ARBOREALITY IN TREE SQUIRRELS (SCIURIDAE) RICHARD W. THORIN.GTON, JR., AMY M.'L MILLER, AND CHARLES G. ANDERSON Department of Vertebrate Zoology, Smithsonian Institution, Washington, DC. 20560 (RWT) Department of Ecology, Evolution, and Marine Biology, University of California, Santa Barbara, CA 93106 (AMLM) Department of Zoology, University of Tennessee, Knoxville, TN 37916 (CGA) ABSTRACT.?The purpose of this paper is to examine the hypothesis that arboreality evolved more than once among extant tree squirrels. The North American fossil, Douglossia (formerly Protosciurus), from the late Eocene is considered to have been an arboreal squirrel (Emry and Thorington, 1982). Thus, arboreality may be primitive for squirrels, although the European Palaeoscinrus from the early Oligocene was probably terrestrial (Vianey-Liaud, 1974). It is un- clear how these are related to the Recent tree squirrels, the Sciurini of the Americas and Northern Eurasia, the Ratuflni and Callosciurini of Southern Asia, and the Protoxerini and Funambulini, mostly of Africa. Three hypotheses of relationships among Recent squirrels im- ply a terrestrial origin for some group of tree squirrels: Moore (1959) hypothesized that Sciwotamias and Tamiasciurus are closely related and that the North American red squirrel is derived from a terrestrial ancestor, Callahan and Davis (1982) hypothesized that Sciwotamias and Ratufa are closely related and suggested a terrestrial origin for the Asian giant tree squir- rels, and Moore (1959) also suggested that the African tree squirrels, Protoxerini and Funambulini, may have evolved from xerine ground squirrels. A review of morphological evi- dence suggests that Sciwotamias is more closely related to chipmunks, the Tamiina, than to either Tamiasciurus or Ratufa. This contradicts the first two hypotheses cited above and the suggestion that arboreality evolved independently in these two tree squirrels. We cite reasons for doubting the third hypothesis, that African tree squirrels have a terrestrial origin, but do not critically examine it in this paper. INTRODUCTION The squirrel family, Sciuridae, includes two radiations of ground squirrels (Marmotini, Xerini) and a larger number of radiations of tree squirrels in South America, North America, Eurasia, and Africa. Formerly, it was presumed that arboreal squirrels evolved from terrestrial squirrels. For example, Black (1963) considered the chipmunks (tribe Tamiini) the basal group from which all other squirrels evolved. Similarly, Moore (1959) thought that the African tree squirrels were derived from the African ground squirrels. However, the earli- est fossil squirrel, Douglossia jeffersoni formerly Protosciurus c.f. jeffersoni), from the late Eocene, 35 Ma, was a tree squirrel (Emry and Thorington, 1982; Emry and Korth, 1996). The earliest chip- munks date from the early Miocene, ca. 25 Ma In M.A. Steele, J. F. Merritt, and D. A. Zegers (eds.). 1998. Ecology and Evolutionary Biology of Tree Squirrels. Special Publication, Virginia Museum of Natural History, 6: 320 pp (Black, 1963; Bruijn, et al., 1980). Thus, it is pos- sible that arboreality is primitive for squirrels and that extant terrestrial squirrels have evolved from tree squirrels once among the Marmotini of North America and Eurasia, independently in the Xerini of Africa and Southern Asia, and also indepen- dently in most tribes of tree squirrels. We here examine several hypotheses that arboreality has evolved more than once in the family Sciuridae. Moore (1959) proposed that Sciurotamias, the Chinese rock squirrel, and Tamiasciurus, the North American red squirrel, are closely related, and that the red squirrel evolved from a terrestrial rock squirrel ancestor, similar to S. davidianus. Callahan and Davis (1982) presented an alterna- tive hypothesis about Sciurotamias, arguing that 119 V VIRGINIA MUSEUM OF NATURAL HISTORY it is closely related to Ratufa They explicitly sug- gested that Ratufa evolved from a terrestrial an- cestor. Thus, the unresolved phylogeny of Sciurotamias, the Chinese rock squirrel, is central to the question of whether arboreality evolved more than once among tree squirrels. Moore (1959) also suggested that the African tree squirrels evolved from the xerine ground squir- rels. This hypothesis has not received careful con- sideration from subsequent investigators. Moore listed four characters shared by the Xerini and the Protoxerini. One of these is the separation of the buccinator and masticatory foramina. This condi- tion is primitive for rodents (Wahlert, 1991) and therefore does not serve as a good phylogenetic indicator. A second is the prominence of the mas- seteric tubercle, but this is found in only one ge- nus in the Protoxerini. Therefore, it is likely that it evolved independently in the Protoxerini and the Xerini. This leaves two characters, long orbits and short interorbital width, purportedly supporting Moore's hypothesis. In details the orbits of the Xerines and the Protoxerines are very different, however, especially in the shapes