S3 
<J 
2 
2 
o 
U 
v 
w 
_c 
'> 
o u 0. 
4> 
ts> O 
?-i 
C 
rj 
W3 
2 
3 
"O 
c 
o 
X 
? 
CO 
?*? 
O 
? 
o 
c 
E 
u 
= 
C 
O 
X 
C 
w *? u 
O u 
C 
E 
o 
E 
3 
K O </! 
c 
CO 
c 
X 
u 
1 
COMPARATIVE MORPHOLOGY 
AND EVOLUTION OF FROGS OF THE 
NEOTROPICAL GENERA ATELOPUS, 
DENDROPHRYNISCUS, 
MELANOPHRYNISCUS, 
AND OREOPHRYNELLA 
By ROY W. McDIARMID 
BULLETIN OF THE LOS ANGELES COUNTY 
MUSEUM OF NATURAL HISTORY 
SCIENCE: NUMBER 12 
DECEMBER 1, 1971 
PROFESSIONAL PUBLICATIONS OF THE LOS ANGELES COUNTY MUSEUM 
The professional publications of the Los Angeles County Museum of Natural History in- 
clude two series, Contributions and Bulletins. In the past, articles, monographs and catalogs 
in the fields of history and science have appeared under various headings?Contributions, 
Science Series, History Leaflet Series and unnumbered catalogs of exhibitions and collec- 
tions. To simplify and to standardize matters, all professional publications of the History 
and Science Division of the Museum will now be issued at irregular intervals either as 
Contributions, or as Bulletins. The former will contain short, technical papers which may 
be occasionally gathered in volumes, octavo in size. The latter will contain longer, separate 
monographs and catalogs, usually quarto in size, although this will depend on the needs of 
the presentation. Papers in each series are to be numbered consecutively. 
These papers are original articles and studies based on the collections and work of the Mu- 
seum, presenting newly acquired information and understanding in the fields of Anthro- 
pology, Botany, Geology, History, Mineralogy, Paleontology, Technology and Zoology. 
GILES W. MEAD, Director 
Los Angeles County Museum 
of Natural History 
VIRGINIA D. MILLER 
Editor 
All communications concerning science manuscripts, exchange of science publications, and 
the purchase of science publications should be sent to the Managing Editor, Los Angeles 
County Museum of Natural History, 900 Exposition Boulevard, Los Angeles, California 
90007. 
TABLE   OF   CONTENTS 
ABSTRACT     1 
INTRODUCTION  2 
Historical Review  3 
Acknowledgments  5 
METHODS AND MATERIALS  5 
ANALYSIS OF CHARACTERS  6 
Myology  6 
Thigh Musculature  6 
Jaw Musculature  8 
Osteology  9 
Skull  9 
Mandibular Arch  20 
Hyoid Apparatus  20 
Vertebral Column  22 
Appendicular Skeleton  26 
Auditory Apparatus  29 
External Morphology  31 
Habitus  31 
Integument  31 
Hands and Feet  32 
Reproductive Biology  33 
Urogenital System  33 
Breeding Biology  36 
Larvae  37 
Ecology and Distribution  38 
SYSTEMATICS  38 
Generic Definitions  38 
Atelopus  39 
Dendrophryniscus  39 
Melanophryniscus  40 
Oreophrynella  41 
EVOLUTION  43 
Evaluation of Character States  43 
Intergeneric Relationships  43 
Evolutionary Trends  49 
Locomotion  49 
Hearing  50 
Aposematic Coloration  51 
Paedomorphosis  52 
Familial Relationships  52 
Bufonidae  56 
HISTORICAL PERSPECTIVE  57 
RESUMEN  59 
LITERATURE CITED  61 
COMPARATIVE MORPHOLOGY AND EVOLUTION OF 
FROGS OF THE NEOTROPICAL GENERA ATELOPUS, 
DENDROPHRYNISCUS, MELANOPHRYNISCUS, AND 
OREOPHR YNELLA 
By ROY W. MCDIARMID1 
ABSTRACT: Atelopodid frogs are one of the most interesting and diverse 
groups of Neotropical Anura. The systematic status and evolutionary history 
of the genera in this family have been poorly understood. The purpose of this 
study is to analyze available knowledge of the morphology and biology of 
these frogs in order to clarify their evolutionary relationships and history. The 
group includes approximately 50 species in four genera: Atelopus, Dendroph- 
ryniscus, Melanophryniscus, and Oreophrynella. Data indicate that Brachy- 
cephalus, a genus originally included in the family, is not closely related to 
the four genera. Its relationships will be discussed elsewhere. 
Representative specimens of all genera and most species were examined. 
Information concerning myology, osteology, and reproductive morphology 
was gathered. All available literature was reviewed and pertinent information 
was assimilated into this report. 
A detailed description of thigh and jaw musculature and osteology of the 
species is presented. The skulls, pectoral girdles, and hyoid apparati are 
described and illustrated. Components of the auditory apparatus, certain 
aspects of their external morphology, reproductive biology, and ecology are 
described. 
Each genus is defined according to 43 characters. Their geographical 
distributions are stated briefly and their included and referred species are 
listed. Atelopus minutus Melin and Atelopus proboscideus Boulenger are 
placed in the genus Dendrophryniscus. Atelopus rubriventris Vellard is placed 
in the genus Melanophryniscus. 
The four genera are discussed and their character states compared. 
Melanophryniscus has the greatest number of primitive states and the least 
number of advanced states and is probably most similar to the ancestral stock. 
Atelopus also has many primitive states but possesses the greatest number of 
advanced states. Atelopus and Melanophryniscus were derived from the same 
lineage, but Atelopus has undergone a significant radiation at the species level 
and has several advancements not found in the other genera. Dendrophrynis- 
cus was derived from the Melanophryniscus line and exhibits parallel evolu- 
tion in some character states with Atelopus. Oreophrynella has more advanced 
character states than either Melanophryniscus or Dendrophryniscus and only 
one less than Atelopus. However, Oreophrynella has the highest number of 
unique states and the lowest number of primitive states. Oreophrynella ap- 
parently was derived from the ancestral stock at a different time from the 
Melanophryniscus-Atelopus-Dendrophryniscus line and has subsequently be- 
come greatly specialized. 
Major evolutionary trends and morphological character shifts apparently 
are associated with changes in means of locomotion; others are the result of 
differential metamorphosis. Biological modifications associated with the loss 
of the middle ear apparatus and the development of aposematic coloration 
are important. The familial status of the Atelopodidae is discussed and re- 
jected. The genera Atelopus, Dendrophryniscus, Melanophryniscus, and Oreo- 
phrynella are placed in the family Bufonidae which is redefined. 
The ancestral stock from which the four genera were derived probably 
was present in South America before the beginning of the Cenozoic. The an- 
cestral Melanophryniscus-Dendrophryniscus-Atelopus stock probably occur- 
red in a savanna or deciduous forest habitat in southeastern Brazil. Melano- 
phryniscus has retained many of the generalized ancestral characteristics and 
currently is found in the same general type of habitat. Dendrophryniscus was 
derived from the Melanophryniscus stock and has adapted to the wet tropical 
forest of eastern Brazil and the Amazon Basin. Atelopus has adapted to a 
stream-side habitat and moved into montane areas which became available 
with the uplift of the Andes in Late Cretaceous and Early Tertiary. This new 
habitat has been successfully exploited by Atelopus and has been a major fac- 
tor contributing to their specific radiation. Oreophrynella is a very specialized 
frog that was derived from an old bufonid stock and subsequently restricted 
to Mount Roraima, an ancient part of the Guiana Shield. 
