard at work on the Sciu- ridae for Hayden" (archival reference H) and asking to use the Smithsonian collection to finish his pinniped work. Allen continued to work on the "History of North American Pinnipeds, a Monograph of the Walruses, Sea-Lions, Sea-Bears and Seals of North America," which finally appeared as num­ ber 12 of the Miscellaneous Publications of the Geological and Geographical Survey in 1880. Records exist indicating pay­ ment and correspondence connected with proofing both copy and plates. In May 1880, Allen wrote to Baird discussing his current work on cetaceans (letter from Allen to Baird dated 11 May 1880; archival reference E). As of September 1881, he had completed the bibliography (some of which was in the hands of the printer) and designed eight original plates to illustrate the monograph text (letter from Allen to Baird dated 26 September 1881; archival reference F). T. Sinclair and Sons produced the plates, a total of 12, which were drawn by James Henry Blake. In 1881, Baird wrote to Allen (archival reference F) asking if he had received the cetacean proofs that Baird forwarded from the public printer. Allen replied in the affirmative. It is not clear why the proofs were originally sent to Baird (archival reference I). ILLNESS Allen came down with a "serious attack of pleurisy" (Allen, 1916:32) in December of 1881. Allen's poor health took him to Colorado for several months and there is correspondence from both Hayden and Coues asking him to publish the edited por­ tion of the bibliography and to finish the remainder when he re­ gained his health. Following the publication of part of the work, correspondence continued, including a letter from Alex­ ander Agassiz to Baird concerning a missing manatee plate from Allen's work (Smithsonian Archives RU 30, box 1). It was clear that Allen's illness strongly affected his working ability. In his autobiography (Allen, 1916:33) he noted that on returning to Cambridge in September of 1882 (after spending the summer in Colorado Springs on the advice of his physicians) he learned that his medical difficulties were caused by a "ner­ vous breakdown." Considering the state of psychiatry in 1882, that diagnosis could encompass a number of disorders. He con­ tinued to work part-time at the MCZ until the spring of 1885. CURATORSHIP AT THE AMERICAN MUSEUM OF NATURAL HISTORY In 1885 the financial affairs of the Museum of Comparative Zoology took a downward turn (Allen, 1916:33). At that time three opportunities presented themselves: (1) Allen could con­ tinue to work at the MCZ but at the risk of the position's being terminated because of lack of money; (2) he could accept a po­ sition on the United States Geological Survey; or (3) he could accept a curatorship that had been offered to him by the Ameri­ can Museum of Natural History. He chose the last alternative, and in his autobiography he indicated that this was with the blessing of Agassiz (Allen, 1916:33). This assertion of Agassiz's approval is in contrast to the sen­ timents Agassiz expressed in a note appended to Allen's letter of resignation (25 April 1885; archival reference G) from his post at the MCZ. In that note Agassiz indicated that he had al­ lowed Allen to remain on full salary for the past three years while he recovered from his illness with the expectation that Allen would resume his post at the MCZ. A note written by Allen in 1908 in his "North Atlantic Right Whale and its near Allies" (pp. 279-280) indicates that, follow­ ing the reorganization of the USGS, there was no money to publish the remaining zoological work begun under Hayden. He also indicates that although ill health was initially the im­ pediment to finishing the work, other interests later took prece­ dence. The AMNH had obtained considerable material of the North Atlantic right whale through the efforts of Roy Chapman Andrews and the shore whalers of southern Long Island (An­ drews, 1908). This prompted Allen to publish the right whale section of his monograph (Allen, 1908). NUMBER 93 359 DEMISE OF THE UNITED STATES GEOLOGICAL AND GEOGRAPHICAL SURVEY OF THE TERRITORIES The United States Geological and Geographical Survey of the Territories, led by F.V. Hayden, was charged with exploring and assessing the geological and agricultural prospects of the areas now known as Colorado, Wyoming, North and South Da­ kota, Nebraska, and Michigan (Nelson et al., 1981). The results of the survey were published in several forms, as annual re­ ports, bulletins, maps, and miscellaneous publications. Hayden was personally involved in the printing of most works and cor­ responded frequently with both the authors and the printers. Thomas Sinclair and Sons, in Philadelphia, did almost all of the plates associated with the survey publications. On 30 June 1879, the Hayden Survey of the Territories was officially merged with the Powell and King Surveys into the United States Geological Survey, with Clarence King as direc­ tor (Records of the 45th Congress; see