of the lacrimal, jugal, and maxillary bones. On the basis of cranial evidence, Lavocat (1973) considered the earliest fossil squirrels in Africa to be tree squirrels de- rived from southwest Asia. This also weakens Moore's speculation; but we will not critically ex- amine it in this paper. In this contribution, we examine and test four hypothesized sister groups: Tarniasciurus-Sciurus, Tamiasciurus-Sciurotamias, Tam.ias-Sciurotam.ias, and Ratufa-Sciurotamias (Fig. 1), challenging the Sciurotamias Tamiasciurus Sciurus two proposed ground squirrel-tree squirrel hypoth- eses, each with an alternative hypothesis. THE STATUS OF THE TAMIASCIURINI Moore (1959) created a tribe, the Tamiasciurini, for Tamiasciurus and Sciurotamias because both have three transbullar septa, a rare condition among squirrels. Both Moore (1959) and Black (1963) noted that Tamiasciurus is polymorphic for two and three transbullar septa. Corbet and Hill (1992) list two transbullar septa for Sciurotamias forresti and three for Sciurotamias davidianus. Thus, both genera are polymorphic for number of transbullar septa. Callahan and Davis (1982) showed that Sciurotamias lacks the specialized reproductive tract of Tamiasciurus. These obser- vations detract from the likelihood that Moore's hypothesis is a good one, but do not test it. The hypothesis would be falsified by finding many de- rived features shared by Sciurus and Tamiasciurus, but not by Sciurotamias. It is not falsified by show- ing that proposed synapomorphies of Tamiasciurus and Sciurotamias are invalid. Bryant (1945) de- scribed many osteological and myological similari- ties between Sciurus and Tamiasciurus, conclud- ing that Tamiasciurus should not be considered a distinct genus from Sciurus. He was very careful to distinguish between primitive and derived char- acters, but he did not study Sciurotamias. Similarity, the immunological study by Hight et al. (1974) and the protein electrophoresis study by Hafner et al. (1994) suggested that Sciurus and Tamiasciurus are closely related, but neither study Sciurus Tamiasciurus Sciurotamias Ratufa Tamias Sciurotamias Tamias Ratufa Sciurotamias C/ D, Fig. 1.?Sister-group hypotheses examined: A. Commonly accepted hypothesis that Sciurus and Tamiasciurus are more closely related to each other than either is to Sciurotamias, B. Moore's (1959) hypothesis that Tamiasciurus and Sciurotamias are more closely related to each other, C. Milne-Edwards' (1871) hypothesis that Sciurotamias is closely related to chipmunks, and D. Callahan and Davis (1982) hypothesis that Sciurotamias is more closely related to Ratufa. 120 included Sciurotamias <=n ?, ^ Moores hypothe^Tfn spi elf 1? T COntrad-t nesses of these "tests" of Mn ! l0gicaI Weak" concept of the Tal ?f M??res hypothesis, his doned (Hoffmann et a" f^ COmP^Y aban- THE STATUS OF THE RATUFiMi - M1asc^^nus^-f the two genera arcloselyS5 ^ SUggGSted tha* be included in the tribe R^/r ^ Should b?th -any ways these t? 1^^ ^ ****** In ferent squirrels. To^s fif* remarkably dif- (1954:479) showing the bac?hgUr^4 ?f Prasad glans penis with rnany ridge"" ?c "^ ^ the cus, looks very different frnSr & VCntral sul" and Davis (1982-45) ^h gUre 3 ofCallahan SciurotamicL X?^*? *?*? Penis of Question the basis forfh ^ Thus' We Sciurotamias in the RatufiniTh P,laceme"t of Pouch and the hoof-Kketl ' Y "St the cheek dence. However &L?S ^ as Sporting evi- ?d Kata/a la^ teTuXmo8 * *"* P?UCh served that Rata/a's-hoonn;,?1"6, We have <*- by a modified firs" ?* - supported terminal phalanx but th^ n * br?ad" flattened not. Agai^ the* crit cTsm's ^ ^oto?- is Pothesis, which has genSv h ' **' their h^ Corbet and Hill, 199| "X^"?*** (e.g., test it, we compare it with Th' ? &1' 1993^ T? that ScturofamL is ^ *e ^ hypothesis chipmunks (Milne-Edwardsist-tn^ t0 the Ellerman 194n- r?~ 18b/- Miller, 1901- 1976). 94?' Gr?mOV' <* "1.. 1965: CallahM.' TH0BINGTONETAL_ARBOREAUTY|NTRE SO thev do nnt ^?.__J. . 'HMtLb ?ts^ and Thorington P1982 oL SCiUrids (E?y Primitive state is listed as fof^fment ?f th^ states are listed as (la) and (ibi f S^ deriVed UJNo septa: ifota/a and noted ^^?2T^ C?^ half transbullar seSa ThT," ^ ^ and one- modal number for sXJsj!fe- the primitive be two or three transbXslT^f' and COuId derived. iar sePta. Lack of septa is CHARACTER LIST AND POLARITY w ASSESSMENTS eluded in studieVby BrZtTfp^ ?MHCtere '"" and others. We have X 'i 45)' M??re < 195^ our studies of wrists anHo,rerCharacters from ?(mn,is is ??, , sessments .rTCeT^T' ^ P?I&rity as~ ^"le ^isTs in aT ^ " ^ ProfosaUrus cranip, . / morphologies of assess. 7? greementwith and Posted 0f SoXsr ^H * BIaCk ^ ^^ r?dents- and Thorington U^T^Pn,^^^^ scribed by Vianey-Liaud (1974 ^^T &S de' -odents, as assessed by WahSU^L?f Primitive consider the closely^ehSL? D / I99l)" We y iated ouglassia and ^ngTon aSrT ^'^ ??"*. of bulla anterior a^sZriorPetri0iedge; porti?n Pollen: Scftinxs. TamiS ? thC meatus ?t (1) Lacks flange^S ;/"*' Rat"f?