'Research Associate, Section of Herpetology, Los Angeles County Museum of 
Natural History; and Department of Biology, University of South Florida, 
Tampa. 
BULLETIN OF LOS ANGELES COUNTY MUSEUM OF NATURAL HISTORY No. 12 
INTRODUCTION 
The South American continent, with its complex 
physiography and vast areas of tropical, subtropi- 
cal, and temperate forest, is a center of diversity for 
anuran amphibians. There are more than 750 
species assigned to ten families and approximately 
105 genera recorded from the continent. One of the 
most interesting of the South American groups is 
the nominal family Atelopodidae which includes 
about 50 species of frogs that are restricted to Cen- 
tral and South America. Most anuran systematists 
consider these frogs to be species of A telopus, Den- 
drophryniscus, Melanophryniscus, Oreophrynella, 
and Brachycephalus. At the outset of this study I 
assumed that these five genera formed a natural 
group (Atelopodidae). As the study progressed it 
became more and more apparent that the mono- 
typic genus Brachycephalus is not closely related 
to the other four genera. Therefore, only A telopus, 
Dendrophryniscus, Melanophryniscus, and Oreo- 
phrynella are considered here. The status of Bra- 
chycephalus is discussed elsewhere (McDiarmid, 
MS). 
Within this group there is considerable biological 
and morphological diversity. Species of A telopus, 
Dendrophryniscus and Oreophrynella are primarily 
tropical forms. One species of A telopus is found at 
sea level in the warm, wet rain forests of the Choco 
in Colombia; another species is adapted to the cold, 
unforested paramos at over 4000 meters elevation 
in the Andes. Species of Melanophryniscus are 
primarily subtropical and south temperate in dis- 
tribution, although one form is known from a 
locality in the Amazon Basin. While species of 
A telopus, Dendrophryniscus, and Melanophrynis- 
cus are wide ranging forms, the two species of 
Oreophrynella are restricted to a single mountain 
massif. Most species are terrestrial as adults, but 
some may be semifossorial, and one species of 
Dendrophryniscus is arboreal and deposits its eggs 
in bromeliads. Adults range from 14 mm in males 
of a small species of Dendrophryniscus to nearly 
60 mm in females of a large species of A telopus. 
Individuals of most species are thin and elongate 
with long, spindly legs; some, however, are short 
and robust. Many species are brightly colored and 
have variable markings of red, green, yellow, 
orange, or black; others may be gray, dull brown, 
or black. Some forms have smooth skin while 
others have rugose and warty skin. Some species 
vocalize; others apparently do not. In some areas 
these frogs are important components of the fauna; 
in other areas they are relatively rare. At certain 
times of the year some species are extremely abun- 
dant and may be the dominant vertebrate in the 
area; at other times, this same species may be nearly 
impossible to find. 
The reproductive biology and life history of only 
four of the more than 40 species of the genus 
A telopus, which has the greatest spectrum of adap- 
tive types, are known. Limited information is avail- 
able on the life histories of two species of Melano- 
phryniscus and two species of Dendrophryniscus, 
but nothing is known concerning the reproductive 
biology of Oreophrynella. 
The generic relationships and systematic status 
of these frogs are poorly understood. Since the first 
species were described in the mid-1800's, they have 
been variously placed in approximately one-half of 
the currently recognized anuran families, and three 
of the four genera have been placed in their own 
families. In one of the classic works on salientian 
phylogeny, Noble (1922) placed the genera, to- 
gether with several genera that now comprise the 
family Dendrobatidae, the firmisternal members of 
the Leptodactylidae, and the Rhinodermatidae, in 
a single family. Most workers, including Noble, 
agree that the atelopodids are related to the Bufoni- 
dae, but no one has demonstrated the nature of 
this relationship. 
It was apparent that this group of frogs pro- 
vided an ideal opportunity for a study of major 
patterns of anuran evolution. This opportunity, 
coupled with my interest in ecology and zoogeogra- 
phy of tropical organisms, was the stimulus for this 
study. The project is designed to answer the follow- 
ing questions: 1) What types of variation, morpho- 
logical or otherwise, are characteristic of each 
genus? 2) What characteristics are most useful in 
defining the genera from the phylogenetic view- 
point? 3) What types of relationship exist among 
the generic categories? 4) What are the evolutionary 
implications of the morphological changes within 
and among the genera? 5) How do the characteris- 
tic features of the genera considered as a unit, com- 
pare with other groups of closely related genera? 
6) What can be said concerning the patterns of 
historical development of the genera? 
This study describes in detail the osteology of 
each genus (based on examination of about 70% 
of the included species) and certain aspects of their 
myology and reproductive morphology. All known 
features of behavior, ecology, and natural history 
are considered. The genera are redefined, and their 
relationships are elucidated together with associated 
morphological and evolutionary trends. Finally, 
this study makes possible the evaluation of the 
familial status of the Atelopodidae. 
1971 MORPHOLOGY AND EVOLUTION OF NEOTROPICAL FROGS 
HISTORICAL REVIEW 
The generic name Atelopus was proposed by 
Dumeril and Bibron in 1841 for the species flaves- 
cens. Two years later Fitzinger (1843) erected the 
family Atelopoda to include Atelopus flavescens 
and Rhinoderma darwinii. 
During the period from 1834 to 1857, several 
other generic names were proposed for species 
related to A. flavescens. Wiegmann (1834) de- 
scribed Phryniscus nigricans from southern Peru. 
Dumeril and Bibron (1841) made reference to 
Phryniscus nigricans from Montevideo. A careful 
comparison of the two descriptions indicates that 
Phryniscus nigricans Wiegmann is not the same as 
Phryniscus nigricans of Dumeril and Bibron. 
Nevertheless, most workers have referred to the 
Montevideo frog as Phryniscus nigricans. 
Wagler (1828) proposed the name Chaunus mar- 
moratus. In 1830 he recognized that Bufo globu- 
losus Spix (1824) was the same as his Chaunus 
marmoratus, and therefore, because of priority, he 
allocated the specific name globulosus to the genus 
Chaunus. Tschudi (1838) ignored the priority of 
Spix's globulosus and recognized Chaunus mar- 
moratus plus a second species formosus. Peters 
(1873) demonstrated that the type of Spix's Bufo 
globulosus was actually a Bufo granulosus. Thus 
Chaunus was established for a species of Bufo and 
became a synonym of Bufo. Tschudi's (1838) refer- 
ence to Chaunus formosus apparently was based on 
specimens in the Paris museum of Phryniscus nigri- 
cans of Dumeril and Bibron. However, as pointed 
out by Boulenger (1894), the name Chaunus for- 
mosus is a nomen nudum. Fitzinger (1843) con- 
sidered Chaunus globulosus a synonym of Bufo 
globulosus. Gunther (1858b), Cope (1867), and 
Boulenger (1882) assigned Chaunus formosus to 
the synonymy of Phryniscus nigricans. 
In 1856 Lichtenstein and von Martens described 
the genus Phrynidium and two species, varium and 
crucigerum, both from Veragoa (= Veragua). Gun- 
ther (1858b) considered Phrynidium a synonym of 
Phryniscus. Cope (1867) maintained Phrynidium 
for the species crucigerum, varium, laevis and 
bibronii and recognized Phryniscus as distinct, in- 
cluding only the species nigricans. Boulenger (1882) 
assigned Phrynidium to the synonymy of Phrynis- 
cus. 