Appendix). The politics surrounding the merger were quite unpleasant, and Hayden was disappointed at not being named director. He did, however, stay on as a geologist. Congress designated $20,000 to finish the publications begun under Hayden. In 1886, Hayden retired from the USGS and moved to Phila­ delphia, where he died on 22 December 1897. His wife, Emma Woodruff Hayden, lived until 1934 but did not deposit his per­ sonal papers with any institution. Edward Woodruff Johnson, her nephew, was beneficiary in her will. There remained a close friendship between the Hayden/Woodruff families and the Allen family, as shown by a gift of Allen's personal copy of his autobiography from his wife, Susan Allen, to one of the Woodruffs. This copy is currently in the library of the AMNH. It is possible that the manuscript has been preserved in the Woodruff family papers, but to eliminate this possibility would require extensive genealogical work and contacting collateral descendants. To establish the preceding chain of events, we carefully ex­ amined the records of the USGS. All incoming and outgoing correspondence of the Hayden Survey as well as the new Geo­ logical Survey under King were examined for the years 1874-1882. Few references to Allen exist. The only letters from him are in support of Hay den's appointment as director or in reference to the rodent or pinniped works. No mention was made of the cetacean work. Hayden wrote and received many letters from both the public printer and T Sinclair and Sons, none of which mention Allen. Records of the USGS (Record Group 57) show accounts and appointments. Hayden's appoint­ ment as geologist is included as well as his notes acknowledg­ ing receipt of his paycheck. Most of these records are on mi­ crofilm. The loose papers yet to be photographed also were examined and yielded nothing more. No mention was made of Allen's being commissioned or receiving payment for his work on cetaceans. Perhaps this explains why the funds specified by Congress for printing the remaining Hayden works did not in­ clude Allen's bibliography. The publication records of the Hay­ den Survey indicated that all printing was done through the Government Printing Office, so the archival material in Record Group 157, containing all GPO records, also were examined. This included incoming and outgoing correspondence and records of orders for printing. The public printers during the years 1880-1882 were John D. Defrees and S.P. Rounds. The chief clerk, involved in most correspondence, was A.F. Childs. In the large volume of correspondence between T. Sinclair and Sons and the GPO, no letters pertain to Allen's work. Among the records of Orders for Executive Printing, how­ ever, is a reference to an order for 3000 copies of "Bulletin Vol. VI, no. 3," dated 3 October 1881. A later entry (17 July 1882) mentions an order for 200 copies of "Hayden's bull. vol. 6, no. 3 (Allen)." This could be the author's reprint request. Each or­ der at the GPO was given its own folder, called a "white jacket," which contained all the information pertaining to that job. Unfortunately these have all been destroyed. Dr. Elliott Coues, an Army surgeon, took over as "acting di­ rector of the ex-Survey" (letter from Coues to Leidy dated 15 July 1879). He also was named Honorary Curator of Mammals at the Smithsonian and seemed to use that office to discharge his duties as editor of the Hayden Survey publications. There are several references by Coues concerning proofs being sent to both Allen and a proof reader, Mr. Young (letter from Coues to Leidy dated 20 January 1879). He maintained a consistent corre­ spondence with both LeConte and Leidy, as well as with Allen. Coues and Allen had coauthored a monograph on North Ameri­ can rodents for the Survey and were close personal friends. All correspondence for 1880-1882 was checked and, other than the above notes, there was no mention of the Allen manu­ script. No letters to or from Allen were found. Coues seems to be the major correspondent concerning Hayden's reports but mentions only volumes 13 and 14. Hayden remained involved until 1880, mostly working on the copy and prints for the 1878 annual report and final report, volume 8. There was some prob­ lem with timing and funding and Sinclair was asked to hold up work on the plates until Hayden came forth with the copy. The time lags and sequences of printing the survey material were quite erratic. It is somewhat curious that Allen is not mentioned at all, as the pinniped work came out in 1880 and the first part of the cetacean bibliography in 1882. Some correspondence refers to proofs but there is only one letter specifically addressing the cetacean bibliography (archi­ val reference D). It also is curious to note that Coues does not appear on the USGS payroll but continued to use the United States Geological and Geographical Survey of the Territories letterhead for related correspondence until at least 1882. It is not clear how the payment of authors or engravers