- -Penor to it sw^tC^ ~ -nd The primitive conditinr, unswollen bulla (WahlertZou ^dents is a" lustration of Pr0toJctimZ*9^11 Black's ^963) and unswollen ^ZTerZeT^ & ^ Scurotamios is consideredf derived C?nditi?n in Character 3. Postglenoid region of the ^n Pierced bv hnth subsquamosal foramina- P?Stglenoid and onskJu!rriffr?r^ Sc^rotamios, TamiJ suPraglenoid foramen- ouite^S^^j-^ndtheg^noidis absent, someumes fused Iyn/y "" som^imes Particularly in SciuZ^^^^ mUlt*le' squirrels is not cleaTand it condition in variable. This is i T ?7uf 1? haVe bee" assessment for ,?S2 WaWert'S ^ C ^aZrJ^^l{?^nat fOJ Present: sSSS T We ^ 2) mnis (most). S' Tam^ciurus, Ratufa species^. SdUroto^ Thn*,.. ^atu/a other 121 VIRGINIA MUSEUM OF NATURAL HISTORY H^rv Fig. 2 ? Ventral views of rostrum and lateral views of skulls of Sciurotamias davidianus, USNM 155219, (above) and Ratufa bicolor, USNM 481228, (below). Arrows indicate fossae behind the incisors of Sciurotamias and the position of the infraorbital foramen of Ratufa. Note also the large sphenopalatine foramen in the orbit and the presence of the temporal foramen in Ratufa. The presence of a foramen at the squamosal- parietal suture is primitive for rodents (Wahlert, 1974), and probably primitive for squirrels (Wahlert, 1991). It is neither mentioned nor figured by Black (1963) in Protosciunis. Character 5. Supraorbital notch (Fig. 2) (0) Absent or inconspicuous: Ratufa (usually). (1) Open and usually trenchant: Sciurus, Tamiasciurus, Sciurotamias, Tamias. Black (1963) comments on the absence of the supraorbital notch in Protosciunis condoni, but it appears to be present in his plate 3, figure la. Vianey-Liaud (1974) shows prominent supra-or- bital notches in Palaeosciurus goti. We have con- sidered its presence to be derived in squirrels, but it is possible that this polarity should be reversed. Character 6. Sphenopalatine foramen (Fig. 2) (0) Large, approximately size of sphenoidal fis- sure: Sciurus (most), Tamiasciurus, Ratufa. (1) Sphenopalatine foramen small: Sciurus granatensis, Sciurotamias, Tamias. Black (1963) illustrates the sphenopalatine foramen in Protosciurus condoni, showing it to be large. We take this to be primitive for squirrels. Character 7. Sharp anterior edge of zygomatic arch (Fig. 2) (0) Does not extend onto the premaxillary bone: Sciurus griseus (the anterior masseter extends onto the premaxillary bone but the sharp ridge does not), Tamiasciurus, Sciiirotarnias, Tamias (most). (1) Extends onto the premaxillary bone: Sciurus (most, but barely in S. anomalus and S. granatensis), Tamias (some), Ratufa. The evolution of sciuromorphy involves the migration of the masseter onto the zygomatic plate and side of the rostrum. The more the masseter musculature extends onto the premaxillary bone, the more derived we consider it. Character 8. Angle of zygomatic plate (Fig. 2) (0) More horizontal: Sciurotamias, Tamias. (1) More vertical: Sciurus, Tamiasciurus, Ratufa. Black (1963) describes the evolution of the zy- gomatic plate from a more horizontal angle to a more vertical angle in his treatments of Miocene Protosciurus rachelae and Sciurus. The derived con- dition is associated with the enlargement of the area for origin of the anterior portion of the deep lateral masseter muscle. 122 Character 9. Length of infraorbital canal (Fig. 2) 0) Moderately long: Sciurus, Tamiasciurus suture^ ^rending t0 m^-premaSLy nJ^SkT 6Xtend^ "to the side of the eratSv WSdUrUS^thC infraorbital canal is mod- erately long, according to Black fl9RS) Th? i -g7p^^ct%xmediai ? ~?Z Srsg01d ridge wea* ?? ? ?<- in. froL^^ Fte7g?ld ndSe Pn>mtoent. divere- r^?cu^d,al "* P?"laterally: Sc^. s prr:sr ?r ^ ^ are considered to be delved ? m?^hol?^ THOR.NGTON ET AL.-ARBOREALITY (N TREE SQU(RRELs the ^i^Srtin^,^^ ^ ?f (Black, 1963- pfate t ft f fghtly behind M3 consider thfs pnrnitfve ^ 1C) ^ We therefore wZtV 12; D?rSal Pr?flle of**"" 0 Flat. Sourotamfas. Tamias, Ratufa (1) Rounded: Scftmxs. Tamiasciurus rounded cranial In , * scluirrels have more Zam^erT13- SquamostosJ*?op/,i!us (Black. SSat They earetea,fddle """*? SpermophilinusbrfovL? .1 ? Present in (Bruin and Mein Ines) A "f"""6 M,ocen<= eheekpoueheSasta?g8Je,tCd0r*ndgly' We treat date from the Early Miocene wh' ?- SSUme ""* in the f08sU reeord or sSCe^rrheT^ are ateo present in r? u , earner- The fossae eep, ^ nrdrfuS:crteanaXeraspnn- Character 15. CcraaM process ?, ^^ 123 VIRGINIA MUSEUM OF NATURAL HISTORY Fig. 3.?Medial views of scapulae of Sciurotamias davidianus, USNM 258508, (left) and Tamiasciurus hudsonicus, USNM 505581, (right). Note length of coracoid process (arrow) and prominence of fossa for teres major muscle. The coracoid of Douglossid was broken, but it was clearly similar to Sciurus and larger than in ground squirrels. Character 16. Axillary ridge relative to the surface of scapula (0?) Perpendicular and high: Sciurus, Tamiasciurus. (la?) Perpendicular and low: Ratufa. (lb?) Perpendicularity intermediate and high: Sciurotamias. (2?) Least perpendicular but high Tamias. Sciurid scapulae are very rare in the fossil record and we know of none that show the axillary ridge. Character 17. Flange near caudal