Oscar Schmidt (1857) described the genus Hylae- 
morphus, a name originally assigned to two species 
by Fitzinger, H. dumerilii from "New-Granada, 
Provinz Veragua...," and H. bibronii from "New- 
Granada, unweit Panama...." In the same paper 
Schmidt described the genus Phirix for the species 
pachydermus, based on a specimen from "Western 
von New-Granada, bei Bonaventura " All three 
were illustrated the next year (Schmidt, 1858). 
Gunther (1858b:43) relegated the name Hylae- 
morphus to the synonymy of Phryniscus, ap- 
parently on the basis of specimens from the Vienna 
Museum. In the appendix of the same book Gun- 
ther (1858b: 136) relegated Hylaemorphus Fitzin- 
ger (in Schmidt) and Phirix Schmidt to the syno- 
nymy of Phryniscus. Cope (1867:196) considered 
Phirix a synonym of Phrynidium and listed the 
species bibronii, described as Hylaemorphus bib- 
ronii Schmidt, as Phrynidium bibronii without any 
mention of Hylaemorphus being synonymous with 
Phrynidium. However, on the same page Cope 
listed Hylaemorphus Fitzinger as a synonym of 
Phryniscus Wiegmann. Boulenger (1882) relegated 
Hylaemorphus and Phirix to the synonymy of 
Phryniscus where they have remained. 
Gunther (1858a, b) recognized only two genera 
of these frogs, Atelopus for flavescens, and Phry- 
niscus for nigricans, laevis, cruciger, varius, olfersii 
(referred to Physalaemus by Parker, 1927), and 
bibronii. Apparently Gunther (1858b) was unaware 
of Cornalia's description of Phryniscus ignescens in 
1849. Phryniscus, family Phryniscidae, was placed 
in the section Brachycephalina, while Atelopus, 
family Rhinodermatidae, was placed in the section 
Bufonina. 
Cope (1865) placed Brachycephalus and Ate- 
lopus in the family Engystomidae and had Phry- 
niscus in the Bufonidae; both families were in the 
suborder Bufoniformia. It is not known whether 
Cope's material of Phryniscus nigricans was the 
same as Wiegmann's or the same as that of 
Dumeril and Bibron. However, in 1867 Cope in- 
cluded all three genera in the family Phryniscidae. 
Although he cites Wiegmann as the source of the 
name Phryniscus, Cope's material came from 
Buenos Ayres (sic) and probably was representa- 
tive of the frogs called Phryniscus by Dumeril and 
Bibron. Keferstein (1867, 1868a, b) assigned a 
Costa Rican frog to Atelopus varius rather than 
Phryniscus varius as proposed by Gunther. How- 
ever, Keferstein followed Giinther's proposal in 
that he considered Atelopus to be a member of the 
family Rhinodermatidae. 
Mivart (1869) placed Brachycephalus and Phry- 
niscus in a single family Phryniscidae, a treat- 
ment not unlike Giinther's classification. However, 
Mivart placed Giinther's Rhinodermatidae, which 
included Atelopus flavescens, in the family Engy- 
stomidae and characterized the group as having 
a perfect ear and thus differing from the Phry- 
niscidae. 
Jimenez de la Espada proposed Dendrophrynis- 
BULLETIN OF LOS ANGELES COUNTY MUSEUM OF NATURAL HISTORY No. 12 
cus brevipollicatus and the family Dendrophrynis- 
cidae in 1870. In the description he mentioned its 
similarities to Atelopus, but because of its dilated 
toe tips, he suggested that its relationships were 
with Giinther's group, the Hylaplesina. He illus- 
trated Dendrophryniscus brevipollicatus in 1875. 
In the same publication Jimenez de la Espada 
(1875) discussed three species of Atelopus as 
members of the family Rhinodermatidae. 
Buchholz and Peters (1875) compared Atelopus 
with Nectophryne in their description of Necto- 
phryne afra. 
Brocchi (1878) suggested that the differences 
between Phryniscus and Atelopus were not enough 
to warrant generic distinction. However, he recog- 
nized both Atelopus and Phryniscus in his 1882- 
1883 publication. 
Boulenger (1882) listed Brachycephalus and 
Phryniscus in the family Engystomatidae. Atelopus 
was considered a synonym of Phryniscus; the latter 
contained eleven species. Dendrophryniscus brevi- 
pollicatus and the leptodactylid Batrachophrynus 
brachydactylus were maintained in their own fam- 
ily Dendrophryniscidae. 
Cope (1887) continued to recognize the family 
Phryniscidae and the genus Atelopus. Boettger 
(1892) followed Boulenger and maintained Phry- 
niscus in the family Engystomatidae. Giinther 
(1900), however, referred Atelopus to the family 
Brachycephalidae. 
In 1894 first Philippi and then Boulenger recog- 
nized that Phryniscus nigricans Wiegmann was not 
the same as Phryniscus nigricans Dumeril and 
Bibron and that Phryniscus Wiegmann was based 
on a Bufo. The next available name for the frogs 
referred to Phryniscus by Dumeril and Bibron and 
others was Atelopus. Philippi (1894) selected the 
specific name formosus published by Tschudi in 
1838 (in the combination Chaunus formosus) to 
replace nigricans. However, as previously indi- 
cated, formosus is a nomen nudum. The only name 
available, as pointed out by Boulenger (1894) is 
Phryniscus stelzneri Weyenbergh (1875); the spe- 
cies hence was designated Atelopus stelzneri. 
From about 1900 to the present several addi- 
tional species of Atelopus have been described, 
most in works by Boulenger (1898, 1902), Werner 
(1899), Peracca (1904), Ruthven (1916), Dunn 
(1933), Shreve (1936), Andersson (1945), Tay- 
lor (1952), and Rivero (1963). 
Boulenger (1895a,b) described Oreophrynella 
quelchii from Mount Roraima in British Guiana. 
He considered Oreophrynella to be closely related 
to Atelopus and placed the genus in the family 
Engystomatidae. In 1900 Boulenger described a 
second species of the genus, O. macconnelli, from 
the same area. Rivero (1961) suggested that mac- 
connelli is a subspecies of O. quelchii. 
Gadow (1901) based his classification on the 
presence or absence of a tongue and teeth, the con- 
dition of the shoulder girdle, and the shape of 
the sacral diapophyses and the vertebral centra. 
On this basis, Phryniscus (Gadow apparently did 
not accept the change proposed by Boulenger 
from Phryniscus to Atelopus), Oreophrynella, and 
Brachycephalus were included in the family En- 
gystomatidae. Dendrophryniscus was maintained 
together with Batrachophrynus in the subfamily 
Dendrophryniscinae of the Cystignathidae. One of 
the major characteristics of the latter family was a 
cylindrical sacral diapophysis. In this character 
Gadow followed Jimenez de la Espada's descrip- 
tion of Dendrophryniscus as having a non-dilated 
sacral diapophysis. 
Barbour and Noble (1920) rejected the family 
Dendrophryniscidae on the grounds that Batra- 
chophrynus was related to Telmatobius and not 
to Dendrophryniscus. It was not until 1926 that 
Noble placed Dendrophryniscus in the Brachy- 
cephalidae. 
Nicholls (1916) introduced the structure of the 
vertebral column into anuran classification. He 
considered Atelopus (two species examined) as a 
member of the Engystomatidae but pointed out 
that they retained procoelous vertebrae. 