was handled. One reference (letter from Coues to LeConte dated 19 November 1879) indicates that Coues sent the bills and origi­ nals to Hayden. A letter from Hayden to LeConte in February 1879 said that he was not able to pay for a commissioned article at that time. A later letter from Hayden to the director of the USGS in December 1879 indicates that he was trying to finally 360 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY settle survey accounts with the treasury department. There were no other ledger accounts except Hayden's salary after 1880. ARCHIVAL SOURCES CONSULTED None of the archival sources yielded the actual manuscript or contained more than limited references to it. It seems curious that, during a time noted for frequent correspondence between colleagues, Allen is not better represented in the papers of Leidy, LeConte, Hayden, or Baird. The material in the National Archives pertaining to the USGS and the GPO has been care­ fully searched and all relevant facts have been discussed above. It is not worth looking there any further. All Philadelphia sources have been covered with the exception of Edward Woodruff Johnson, a descendant of Hayden, who may still be alive and possibly in possession of some of Hayden's personal papers. Letters to other archival holdings of Hayden' papers have revealed little. Tracing the plates through Thomas Sinclair and Sons provided information indicating that any letter copy was done by the GPO. Plates for the Allen Monograph While the senior author (Mead) was a student at Yale Univer­ sity in the mid-1960s, he developed a rapport with the staff of the MCZ in Cambridge. Mead was collecting loose litho­ graphic plates that Othniel Charles Marsh, of the Yale Peabody Museum, had produced. Mead followed a lead given him by William D. Sill, a student in Vertebrate Paleontology at the MCZ, and approached the librarian of that museum, who was rumored to have several sets of loose lithographic plates for sale. This turned out to be the case and Mead bought one set. When he opened it he was delighted to find that it contained several Marsh plates that he did not have plus some other mis­ cellaneous plates. It was common practice in the late nineteenth century for authors of monographs to have the plates printed separately from the text and to make extra copies of them to give to colleagues. It was not until Mead was going through the miscellaneous plates in the mid-1970s that he discovered a complete set of the plates for Allen's monograph on the Ceta­ cea. The plates were printed as quarto illustrations, indicating that Allen had planned a quarto monograph and not an octavo, as was the case with his monograph on pinnipeds. They are re­ produced herein (with the exception of Plate VIII, which was not found), and as complete a description of them as possible is given in the plate legends. The legends include the scientific names as they appear on the original plates, with the currently recognized names in brackets. CONCLUSIONS Some facts definitely have been established: (1) The mono­ graph was assigned to be volume 6, part 3 of the Bulletin of the United States Geological and Geographical Survey of the Ter­ ritories (Hayden Survey). The first part of the Bibliography was published under this reference. (2) Printing of the bulletin was done by the GPO. It is unclear who printed the reprint of Allen's partial bibliography, but most probably it, too, was printed in Washington, D.C. Reference was found to a law against printing anything at government expense except by the GPO. (3) Both Baird and Allen had access to the galley proofs at one time. Probably Coues and Hayden maintained copies also. Anything sent back to the GPO was either printed or de­ stroyed. (4) A copy of the proofs was still in Allen's possession as late as 1916, as he mentions it in his autobiography and bib­ liography. Most of Allen's personal papers are located at the AMNH. The main archives have two scrapbooks as well as miscella­ neous correspondence, none of which pertains to the cetacean bibliography. Some manuscripts have been cataloged in the main library, but Allen's is not among them. The Ornithology Department has several file drawers full of Allen's correspon­ dence. These were searched thoroughly, but unsuccessfully, for the manuscript and for letters to Coues, Hayden, Baird, and Thomas Sinclair and Sons. The archives of the Ornithology Department has a large collection of letters between Allen and Coues. It was here that mention was made of the bibliography and of Allen's health problems. The Osborn Library in Verte­ brate Paleontology has copies of the plates but no manuscript. The Mammalogy Department archives have recently been reorganized and, although the actual manuscript was not found, several interesting facts appeared. In 1921, all of Allen's books were purchased by the department for $200. Cleveland Allen wrote in the 1930s asking for permission to look at his father's letters and papers. Dr. Frederick