angle of scapula for teres major muscle (Fig. 3) (0?) Prominent: Sciurus, Tamiasciurus, Ratufa. (1?) Small: Sciurotamias. (2?) Almost absent: Tamias. Character 18. Subscapular spine (Fig. 3) (0?) Single: Sciurus, Tamiasciurus, Sciurotamias, Tamias. (1?) Sometimes double: Ratufa. The broken scapula of Douglassia jeffersoni shows a single subscapular spine. Character 19. Metacromion process of scapula (0?) Less broad and at angle to plane of scapula: Sciurus, Tamiasciurus, Sciurotamias. (la?) Narrow and abruptly parallel to plane of scapula: Ratufa. (lb?) Broad and most parallel to plane of scapula: Tamias. The metacromion process is unknown for fos- sil sciurids. Character 20. Acromion process of scapula (0?) Broad: Sciurus, Tamiasciurus, Ratufa. (1?) Narrow: Sciurotamias, Tamias. Character 21. Deltoid and pectoral ridges of humerus (0) Deltoid ridge weak proximally: Sciurus, Ratufa. (1) Deltoid ridge prominent proximally, diverg- ing from pectoral ridge: Tamiasciurus, Sciurotamias, Tamias. In Douglassia the deltoid and pectoral ridges are damaged, but appear similar to those of Sciurus. Character 22. Orientation of delto-pectoral crest of humerus (0) Directed medially with prominent bicipital groove: Sciurus, Tamiasciurus, Ratufa. (1) Directed laterally with less distinct bicipi- tal groove: Sciurotamias, Tamias. This region was broken on Douglassia but ap- peared to be similar to Sciurus. Character 23. Entepicondylar foramen of humerus (0) Usually present: Sciurus, Tamiasciurus, Ratufa, Sciurotamias (1) Usually absent: Tamias striatus. The entepicondylar foramen is commonly present in rodents, including most Recent squir- rels, Douglassia, and Paleosciurus. Character 24. Medial epicondyle of humerus (0) Elongate medially: Sciurus, Tamiasciurus, Sciurotamias, Ratufa. (1) Not elongate: Tamias. The medial epicondyle was elongate in Douglassia and in Paleosciurus. Character 25. Extent of the radial notch on ulna (0) Approximately one-third the width of the semilunar notch: Sciurus, Tamiasciurus, Ratufa. (1) Approximately one-half of the width of the semilunar notch: Sciurotamias. (2) More than one-half of width of the semilu- nar notch: Tamias. This feature in Douglassia is very similar to that in Sciurus. 124 THORINGTON ET AL? ARBOREALITY IN TREE SQUIRRELS Character 26. Prominent ridge for pronator quadratus muscle on ulna (0) Present: Tamias, Sciurotamias. (1) Absent: Sciurus, Tamiasciurus, Ratufa. The ridge is prominent in Douglassia. Character 27. Groove on pisiform bone (0) Pisiform ungrooved: Sciurus, Tamiasciurus, Ratufa. (1) Radial side of pisiform grooved near palmar end: Sciurotamias, Tamias. The pisiform of Douglassia is ungrooved, so this is presumed to be the primitive condition. The de- rived condition is not unique to Sciurotamias and Tamias, it is also found in Callosciurus to which they are not closely related (based on other evi- dence). Character 28. Shape of triquetrum (0) Triquetrum elongate and gracile: Sciurus, Tamiasciurus, Tamias, Ratufa. (1) Triquetrum short and robust: Sciurotamias. The shape of the triquetrum in Sciurotamias appears to be unique among squirrels and we think that it is derived. Character 29. Ventral articulation of metacarpal HI and metacarpal IV (0) Present: Sciurus carolinensis, S. griseus, S. granatensis, Sciurotamias davidianus, Tamias, Ratufa. (1) Absent: Sciurus niger, Tamiasciurus hudsonicus. Loss of this articulation is rare and is presumed to be derived. The presence of the derived condi- tion in Tamiasciurus and only part of Sciurus can be interpreted in two ways. First, it may be inde- pendently derived in the two genera. Second, it may indicate that some species of Sciurus are more closely related to Tamiasciurus than they are to other species of Sciurus. Character 30. Centrale-greater Multangular articulation (0) Absent: Sciurus, Tamiasciurus, Sciurotamias. (1) Present: Tamias, Ratufa. Primitively in squirrels the centrale articulates distally with capitate, metacarpal II, and lesser multangular. In several lineages it extends radi- ally, toward the pollex, and also articulates with greater multangular. Character 31. Articulation of metacarpal ID and hamate. (0) Corner of hamate beveled: Sciurotamias, Tamias, Ratufa. (1) Corner of hamate square: Sciurus, Tamiasciurus. The hamate is slightly beveled in Douglassia. The loss of this feature appears to be correlated with the narrowing of the hand and particularly metacarpal III. Character 32. Lateral iliac ridge (0) Broad, distinct ridge: Sciurus, Tamiasciurus. (la) Broad, indistinct ridge: Tamias. (lb) Narrow, distinct ridge: Sciurotamias Ratufa. The iliac ridge of Douglassia is more pro- nounced than in most Recent squirrels, but is ap- proximated by some Sciurus. Character 33. Dorso-ventral depth of pelvic basin (0?) Deep: Sciurus, Tamiasciurus, Ratufa. (1?) Shallow: Sciurotamias, Tamias. It is not possible to determine this in Douglassia. Character 34. Pubic symphysis (0?) Normal: Sciurus, Tamiasciurus Sciurotamias, Ratufa. (1?) Short: Tamias. The pubic symphysis of chipmunks is shorter than the ascending ramus of the pubis, and the anterior end does not extend cranially to the pos- terior edge of the acetabulum. This seems to be unique to Tamias and is probably derived. Character 35. Lesser trochanter of femur (Fig. 4) (0) Directed medially: Sciurus, Tamiasciurus, Ratufa. (1) Directed postero-medially: Sciurotamias, Tamias. A B Fig. 4.? (A) Posterior views of proximal ends of femora, and (B) medial views of distal ends of tibiae of Sciurotamias davidianus, USNM 258508, (left) and Tamiasciurus hudsonicus, USNM 564083' (right). Arrows indicate lesser trochanter of femur and the anterior process of the medial malleolus of the tibia. 