In 1922, Noble used the name Brachycephalidae 
for frogs having a bufonid type of thigh muscula- 
ture, procoelous centra, arcifero-firmisternal or 
firmisternal pectoral girdle, cylindrical or dilated 
sacral diapophyses, five to eight presacral verte- 
brae, and terminal phalanges that are never claw 
shaped. He placed Sminthillus, Geobatrachus, 
Brachycephalus, Phyllobates, Hyloxalus, Dendrob- 
ates, Atelopus, and Rhinoderma into this family. 
Noble (1922:39) considered Atelopus stelzneri to 
be a Bufo. 
In 1926 Noble reviewed the brachycephalid 
frogs and included Dendrophryniscus brevipollica- 
tus in the family. Noble pointed out that the sacral 
diapophyses are broadly dilated in Dendrophrynis- 
cus. In this same paper he assigned A. stelzneri, 
earlier considered a Bufo, and A. moreirae to the 
genus Dendrophryniscus. He defined Dendrophry- 
niscus as differing from Atelopus only in the par- 
tially arciferal pectoral girdle. 
According to Noble (1926), the Brachycephali- 
dae consisted of three distinct groups of genera 
each of which was independently derived from 
bufonid ancestors. However, because all three arose 
in the same general region and from the same fam- 
1971 MORPHOLOGY AND EVOLUTION OF NEOTROPICAL FROGS 
ily, he considered them a natural group. The first 
group included Atelopus, Dendrophryniscus, Oreo- 
phrynella, and Brachycephalus. The second group 
included Hyloxalus (=Colostethus), Phyllobates, 
and Dendrobates. The third group included Smin- 
thillus, Rhinoderma, and Geobatrachus. 
Davis (1935) questioned Noble's consideration 
of the Brachycephalidae as a natural, even though 
composite, family. Davis elevated each of Noble's 
groups to familial status. He recognized the Atelo- 
podidae ( = Brachycephalidae), the Rhinodermati- 
dae, and the Dendrobatidae. In the same paper he 
placed the Asian Cacophryne borbonica in the fam- 
ily Atelopodidae, primarily on the basis of its fir- 
misternal girdle and its lack of a Bidder's organ. 
However, Cacophryne later was shown to have a 
Bidder's organ (Dubois, 1947; Griffiths, 1954b). 
Griffiths (1954b) removed Cacophryne from the 
Atelopodidae and put it in the Bufonidae. He con- 
sidered the firmisternal condition of the girdle as 
independently derived from that of a Pedostibes- 
like bufonid ancestor. This treatment had been 
considered, but rejected earlier by Davis (1935). 
In 1959 Griffiths restricted the Atelopodidae to 
include two genera, Atelopus and Brachycephalus. 
He placed Oreophrynella and Dendrophryniscus 
(based on D. stelzneri) in the Bufonidae. In this 
same work Griffiths placed the three genera of the 
family Rhinodermatidae into the family Leptodac- 
tylidae and considered the Dendrobatidae a sub- 
family of the Ranidae. In general, Cochran (1961), 
Goin and Goin (1962), and Cochran and Goin 
(1970), have followed Griffiths' classification. 
Baldauf (1959) postulated the derivation of the 
Atelopodidae from a bufonid stock similar to the 
B. valliceps, marinus, debilis, punctatus group or 
possibly from a B. quercicus type. 
Gallardo (1961a, b) proposed the genus Mel- 
anophryniscus to include the species stelzneri, 
moreirae, and tumifrons, thus restricting Dendro- 
phryniscus to a single species, brevipollicatus. 
ACKNOWLEDGMENTS 
My sincere appreciation goes to Jay M. Savage, 
who offered many valuable suggestions and criti- 
cisms throughout this study.1 Several other people, 
including Charles L. Hogue, Basil G. Nafpaktitis, 
John W. Wright, and Chester Hymen, read and 
criticized an earlier draft of this paper. Discus- 
sions with several colleagues, especially W. Ronald 
Heyer, Andrew Starrett, Priscilla H. Starrett, and 
'This report is a revised edition of a doctoral dis- 
sertation completed at the University of Southern 
California. 
David B. Wake, have influenced my thinking and 
greatly added to this report. 
The following people and institutions have made 
material available either through loans or ex- 
change: Villy Aellen, Museum D'Histoire Natu- 
relle, Geneve; Werner C. A. Bokermann, Sao 
Paulo, Brazil; Howard W. Campbell, Florida State 
Museum, Gainesville; Antenor Leitao de Carvalho, 
Museu Nacional, Rio de Janeiro; William E. Duell- 
man, Museum of Natural History, University of 
Kansas; Robert F. Inger and Hymen Marx, Field 
Museum of Natural History, Chicago (FMNH); 
Eugenio Izecksohn, Universidade Federal Rural do 
Rio de Janeiro; Raymond Laurent, Fundacion- 
Instituto Miguel Lillo, San Miguel de Tucuman, 
Argentina; James A. Peters, United States National 
Museum, Washington, D. C; Douglas C. Robin- 
son, Universidad de Costa Rica; Jay M. Savage 
and Philip A. Silverstone, University of Southern 
California; Norman J. Scott, University of Con- 
necticut; Charles F. Walker, Museum of Zoology, 
University of Michigan; Ernest E. Williams and 
Benjamin Shreve, Museum of Comparative Zool- 
ogy, Harvard University; John W. Wright, Los 
Angeles County Museum of Natural History 
(LACM); and Richard G. Zweifel, American 
Museum of Natural History, New York. Valuable 
information concerning the behavior, ecology and 
life history of some species of Atelopus was pro- 
vided by Howard W. Campbell, Martha Crump, 
Philip A. Silverstone, and Jay M. Savage. The 
Department of Biological Sciences and the Allan 
Hancock Foundation at the University of Southern 
California provided library facilities, laboratory 
space, equipment, and supplies during the study. 
Field work in Costa Rica in 1964 was supported 
in part by a research grant from the Organization 
for Tropical Studies and by the National Science 
Foundation in the form of a Summer Fellowship 
for Graduate Teaching Assistants. Finally, my 
wife, Mercedes, has provided constant encourage- 
ment and given unselfishly of her time in read- 
ing and typing several drafts of the manuscript. 
Mrs. Peggy Walsh and Mrs. Jenny Harding of 
the Department of Biology, University of South 
Florida typed the final manuscript. To all of these 
people I extend my sincerest thanks. 
METHODS AND MATERIALS 
Specimens used in the osteological analysis were 
prepared in several ways. Most preparations were 
made from specimens preserved in 10 per cent 
formalin and stored in ethyl alcohol. This material 
was measured, sexed and tagged. The frogs were 
skinned, eviscerated, and then cleared in a 2 to 4 
BULLETIN OF LOS ANGELES COUNTY MUSEUM OF NATURAL HISTORY No. 12 
per cent solution of potassium hydroxide, stained 
with Alizarin Red-S, and stored in glycerine. Some 
specimens were dehydrated and cleaned using der- 
mestid beetles or cleaned by hand and bleached in 
sodium hypochlorate solution. Osteological infor- 
mation from additional species not available for 
direct osteological examination was obtained from 
X-ray radiographs, mostly stereoscopic. The radio- 
graphs were exposed from 20 to 30 seconds at in- 
strument readings of 20 or 25 kilovolts and five DC 
milliamperes. All radiographs are on file at the 
Department of Biological Sciences, University of 
Southern California. 