Lucas petitioned to have an­ other of Allen's papers published posthumously. There also was a request from the South Dakota State College library for letters between Coues and Allen. It is certainly worth checking with them for further leads. The authors would like to note that they have found Allen's published bibliography (Allen, 1882) so useful that they have put it into computer-readable format and used the extensive an­ notations to keyword it. It is available on-line as a part of the Smithsonian Institution Research Information System (SIRIS). NUMBER 93 361 Appendix Archival References A. Geological Survey, Records of the 45th Congress, Session III, Chapter 182, 1879, pages 394-395. B. Letter from Elliott Coues to Joseph Leidy, dated 15 July 1879, in the Leidy Collection of the Hayden Papers (567), Academy of Natural Sciences, Philadelphia. C. Letter from Elliot Coues to Joseph Leidy, dated 20 January 1879, in the Leidy Collection of the Hayden Papers (567), Academy of Natural Sciences, Philadelphia. D. Letter from Elliott Coues to John LeConte, dated 19 No­ vember 1879, in the Elliott Coues Papers, American Philo­ sophical Society, Philadelphia. E. Letter from Joel Allen to Spencer Baird, dated 11 May 1880, in the records of the Office of the Secretary (Spencer F. Baird), 1879-1882 (Smithsonian Archives, RU 28, box 14). F. Letter from Joel Allen to Spencer Baird, dated 26 Septem­ ber 1881, in the records of the Office of the Secretary (Spencer F. Baird), 1879-1882 (Smithsonian Archives, RU 28, box 37, folder 19). G. Letter from Joel Asaph Allen to Louis Agassiz, dated 25 April 1885, resigning Allen's position at the Museum of Comparative Zoology; with a note in Agassiz's hand ex­ pressing dismay at the letter (Special Collections, MCZ, bag 33.10.1). H. Letter from Joel Asaph Allen in Cambridge, Massachusetts, to Spencer Fullerton Baird, Assistant Secretary of the Smithsonian Institution, dated 9 January 1877, in the records of the office of the Assistant Secretary (Smithso­ nian Archives, RU 52, box 38, folder 192). I. Letter from Spencer Fullerton Baird to Joel Asaph Allen, dated 14 December 1881, about the proofs that Baird re­ ceived from the public printer and was forwarding to Allen (Smithsonian Archives, RU 33, file 115b-480). Literature Cited Allen, J.A. If 1882. 1908. History of North American Pinnipeds: A Monograph of the Wal­ ruses, Sea-lions, Sea-bears and Seals of North America. United States Geological and Geographical Survey of the Territories, F. V. Hayden, Geologist-in-Charge, Miscellaneous Publications, 12: xvi + 785 pages, 60 woodcuts. Preliminary List of Works and Papers Relating to the Mammalian Orders Cete and Sirenia. Bulletin of the United Stales Geological and Geographical Survey of the Territories, 6(3):399-562. [Re­ printed by Arno Press, New York, 1974.] The North Atlantic Right Whale and Its Near Allies. Bulletin of the American Museum of Natural History, 24:277-329. 1916. Autobiographical Notes and a Bibliography of the Scientific Publi­ cations of Joel Asaph Allen, xi + 215 pages. New York: American Museum of Natural History. Andrews, R.C. 1908. Notes upon the External and Internal Anatomy of Balaena glacialis Bonn. Bulletin of the American Museum of Natural History, 24(10): 171-182. Nelson, CM., M.C. Rabbitt, and FM. Fryxell 1981. Ferdinand Vandeveer Hayden: The U.S. Geological Survey Years, 1879-1886. Proceedings of the American Philosophical Society, 125(3):238-243. 362 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY . -JPVEY OF THE TERRITORIES. Pt.ATE I. .1 !l H M . - . l . - i D A U ^ U A MYSTICETUS PLATE I.—Balaena mysticetus: Right lateral view of skull and mandible; lateral view of right scapula; medial view of right flipper. Specimen unknown. NUMBER 93 363 U 6. GEOL SURVEY OP THE TERRITORIES PLATE II J H.BIakrdal k s»B :,iih »r.ii«ac B A L A E N A C1SARCTICA PLATE II.—Balaena cisarctica [Eubalaena glacialis]: Left lateral view of skull and mandible. Skull in MCZ, taken in April 1864 at Provincetown, Massachusetts. Reproduced as Plate XIX in Allen, 1908. US GEOL, SURVEY OF THE TERRITORIES SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY PLATE m T i i n c l u i b SOT-. Liu BALAENA CISARCT1CA PLATE III.—Balaena cisarctica [Eubalaena glacialis]: 1, dorsal view of skull and mandibles; 2, dorsal view of left mandible. Skull in MCZ, taken in April 1864 at Provincetown, Massachusetts. Reproduced as Plate XX in Allen, 1908. NUMBER 93 365 U S . GEOL SURVEY OF THE TERRITORIES PI.ATE IV J.H BUkedrl T Smclvr k Sou Laih T i^lftAi BALAENA C1SARCTICA PLATE IV—Balaena cisarctica [Eubalaena glacialis]: 1, posterior view of skull and mandibles; 2, anterior view of right tympanoperiotic; 3, medial view of right tympanoperiotic; 4, lateral view of right tympanoperiotic. Skull in MCZ, taken in April 1864 at Provincetown, Massachusetts. Reproduced as Plate XXI in Allen, 1908. 366 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY U S GEOL SURVEY OF THE TERRITORIES PLATE V T SiaolUT lc Sot. Ll