125 VIRGINIA MUSEUM OF NATURAL HISTORY The lesser trochanter of Douglassia is directed medially. This appears to be the case in Palaeosciurus goti as well (Vianey-Liaud, 1974). Character 36. Popliteal fossa of tibia (0) Deep with medial and lateral ridges: Tamiasciurus, Sciurotamias, Tamias. (1) Shallow and unridged, sometimes with medial ridge: Sciurus, Ratufa. In Douglassia, the popliteal fossa is prominent. Character 37. Relative lengths of the ante- rior and posterior processes of the medial malleolus of tibia (Fig. 4) (0) Anterior process slightly shorter than the posterior process: Tamias, Sciurotamias. (1) Anterior process distinctly shorter than the posterior process: Sciurus, Tamiasciurus, Ratufa. The anterior and posterior processes are ap- proximately the same length in Douglassia and Palaeosciurus goti. Character 38. Tarsalfoot pads (0?) Two: Sciurotamiasforresti, Ratufa, Sciurus niqer (rare), S. vulgaris (rare). (1?) One (hallucal): Sciurus (most), Tamiasciurus hudsonicus (rare). (2?) None: Tamiasciurus douglassi, i. hudsonicus (most), Sciurotamias davidianus, Tamias. r . We hypothesize that the presence of two tarsal pads is primitive and that loss of one or both is derived. However, in view of the great variation in this character, we are not very confident of this polarity. Character 39. Two calcaneal facets on astragalus (0) Separated by a shallow groove: Ratufa. (1) Confluent. Sciurus, Tamiasciurus, Sciurotamias, Tamias. In Douglassia the two facets are separated by a groove. Character 40. Reproductive tract (0) Large bulbo-urethral glands and presence of penile duct: Sciurus, Sciurotamias, Tamias, ^ U) Absence of separate bulbo-urethral glands and absence of penile duct. Tamiasciurus Large, distinctive bulbo-urethral glands occur in the flying squirrels and five tribes of the Sciurinae, as listed below, suggesting that they are primitive for the family Sciuridae. Pteromyinae (= Petauristinae): Glaucomys, Mossman, et al., 1932; 126 Protoxerini: Heliosciurus illustrated but misidentifiedby Kingdon, 1974; 372; Funambulini: Funisciurus illustrated but misidentified by Kingdon 1974; 373; Funambutus (in part), Siddiqi, 1938- Ratufini: Ratufa, Prasad, 1954; Sciurini: Sciurus, Mossman, et al., 1932; Marmotini: Tamias, Spermophilus, Mossman, et al., 1932. Large bulbo-urethral glands and a penile duct are absent in both Tamiasciurus (Mossman, et al., 1932) and Funambulus palmarum (Prasad, 1954). Character 41. Shape ofbaculum (0?) Baculum nearly symmetric and simple ? almost rod shaped, with distal end bent dorsally: Tamias, Sciurotamias, and Ratufa. (la?) Baculum asymmetric and flattened at tip. Sciurus. . (lb?) Baculum absent or minuscule. Tamiasciurus. There is no clear evidence on which to base a polarity assessment for the baculum of squirrels A simple, symmetric baculum would be a good model from which to derive the complex symmetri- cal bacula of ground squirrels and the complex asymmetric bacula of some other squirrels. An- other alternative is that the bacular morphology of Sciurus is close to primitive, because it is found in such a diverse group of squirrels as Sciunltus, Rheithrosciurus, and Petaurista. A feature not included in our list above is the external morphology of the glans penis, cited by Callahan and Davis (1982) as their justification for placing Sciurotamias in the Ratufini. We think that the penile morphologies of Ratufa and Sciurotamias are both derived but not homologous with one another. The morphology of the glans penis of Ratufa is illustrated by Hill (1936, 1940) and Prasad (1954) but interpretation of these draw- ings is not straightforward. The difference between the illustrations of Hill (1936, 1940) and Prasad (1954) appears to be in their definitions of dorsal and ventral. In comparing the three papers, Prasad (1954) should be viewed upside down. It also seems that the baculum illustrated in Hill (1936) is upside down relative to his illustrations of the penis but in either orientation it is difficult to visualize how the baculum could fit inside the il- lustrated penis. All three publications clearly illustrate 12-Zd ridges or folds, with a ventral sulcus on the glans penis of Ratufa. According to Callahan and Davis (1982), the ornamentation of the penis of Sciurotamias consists of four annuli, three going completely around the penis. No ventral sulcus is shown or described. To