Osteological material was available for every 
genus. The species included, followed by the num- 
bers of specimens examined, are listed below (N = 
skeletons or cleared and stained specimens, XN = 
X-rays): Atelopus boulengeri 1; A. carrikeri X6; 
A. certus X2; A. chiriquiensis 1; A. cruciger 1, 
X15; A. ebenoides 1, X5; A. elegans 2, X8; A. 
exigua I; A. flavescens 2, X10; A. glyphus 1, XI; 
A. ignescens 12, X25; A. longirostris 1; A. oxy- 
rhynchus I; A. pachydermus I; A. spumarius 1; 
A. senex 2, X15; A. spurrelli 2, X10; A. varius 13, 
XI1; A. walkeri X4; A. zeteki 4; Dendrophrynis- 
cus brevipollicatus 4, X6; D. minutus 2, XI1; Mel- 
anophryniscus moreirae 4, X14; M. rubriventris 2; 
M. stelzneri 3, X9; M. tumifrons 2; Oreophrynella 
quelchii 3, X23. In addition, one or two specimens 
each of four undescribed species of Atelopus were 
cleared and stained and are included in the analy- 
sis. Limited dissections provided some osteological 
data for Dendrophryniscus leucomystax and D. 
proboscideus. 
Skeletal material was also available for the fol- 
lowing genera: Bufo, 8 species; Brachycephalus, 
1 species; Ansonia, 2 species; Cacophryne, 1 spe- 
cies; Crepidophryne, 1 species; Rhinoderma, 1 
species; Dendrobates, 2 species; Leptodactylus, 2 
species. 
In addition to the above material, other speci- 
mens were dissected to study musculature and the 
hyoid apparatus. When necessary, specimens were 
stained 48 hours in a 1:8 solution of borax carmine 
and differentiated in one per cent picric acid alcohol 
solution. All my osteological material was de- 
posited in the collections of the Los Angeles 
County Museum of Natural History. 
Preserved museum specimens of nearly all the 
species of Atelopus were examined. Some informa- 
tion was added from observations of populations 
of Atelopus varius in Costa Rica and several live 
specimens of Brazilian Melanophryniscus moreirae 
in the laboratory. 
My use of the word atelopodid is in reference 
to the genera Atelopus, Dendrophryniscus, Mela- 
nophryniscus, and Oreophrynella and does not 
imply a formal taxonomic category. All references 
to a character or character state of a species or 
a genus are based on individuals that are con- 
sidered typical of that taxon and should be read 
accordingly. 
ANALYSIS OF CHARACTERS 
The majority of anuran systematists agree that 
taxonomic categories above the specific level should 
be based on evolutionary relationships inferred 
from data on the evolutionarily conservative and 
relatively stable characters of internal morphology 
(Brattstrom, 1957), rather than external mor- 
phology, coloration and habitus, supplemented 
with whatever information can be obtained from 
reproductive ethology, general behavior and ecol- 
ogy, genetic compatibility, and biochemistry. Only 
a cursory examination of the major schemes of 
frog classification, including the early ideas of 
Cope (1864, 1865, 1866) and Boulenger (1882), 
the major contributions of Noble (1922, 1931), 
the recent comprehensive work of Griffiths (1963) 
and the phylogenies presented by Inger (1967) 
and Kluge and Farris (1969), is needed to compre- 
hend the importance of careful consideration and 
evaluation of different facets of comparative mor- 
phology. The emphasis on internal morphology, 
especially osteology, is extremely important in 
other vertebrate groups because of the abundance 
of fossil material with which modern forms can be 
compared. With frogs, however, the fossil record 
has added little to our understanding of phylo- 
genetic relationships of modern groups (Hecht, 
1963; Inger, 1967). Blair (1962) and Inger (1967) 
pointed out some of the problems that face the evo- 
lutionary morphologist in dealing with anuran clas- 
sification. In spite of these drawbacks, it has been 
amply demonstrated that detailed study of a par- 
ticular group of related organisms (Ritland, 1955a, 
b; Wake, 1966; Liem, 1970) or a detailed study of 
a specific morphological character (Bhaduri, 1953; 
Orton, 1957; Wake, 1968) or system through a 
wide spectrum of unrelated organisms (Trewavas, 
1933; Dunlap, 1960; Starrett, 1968) adds signifi- 
cantly to our understanding of evolutionary rela- 
tionships at generic and higher levels. 
MYOLOGY 
Thigh Musculature 
Early works by Duges (1835), DeMan (1874), 
Perrin (1893), Gaupp (1896), and Nussbaum 
(1898), though somewhat fragmentary in most 
cases, set the foundation for comparative study of 
1971 MORPHOLOGY AND EVOLUTION OF NEOTROPICAL FROGS 
the pelvic limb musculature of frogs. Noble (1922) 
was the first to apply characteristics of the thigh 
musculature from a comparative aspect in an over- 
all scheme of frog phylogeny. Nobel's scheme has 
been criticized, particularly with reference to his 
use of the arrangement of the distal tendons of the 
thigh (Latsky, 1930; Parker, 1940; Colifax, 1956; 
Griffiths, 1959), but has been followed generally. 
Griffiths (1963:262) argued against defining major 
taxonomic groups on the evidence of the thigh 
complex alone but pointed out the value of these 
characters in determining directions of evolution- 
ary trends within groups whose relationships are 
well defined by other characters. 
A recent comparative study of pelvic myology 
(Dunlap, 1960) emphasized the usefulness of 
characteristics of thigh musculature in evaluating 
relationships among different groups of frogs. I 
have followed Dunlap's terminology in this study, 
although, as pointed out by Noble (1922) and 
Dunlap (1960), the names are largely derived 
from mammalian muscles and in many cases hom- 
ologies are unlikely. 
Tensor fasciae latae.? In most frogs, this muscle 
is wide and thin, originates on the ventro-lateral 
aspect of the ilium and inserts on the dorsal sur- 
face of the cruralis or the latero-dorsal surface of 
the glutaeus magnus. Noble (1922) and Dunlap 
(1960) indicated that there is a general tendency 
for this muscle to be better developed in what they 
considered to be the more primitive frogs. 
The position and origin of the tensor fasciae 
latae follow two basic patterns in the specimens 
studied (Fig. 1). In most species of Atelopus stud- 
ied, the muscle is straplike, much elongated, and 
has its origin on the ventro-lateral anterior surface 
of the ilium below the sacral diapophysis. In A. 
flavescens, spumarius, spurrelli, and an undescribed 
species, this muscle arises just posterior to the point 
of connection between the sacral diapophysis and 
the ilium. In all forms the muscle is covered along 
its anterior half by the dorso-lateral extension of 
the external oblique muscle. The insertion of the 
tensor is on the cruralis; the tendon of insertion 
extends distally to fuse with the aponeurosis of the 
knee. The position of the point of insertion is vari- 
able. In some species the muscle inserts at a point 
between the proximal one-third to one-half of the 
foreleg; in others the insertion may be along the 
distal one-half to one-third of the foreleg. Gen- 
erally, the shorter legged species (e.g., A. ignescens, 
boulengeri, ebenoides, pachydermus) have the in- 
sertion closer to the distal end of the foreleg. 
The second basic pattern is found in species of 
Dendrophryniscus, Melanophryniscus, and Oreo- 
TENSOR FASCIAE 
LATAE 
CRURALIS 
B 
FIGURE 1. Diagrammatic representations of the condi- 
tions of the tensor fasciae latae muscle in ventral view. 
Condition A characteristic of species of Dendrophry- 
niscus, Melanophryniscus, Oreophrynella and Bufo. 