us this seems very different THORINGTON ET AL?ARBOREALITY IN TREE SQUIRRELS from the morphology of Ratufa, and we do not consider the two to be homologous. CHARACTERS SUPPORTING SISTER-GROUP HYPOTHESES We examined two competing hypotheses (Figs. 5A and 5B) for the phylogenetic placement of Tamiasciurus, Moore's hypothesis that Tamiasciurus and Sciurotamias form a sister group, and the alternative hypothesis that Tamiasciurus forms a sister group with Sciurus. The more strongly supported hypothesis is that it forms a sister group with Sciurus. Two characters, 5 and 39, are shared by both pairs and hence are shown at the base of the tree. The remaining six charac- ters include three in the cranium, two in the fore- limb, and one in the hindlimb. Derived characters shared by Tamiasciurus and Sciurus: Character 5: Supraorbital notch present. Character 8: Vertical angle of zygomatic plate. Character 10: Posteriorly diverging lateral pterygoid ridges. Character 12: Rounded skull profile. Character 26: Reduced ridge on ulna for pronator quadratus muscle. Character 31: Absence of beveled corner of hamate. Character 37: Anterior process distinctly shorter than the posterior process of medial mal- leolus of tibia. Character 39: Confluent astragalus facets. Sciurotamias Tamiasciurus Sciurus 8, 10,12.26,31,37 Sciurus Tamiasciurus Sciurotamias Ratufa Tamias Sciurotamias Tamias Ratufa Sciurotamias 4,5,6,9,14,15,21 22,25,27,35,39 Fig. 5.? Sister-group hypotheses examined. A.-D. as in Fig. 1, but with shared derived characters plotted on them. Numbers correspond to character list in text. Uniquely de- rived characters not plotted. Derived characters shared by Tamiasciurus and Sciurotamias: Character 5: Supraorbital notch present. Character 21: Deltoid and pectoral ridges di- verge proximally. Character 39: Confluent astragalus facets. The hypothesis proposed by Callahan and Davis (1982), that Sciurotamias forms a sister group with Ratufa (Fig. 5D), is supported by only one character. The original hypothesis of Milne- Edwards (1871), that the Chinese rock squirrel is more closely related to chipmunks (Fig. 5C), is sup- ported by twelve derived characters not shared with Ratufa. Nine of these characters are not shared with either Ratufa or Tamiasciurus - four of the cranium, four of the forelimb, and one of the fe- mur. Derived characters shared by Sciurotamias and Tamias: Character 4: Temporal foramen absent. Character 5: Supraorbital notch present. Character 6: Small sphenopalatine foramen. Character 9: Short infraorbital canal. Character 14: Cheek pouch present. Character 15: Short coracoid process of scapula. Character 21: Deltoid and pectoral ridges of humerus diverge proximally. Character 22: Delto-pectoral crest directed laterally. Character 25: Character 27: Character 35: Radial notch on ulna large. Pisiform grooved. Lesser trochanter of femur di- rected postero-medially. Character 39: Confluent astragalus facets. Derived character shared by Sciurotamias and Ratufa: Character 32: Narrow, distinct lateral iliac ridge. The characters we consider to be strongest are those associated with the zygomatic plate, the infraorbital canal, and the cheek pouch. These appear to have evolved early in the evolutionary history of squirrels (Black, 1963) and, based on our personal observations, show little interspecific and intergeneric variation. A smaller number of derived characters supports the tree squirrel hy- pothesis than the chipmunk-rock squirrel hypoth- esis. This probably results from tree squirrels re- taining more features of the primitive squirrel morphology. 127 VIRGINIA MUSEUM OF NATURAL HISTORY Tamias lacks several derived features of other members of the Marmotini, as described by Bryant (1945)- a second subscapular spine, the triangu- lar cross-section of the ilium, and the flaring of the ilium, for example. In these characters Sciwotamias is like Tamias, appearing to be primi- tive within the tribe. DISCUSSION The taxonomic position of Sciurotamias, its behavior, and its use of habitat, are central to the question of whether arboreality evolved more than once among tree squirrels. Moore (1959) placed it with Tamiasciurus on the basis of the presence of three transbullar septa. Callahan and Davis (1982) placed it with Ratufa on the basis of their percep- tion of the similarity of bacular and penile mor- phologies. We contend that the similarities in transbullar septa and bacula are primitive and that the features of the penis are not homologous. Our observations lead us to conclude that Sciurotamias belongs within the tribe Marmotini. While we con- tend that the characters shared by Sciurotanuas and Tamias are derived features of the Marmotini, we do not consider them to be derived features linking Sciurotamias with Tamias within the Marmotini. Thus, we provisionally place Sciurotamias within its own subtribe, the Sciurotamiina. Although one species of Sciurotamias, S. davidianus, is called a "rock squir- rel" and is considered to be terrestrial, the other species, S.forresti is described