Condition B characteristic of some species of Atelopus. 
Condition C characteristic of most species of Atelopus. 
8 BULLETIN OF LOS ANGELES COUNTY MUSEUM OF NATURAL HISTORY No. 12 
phrynella. In specimens of Melanophryniscus the 
muscle is short and broad and originates on the 
ventral surface of the ilium about one-third the 
distance from the posterior end. The insertion lies 
along the lateral surface of the distal half of the 
cruralis. There is, however, some variation among 
the species in the width of the tensor muscle at the 
point of insertion. 
The tensor muscle of Oreophrynella quelchii is 
similar to that found in Melanophryniscus. The 
origin is about midway along the ilium, slightly 
more anterior than in Melanophryniscus; the inser- 
tion is about midway along the cruralis muscle. 
In specimens of Dendrophryniscus the origin is 
on the ventro-lateral surface of the ilium posterior 
to the midpoint; the insertion is on the cruralis 
muscle medial to the midpoint. As with some mem- 
bers of the genus Atelopus, the longer legged frogs 
of the genus Dendrophryniscus have the insertion 
nearer the body than do the shorter legged frogs of 
the genera Melanophryniscus and Oreophrynella. 
Sartorius, semitendinosus, gracilis major and 
gracilis minor.?These muscles were considered to 
be of great taxonomic importance (Noble, 1922) 
not so much because of their individual character- 
istics, but rather because of their relationship to 
each other, especially with reference to the dispo- 
sition of their distal tendons. 
The sartorius is a superficial muscle lying across 
the ventral surface of the thigh. It is thin and nar- 
row in species of Atelopus and Dendrophryniscus, 
thin, but broader in Oreophrynella, and somewhat 
thicker and broader in Melanophryniscus. It ap- 
pears that the sartorius in the longer legged frogs 
(Atelopus, Dendrophryniscus) is thinner and nar- 
rower than in the shorted legged forms (Melano- 
phryniscus, Oreophrynella). 
Primitively, the sartorius and the semitendinosus 
in most frogs are derived from a common muscle 
(Noble, 1922; Dunlap, 1960). The semitendinosus 
is ventrally situated and lies between the gracilis 
major and the adductor magnus, where it is visible 
superficially as a thin slip along the distal one-half 
to two-thirds of its length in all species of Atelopus, 
Melanophryniscus and Oreophrynella. In Dendro- 
phryniscus the semitendinosus muscle lies slightly 
deeper between the gracilis major and adductor 
magnus muscles and is not visible superficially. 
The distal end of the semitendinosus forms a 
slender tendon which inserts on the aponeurosis of 
the knee. This tendon is proportionally longer in 
Dendrophryniscus brevipollicatus than it is in D. 
minutus and species of Atelopus, Melanophrynis- 
cus, and Oreophrynella. 
The gracilis major and minor muscles are gen- 
erally the same in all species studied. The gracilis 
major has its origin from the postero-ventral edge 
of the ischial rim of the pelvis and its insertions by 
one tendon on the aponeurosis of the knee and by 
a second tendon on the proximal part of the tibia. 
The gracilis minor has its origin in two heads, one 
from the ischiac region of the pelvis and the other 
from the skin. 
The sartorius inserts, in part, onto the insertion 
tendon of the semitendinosus via a distal tendon. 
In all specimens studied the insertion tendon of the 
semitendinosus passes ventrally and superficially 
with respect to the insertion tendon of the gracilis 
major. 
Adductor longus and pectineus.?That the adduc- 
tor longus and pectineus are derivatives of a single 
muscle mass has been demonstrated unequivocally 
by Noble (1922) and Dunlap (1960, 1966). The 
adductor longus arises on the iliac portion of the 
ventral pelvic rim and inserts on the aponeurosis 
of the knee. It is visible superficially and lies be- 
tween the cruralis and the adductor magnus. The 
pectineus is a short, fan-shaped muscle which in- 
serts along the ventral surface of the femur shaft. 
It lies beneath the adductor magnus posterior to the 
cruralis. 
Inger (1967, Table 1) reported the presence of 
the adductor longus muscle in the Atelopodidae. 
However, the adductor longus muscle is absent 
from all specimens of Atelopus, Dendrophrynis- 
cus, Melanophryniscus, and Oreophrynella which 
I examined. 
The pectineus is present in all specimens exam- 
ined. There is considerable variation in the distal 
extent of the insertion of this muscle. In the long- 
legged species of Atelopus (e.g., cruciger, longi- 
rostris, oxyrhynchus, varius) the distal end inserts 
along the proximal one-third to one-half of the 
femoral shaft. The same condition is found in the 
long-legged species of Drendrophryniscus. How- 
ever, the short-legged species of Atelopus (ignes- 
cens, pachydermus, etc.) and all species of Melano- 
phryniscus and Oreophrynella have the insertion 
of the distal end of the pectineus between the distal 
one-half to three-fourths of the femoral shaft. No 
attempt was made to categorize this variation other 
than the apparent correlation between the distal 
insertion and relative length of the femur. 
Jaw Musculature 
The first reference to the value of jaw muscula- 
ture in anuran systematics was made by Griffiths 
(1954b, 1959) who divided the phaneroglossid 
anurans into three groups according to the origin 
of the depressor mandibulae muscle and the angle 
1971 MORPHOLOGY AND EVOLUTION OF NEOTROPICAL FROGS 
which the squamosal-quadrate complex forms with 
the mandibular arm. Starrett (1960) and Limeses 
(1965) demonstrated the usefulness of jaw mus- 
culature in defining genera and evaluating generic 
relationships. Recently Starrett (1968) discussed 
and surveyed the variation in jaw musculature in 
frogs. The terminology and character state assign- 
ments used here generally follow her work. 
Depressor mandibulae.?-The three groups that 
Griffiths (1954b) recorded using the disposition 
of the depressor mandibulae are: (1) originating 
from the edge of the posterior squamosal arm and 
the lateral edge of the otic arm; (2) originating 
from the squamosal, the annulus tympanicus and 
the dorsal fascia; (3) originating from the dorsal 
fascia only. Starrett (1968) found much more vari- 
ation in the origins of the depressor mandibulae 
than mentioned by Griffiths. Griffiths (1954b:44) 
stated that the depressor mandibulae has its origin 
on the supra-otic (Bufonidae) and on the latero- 
otic (Atelopodidae) of the squamosal arm. Baldauf 
(1959:536) demonstrated that the depressor man- 
dibulae muscle may arise from the crista parotica, 
from the squamosal or from both in species of 
Bufo. On this evidence Baldauf questioned the 
importance of the origin of the muscle in clarifying 
bufonid phylogeny. 
In all species examined of the four genera, the 
depressor mandibulae muscle is typical of Griffiths' 
group one; the muscle has its origin on the squa- 
mosal arm and the lateral portion of the prootic, 
or the crista parotica, inserts on the posterior por- 
tion of the mandible, and is referred to as condi- 
tion SQ. Atelopus flavescens, spumarius, and an 
undescribed species and Dendrophryniscus minu- 
tus have variations of the basic atelopodid pattern. 
In these three species of Atelopus a few fibers of 
the depressor mandibulae arise on the postero- 
ventral edge of the annulus tympanicus; this con- 
dition is referred to as SQat. In D. minutus the 
depressor mandibulae arises on the anterior pro- 
jection as well as the posterior arm of the squamo- 
sal and on the lateral portion of the prootic; this 
modified condition is referred to as SQm. 