as an arboreal tree squirrel. Unfortunately, little is known about its anatomy or ecology. Similarly, some species of Tamias are quite arboreal, and others forage readily in trees. We suspect that this says more about the versatility of small squirrels and their retention of climbing abilities than it does about any indepen- dent evolution of arboreality. However, if cheek pouches evolved as an adaptation for terrestrial foraging, then arboreal squirrels with cheek pouches are potentially very interesting. Accord- ingly, the biology of Sciurotamias forresti of south- ern China deserves study. We agree with Black (1963) that the Tamiasciurini is not a valid tribe. We concur with his return of Tamiascvurus to the Sciurini, but not with his placement of Sciurotamias in the Callosciurini. Qiu (1991) also places the Sciurotamias lineage in the Callosciurini, although be argues convincingly that Sinotamias, a fossil from the upper Miocene of Inner Mongolia, is an- cestral to Sciurotamias and is closely related to Spermophilus, of the Marmotini. We think his evi- dence better supports our conclusion. Our place- ment of Sciurotamias close to Tamias agrees with the earlier assessments of Milne-Edwards (1871), Miller (1901), Ellerman (1940), Gromov et al. (1965), and the initial conclusions of Callahan (1976). Emry and Thorington (1982) presented evi- dence that Douglassia jejfersoni (of the late Eocene) was an arboreal squirrel. Black (1963) argued that Protosciurus rachelae of the early Miocene was a good morphological ancestor of the primitive Sciurus of the middle Miocene. Therefore, it is rea- sonable to contend that the tribe Sciurini has ar- boreal roots that extend back into the Eocene. The other tribes of tree squirrels probably share an- cestry with the Sciurini in the Oligocene. Thus, it is probable that their common ancestor was arbo- real, and that arboreality evolved only once in the extant Sciuridae. ACKNOWLEDGMENTS We thank V. Louise Roth, Robert S. Hoffmann, and two anonymous reviewers for their helpful recommendations. Special thanks are due to Karolyn Darrow, who assisted with the extensive final revision of the manuscript and prepared all the figures. LITERATURE CITED BLACK, C. G. 1963. A review of the North American Tertiary Sciuridae. Bulletin of the Museum of Comparative Zoology, Harvard University, 130:100-248. BRUIJN, H. DE AND P. MEIN 1968. On the mam- malian fauna of the Hipparion beds in the Calatayud-Teruel basin (Prov. Zaragoza, Spain). Part V. The Sciuridae. Proceedings of the Koninklijke Nederlandse Akademie van Wetenschappen, series B, 71:74-90. BRUIJN, H. DE., A. J. VAN DERMEULEN, AND G. KATSIKATSOS. 1980. The Mammals from the Lower Miocene of Aliveri (Island of Evia, Greece). Proceedings of the Koninklijke Nederlandse Akademie van Wetenschappen, series B, 83:241-261. BRYANT, M. D. 1945. Phylogeny of Nearctic Sciuridae. The American Midland Naturalist, 33:257-390. CALLAHAN, J. R. 1976. Systematics andbiogeog- raphy of the Eutamias obscurus complex (Rodentia: Sciuridae). Ph.D. dissertation., Uni- versity of Arizona, Tucson, 184 pp. (as cited in Callahan and Davis, 1982). CALLAHAN, J. R, AND R. DAVIS. 1982. Repro- ductive tract and evolutionary relationships of 128 THORINGTON ET AL?ARBOREALITY IN TREE SQUIRRELS the Chinese rock squirrel, Sciurotamias davidianus. Journal of Mammalogy, 63:42-47. CORBET, G. B.. AND J. E. HILL. 1992. Mammals of the Indomalayan Region. Oxford University Press, New York, 488 pp. ELLERMAN, J. R. 1940. The families and genera of living rodents. Volume I. Rodents other than Muridae. Trustees of the British Museum (Natural History), London, 689 pp. EMRY, R. J., AND W. W. KORTH. 1996. The Chadronian Squirrel "Sciurus" jeffersoni Douglass, 1901: A new generic name, new material, and its bearing on the early evolu- tion of Sciuridae (Rodentia). Journal of Verte- brate Paleontology, 16:775-780. EMRY, R. J., AND R. W. THORINGTON, JR. 1982. Descriptive and comparative osteology of the oldest fossil squirrel, Protosciurus (Rodentia: Sciuridae). Smithsonian Contributions to Paleobiology, 47:1-35. GROMOV, I. M., D. I. BIBIKOV. N. I. KALABUKOV, AND M. N. MEIER. 1965. Nazemnye belich'i (Marmotinae). Fauna SSSR, 3:1-467. (In Rus- sian). HAFNER, M. S., L. J. BARKLEY, AND J. M. CHUPASKO. 1994. Evolutionary genetics of New World tree squirrels (tribe Sciruini). Journal of Mammalogy, 75:102-109. HILL, W. C. 1936. The penis and its bone in Ceylonese Squirrels; with special reference to its taxonomic importance. Ceylon Journal of Science (B), 20:99-113. . 1940. On the penis of Ratufa macroura melanochra. Ceylon Journal of Science (B), 22:131-134. HIGHT, M. E., M. GOODMAN, AND W. PRYCHODKO. 1974. Immunological studies of the Sciuridae. Systematic Zoology, 23:12-25. HOFFMANN, R. S., C. G. ANDERSON, R. W. THORINGTON, JR., AND L. R. HEANEY. 