A second characteristic which Griffiths consid- 
ered in conjunction with the condition of the de- 
pressor mandibulae muscle was the angle formed 
by the connection of the squamosal quadrate com- 
plex and the mandible. He indicated that all forms 
with a depressor mandibulae muscle in condition 
(1), also have a squamosal angle ranging from 55 
to 70 degrees. All other frogs that have a different 
depressor mandibulae condition have a squamosal 
angle lower than 50 degrees. He considered the 
combination of the group (1) depressor mandi- 
bulae muscle and the high value for the squamosal 
angle as the primitive condition. Starrett (1968) 
showed that there is much greater variation in the 
squamosal angle than indicated by Griffiths and 
rejected the use of this character in discussing 
relationships because of the wide range of overlap 
between most higher groups of frogs. 
Adductor mandibulae complex.?Starrett (1960, 
1968) demonstrated that there are basically three 
different relationships between the adductor mandi- 
bulae muscles and the mandibular branch of the 
trigeminal nerve (Fig. 2). Some groups of frogs 
have a single adductor mandibulae muscle, others 
have two adductor mandibulae muscles, and a few 
species have none. When present the adductor 
muscles arise from the anterior arm of the squamo- 
sal and insert on the lateral part of the mandible 
anterior to its articulation with the squamosal. If 
two muscles are present, the mandibular branch of 
the trigeminal nerve passes medial to the adductor 
mandibulae externus superficialis and lateral to the 
adductor mandibulae posterior subexternus, that is, 
between the two muscles; this condition is referred 
to as S and E. In some frogs with one muscle, the 
mandibular branch of the trigeminal nerve lies 
across the lateral face of the muscle, so that only 
the subexternus is present; this condition is desig- 
nated S. In other frogs with one muscle, the nerve 
passes medial to the muscle so that only the exter- 
nus is present; this condition is designated E. 
All species of Atelopus, Melanophryniscus, and 
Dendrophryniscus have the S condition. In Oreo- 
phrynella quelchii the adductor mandibulae pos- 
terior subexternus is absent, and the mandibular 
branch of the trigeminal nerve passes behind the 
adductor mandibulae externus superficialis, the E 
condition. 
OSTEOLOGY 
Skull 
While there are several papers which treat the 
anuran skull in general, no author has considered 
the species examined in this study from a compara- 
tive aspect. In the classic study by Parker (1882) 
the skulls of three species of Atelopus were de- 
scribed and illustrated. In general Parker's work is 
adequate in view of the techniques he utilized in 
preparing specimens. Unfortunately some of his 
specimens were misidentified. Parker's (1882:233 
and Plate 41, figs. 1-5) description and illustrations 
of Atelopus cruciger are based on a male specimen 
from the "Interior of Brazils." Yet Atelopus cruci- 
ger occurs only in northern South America. The 
Catalogue of the Batrachia in the British Museum 
10 BULLETIN OF LOS ANGELES COUNTY MUSEUM OF NATURAL HISTORY No. 12 
(Boulenger 1882: 154) lists no specimens of A. 
cruciger from Brazil. However, four specimens of 
A. flavescens from the "Interior of Brazil" are 
listed. A comparison of material of both A. cruci- 
ger and A. flavescens with Parker's illustrations 
a.m. p. I 
a. m.e.s 
FIGURE 2. Representations of the three different ar- 
rangements of the adductor mandibulae muscles. 
Upper?S and E condition typical of Bufo; Middle?S 
condition typical of Atelopus, Dendrophryniscus, 
Melanophryniscus; Lower?E condition typical of 
Oreophrynella. a. m. e. s.?adductor mandibulae ex- 
ternus superficialis; a. m. p. 1.?adductor mandibulae 
posterior lateralis; a. m. p. sub.?depressor mandi- 
bulae; NTm?mandibular branch of trigeminal nerve; 
sq.?squamosal; tymp.?annulus tympanicus. 
of A. cruciger leaves no doubt that he was deal- 
ing with a specimen of A. flavescens and not A. 
cruciger. 
The only other work dealing with the cranial 
osteology pertinent to this study is that by Baden- 
horst (1945). She worked with the cranial os- 
teology of Melanophryniscus moreirae, relying 
primarily on sectioning techniques. Laurent (1942) 
discussed certain aspects of the osteology of Atelo- 
pus varius. Griffiths (1954b) presented a diagram 
of the skull of A telopus ignescens in his discussion 
of the otic element in Amphibia. Several other 
papers dealing with cranial osteology of bufonid 
frogs, including those by Parker (1876), Rama- 
swami (1937), Sedra (1949), Sanders (1953), 
Baldauf (1959), and Tihen (1962a), were con- 
sulted and found useful in interpreting the rela- 
tionships between the atelopodid genera and genera 
referred to the Bufonidae. 
The cranium can be divided conveniently into 
two basic components: a dermatocranium, that 
part of the skull formed from elements of dermal 
or membrane origin; and a chondrocranium, that 
part formed from elements of endochondral origin. 
Nasal.? The nasal bones in these frogs are sub- 
ject to considerable interspecific variation. In spe- 
cies of Atelopus (Fig. 3) the nasals are broadly 
triangular in shape and separated on the midline. 
Dorsally their anterior borders form the antero- 
lateral edge of the skull. In a few species (e.g., A. 
longirostris) the nasals are nearly in contact an- 
teriorly; in most they are separate. The nasals 
diverge posteriorly on the midline and pass later- 
ally to form an elongate maxillary process. In some 
species this postero-medial divergence is gradual; 
in others it forms an obtuse angle just anterior to 
the prefrontals before passing latero-ventrally into 
the maxillary process. On the lateral edge anterior 
to the maxillary process there usually is a septo- 
maxillary process that forms the postero-lateral 
structure of and gives rigid support to the nasal 
capsule. The nasals are convex and usually free 
from the sphenethmoid complex. There is a ten- 
dency in some individuals of a few species for the 
postero-medial edges of the nasal bones to fuse 
with the underlying chondocranium. 
In Dendrophryniscus (Fig. 4) the nasal bones 
are similar in most characteristics to those de- 
scribed for Atelopus. They differ in that they are 
strongly convex dorsally, nearly in contact ante- 
riorly and do not diverge posteriorly along the 
median as sharply as in most species of Atelopus. 
In large adults of Melanophryniscus the nasals 
are fused along the midline (Fig. 5) and, with the 
exception of M.  rubriventris and M.  tumifrons, 
1971 MORPHOLOGY AND EVOLUTION OF NEOTROPICAL FROGS 11 
NASAL CARTILAGE 
SPHENETHMOID 
NASAL 
MAXILLA 
PALATINE 
FRONTOPARIETAL 
ORBIT 
PTERYGOID 
OCCIPITAL GROOVE 
SQUAMOSAL 
PROOTIC 
FORAMEN MAGNUM 
OCCIPITAL CONDYLE 
NASAL CARTILAGE 
PREMAXILLA 
VOMER 
SPHENETHMOID 
PALATINE 
MAXILLA 
PTERYGOID 
ORBITOSPHENOID 
OPTIC FORAMEN 
QUADRATOJUGAL 
OCULOMOTOR FORAMEN 
TRIGEMINAL FORAMEN 
PARASPHENOID 
AUDITORY CAPSULE OPERCULUM 
PROOTIC 
OCCIPITAL CONDYLE 
FIGURE 3.  Dorsal and ventral views of the skull of Atelopus varius, LACM 64437, $. Line 
equals 5 mm. 