1993. Family Sciuridae. Pp. 419-465, in Mammal Species of the World: a taxonomic and geo- graphic reference, (D.E. Wilson and D.M. Reeder, eds.), Smithsonian Institution Press, Washington, D.C., 1206 pp. KINGDON, J. 1974. East African Mammals: an at- las of evolution in Africa. Vol. II. Part B (Hares and Rodents). Academic Press, London. LAVOCAT, R. 1973. Les rongeurs du Miocene d'Afrique orientale. Memoires et Travaux de l'lnstitut Montpellier, 1:1-284. MILLER, G. S., Jr. 1901. The subgenus Rhinosciurus of Trouessart. Proceedings of the Biological Society of Washington, 14:22. MILNE-EDWARDS, A. 1867. Description de quelques especes nouvelles d'ecureuils de l'ancien continent. Revue et Magasin de Zoologie, 2:193-232. . 1871. Etudes pour servir a l'histoire de la faune mammalogique de la Chine. Pp. 67- 229, in Recherches pour servir a l'histoire naturelle des mammifers: comprenant des considerations sur la classification de ces animaux, (H.Milne-Edwards and A. Milne- Edwards, 1868-1874). Paris: G, Masson, 394pp. MOORE, J. C. 1959. Relationships among the liv- ing squirrels of the Sciurinae. Bulletin of the American Museum of Natural History, 118:153-206. MOSSMAN, H. W., J. W. LAWHAH, AND J. A. BRA- DLEY. 1932. The male reproductive tract of the Sciuridae. American Journal of Anatomy, 51:89-155. PRASAD, M. R. N. 1954. The male genital tract of two genera of Indian squirrels. Journal of Mammalogy, 35:471-485. QIU, Z. 1991. The Neogene mammalian faunas of Ertemte and Harr Obo in Inner Mongolia (Nei Mongol), China.-8. Sciuridae (Rodentia). Senckenbergiana lethaea 71:223-255. SIDDIQI, M. A. H. 1938. The genito-urinary sys- tem of the Indian ground squirrel (Funambulus palmarum). Proceedings of the Na- tional Academy of Science of India, 1:1-10. VLANEY-LIAUD, M. 1974. Palaeosciurus goti nov. sp., ecureuil terrestre de l'Oligocene moyen du Quercy. Donnees novelles sur l'apparition des Sciurides en Europe. Annales Paleontologique (Vertebrata), Paris, 60:103-122. WAHLERT, J. H. 1985. Cranial foramina of rodents. Pp. 311-332, in Evolutionary relationships among rodents: A multidisciplinary analysis (W. P. Luckett and J-L. Hartenberger, eds.). Plenum Press, New York, 721 pp. . 1991. The Harrymyinae, a new Heteromyid subfamily (Rodentia, Geomorpha), based on cranial and dental morphology of Harrymys Munthe, 1988. American Museum of Natural History Novitates, 3013:1-23. SPECIMENS EXAMINED (all USNM): Ratufa affinis?post-crania: 151757, 198121. Ratufa affinispyrsonota?skulls: 251673, 257716, 257720. Ratufa affinis sandakanensis?skulls: 292564, 292565. 129 VIRGINIA MUSEUM OF NATURAL HISTORY Ratufa bicolor?post-crania: 464512. Ratufa bicolor fretensis?post-crania: 49703. Ratufa bicolor palliata?penis: 546334. Ratufa bicolor smithi ?skulls: 320803-320807. Ratufa indica?skulls: 38010, 355785; penis/bacu- lum: 548661. post-crania: 308415. Sciurotamias davidianus davidianus?skulls: 155110, 155219, 155125-155127, 548431; post-crania: 258505, 258510, 285511, 258516. Sciurotamias davidianus consobrinus -- skulls: 258511-258513, 544436; post-crania: 258506, 258509. Sciurotamias Jorresti?skulls: 255138 (occipital region broken). Sciurus aberti aberti?skulls: 158892, 159332, 167027. Sciurus anomalus?skulls: 37412, 152748, 152749. Sciurus carolinensis carolinensis?skulls: 234368- 234370; post-crania: 256047, 397180. Sciurus carolinensis pennsylvanicus ? post-crania: 297850, 505573, 505575, 548048. Sciurus granatensis?skulls: 318405, 318408, 318409; post-crania: 387805, 540703. Sciurus griseus griseus?skulls: 43041, 242332, 274351. Sciurus niger?skulls: 167740, 177744, 177801; skin: 248132 (only one tarsal pad); postc r a - nia: 347957, 397159. Sciurus vulgaris?skulls: 105106, 105107, 121351; skin: 121351 (juv., has two tarsal pads). Tamias dorsalis dorsalis?skulls: 23695, 23696, 24882, 32090, 32093. Tamias dorsalis merriami?skulls: 22723, 22808, 41773, 41776, 41783. Tamias striatus fisheri?skulls: 62602, 86680, 86686, 86834, 260250; penises/bacula: 2 uncatalogued specimens from Mountain Lake, VA; post-crania: 364947, 396281, 505612, 505613. Tamias striatus griseus?skulls: 17313, 226948. 227426, 229004, 232129; post-crania: 349628. Tamias striatus lysteri?skulls: 30225, 30230, 30239, 43415, 96939; post-crania: 500999, 564115. Tamias striatus ohionensis?skulls: 308622, 308623, 308626, 308636, 308640. Tamias townsendi ochrogenys?skulls: 96110, 96112, 97146, 97337, 97339. Tamias townsendi townsendi?skulls: 24423, 24424, 57124, 69373, 69375. Tamiasciurus douglasii albolimbatus?skulls: 548848, 548849. Tamiasciurus douglasii douglasii?skulls: 166892, 166893, 231804, 231805. Tamiasciurus douglasii mearnsi?skulls: 25169 25171. Tamiasciurus douglasii mollipilosus?skulls: 23992, 24028. Tamiasciurus hudsonicus albieticola?skulls: 55796, 50853, 268992, 294450. Tamiasciurus hudsonicus baileyi?skulls: 66447, 168950, 168951. Tamiasciurus hudsonicus columbiensis?skulls: 202835, 202836. Tamiasciurus hudsonicus dakotensis?skulls: 213689, 213690. Tamiasciurus hudsonicus dixiensis?skulls: 158041, 158042. Tamiasciurus hudsonicus fremonti?skulls: 48209, 48210; post-crania: 564078. Tamiasciurus hudsonicus loquax? post-crania: 397070, 397151, 397152, 505579, 505587, 506645, 551803, 564084. Tamiasciurus hudsonicus hudsonicus?post-crania: 564083. 130