12 BULLETIN OF LOS ANGELES COUNTY MUSEUM OF NATURAL HISTORY No. 12 
FIGURE 4. Dorsal and ventral views of the skull of 
Dendrophryniscus brevipollicatus, LACM 64438, $. 
Line equals 5 mm. 
also with the frontoparietals. Smaller individuals 
of M. moreirae (smaller individuals of other spe- 
cies not available) that exhibit less cranial ossifi- 
cation have the nasals separated, suggesting an 
ontogenetic trend towards fusion. In addition to 
the medial fusion, the nasal bones are more convex 
than in Atelopus. In cross section the nasal regions 
look like two adjacent hills (the nasal bones) that 
are separated by a shallow valley (the suture zone). 
Dorsal ornamentation is characteristic of these 
species. Dorsally the nasals are fused with the 
sphenethmoid complex but laterally in the maxil- 
lary area the two processes are separate. There is 
no septomaxillary process. 
In Oreophrynella (Fig. 6) the nasals are gener- 
ally reduced, widely separated and free from the 
underlying sphenethmoid complex and the fronto- 
parietals. The maxillary process is present but 
reduced in length. The overall shape is quadrangu- 
lar; there is a definite septomaxillary process. The 
typical convex shape of the nasals is suggested, but 
there is no indication of any dorsal ornamentation. 
Frontoparietal ?The frontoparietal bone covers 
most of the dorsal surface of the anuran skull. In 
Atelopus it extends from the anterior border of the 
orbit posteriorly to a point just in front of the fora- 
men magnum and laterally above the auditory 
capsules (Fig. 3). In some specimens the lateral 
and posterior margins of the frontoparietal are 
fused with the underlying chondrocranium. The 
frontoparietal folds ventrally to form the mid- 
dorsal wall of the orbit. Each occipital artery lies 
in a canal near the point at which the frontoparie- 
tals pass laterally over the otic capsule on each side. 
In all species of Atelopus examined the occipital 
grooves are roofed over with bone for at least half 
of their length. In some, the groove is completely 
covered, opening only near the anterior end. In 
one specimen of A. flavescens the groove on one 
side is open along its entire length. Both grooves 
are open their entire lengths in a specimen of an 
undescribed but closely related species from French 
Guiana. These grooves converge slightly and open 
along the posterior edge of the frontoparietal. Me- 
dially the occipital groove delimits the otic portion 
of the frontoparietal. In most species of Atelopus 
a lateral flange extends out over the orbit at the 
angle between the main body of the frontoparietal 
and its otic portion. 
Griffiths (1954a) demonstrated that ossification 
of the frontoparietal element originates in two 
centers in some amphibians and from a single 
center in others. Notably, most bufonids have a 
single center. Griffiths argued that single centers 
of frontoparietal ossification probably have re- 
1971 MORPHOLOGY AND EVOLUTION OF NEOTROPICAL FROGS 13 
suited from the fusion of the center of ossification 
of the frontal and the center of ossification of the 
parietal rather than from the loss of either. 
Three Atelopus ignescens, measuring 12, 18, and 
20 mm, were used to determine morphogenesis of 
the frontoparietal bone. The frontoparietal bone is 
derived from a single center of ossification on 
either side of the midline. The two frontoparietal 
elements have already assumed their general out- 
line in the 12 mm specimen. Between the 12 and 
20 mm stages, the elements complete their growth 
anteriorly. The otic portion, not apparent at the 
earliest stage, takes on its general appearance lat- 
eral to the occipital groove. Beginning posteriorly, 
the two elements ossify along the midline. The two 
halves of the frontoparietal have all degrees of 
medial fusion in individuals of the same or different 
species. In adults of a single species the fronto- 
parietal elements may be completely separate, or 
they may be fused along their entire length. 
In Dendrophryniscus the frontoparietal is essen- 
tially the same shape as it is in Atelopus. Only in 
the largest male examined is there any fusion pos- 
teriorly between the frontoparietal elements; in all 
other specimens they are completely separate. The 
occipital groove is open along its entire length 
across the frontoparietal (Fig. 4). In the largest 
individuals the otic portion of the frontoparietal 
extends laterally nearly to the dorsal section of the 
squamosal arm. Dendrophryniscus minutus and 
most specimens of D. brevipollicatus have some 
fusion of the frontoparietal to the prootic. In one 
large specimen of D. brevipollicatus, fusion is com- 
plete. The medial part of each frontoparietal is 
noticeably convex above the postero-lateral por- 
tions of the frontal fontanelle. 
The frontoparietal of Melanophryniscus is simi- 
lar in shape to that of Atelopus. There is always 
medial fusion of the frontoparietal posterior to the 
transverse tectum. Anterior to the tectum the fron- 
toparietal may form a medial V, exposing part of 
the frontal fontanelle (Fig. 5), or it may be fused 
completely. In the large adults of M. moreirae and 
stelzneri, the frontoparietal and the nasal bones are 
fused. The occipital groove is open except for a 
very short distance posteriorly. Some individuals, 
especially of M. tumifrons, have considerable 
sculpturing and secondary dermal ornamentation 
on the frontoparietal. In these instances the occipi- 
tal groove usually remains open although it some- 
times is set deeply into the surface of the bone. 
The otic portion of the frontoparietal does not fuse 
with the prootic lateral to the occipital groove. 
The frontoparietal of Oreophrynella quelchii 
(Fig. 6) is unique among the frogs studied. The 
triangular frontal fontanelle and the two parietal 
fontanelles are exposed in the adults. The fronto- 
parietal bone extends laterally on each side of the 
frontal fontanelle to the anterior level of the orbit. 
The anterior edges are longest laterally; medially 
they converge towards the anterior portion of the 
fontanelle. Although the frontoparietal is fused 
across the transverse tectum in one specimen and 
separate in another, in both it surrounds the 
parietal fontanelles on all sides and above the 
FIGURE 5. Dorsal and ventral views of the skull of 
Melanophryniscus tumifrons, LACM 64439, $. Line 
equals 5 mm. 
14 BULLETIN OF LOS ANGELES COUNTY MUSEUM OF NATURAL HISTORY No. 12 
FIGURE 6. Dorsal and ventral views of the skull of 
Oreophrynella quelchii, LACM 64442,9 . Line equals 
5 mm. 
medial tectum that separates the two fontanelles. 
The occipital groove is open along its entire length 
although there is a tendency for calcified cartilage 
to roof over the groove at its posterior end. Lateral 
to the groove the otic portions of the frontoparietal 
apparently fuse with the prootic. Only the anterior 
and medial portions of the frontoparietal are dis- 
tinct from the underlying chondrocranium. 
Premaxilla.?The premaxilla is essentially the 
same in all specimens examined. It consists of a 
basal platform and a dorsal or nasal process. Atelo- 
pus, Dendrophryniscus, Melanophryniscus, and 
Oreophrynella have a premaxillary platform that 
is rectangular in shape. The anterior edge is nearly 
twice as long or longer than the medial edge. There 
is always a well-defined extension into a palatal 
process along the medial edge; there is no well- 
developed lateral process. There is a tendency in 
Melanophryniscus for the palatal process to termi- 
nate in a crescent, the ends of which point to the 
medial line. The palatal processes are longest in 
Oreophrynella and shortest in certain species of 
Atelopus. The palatal processes usually turn dor- 
sally. The posterior edge of the premaxilla has a 
wavy appearance, the arched part of the wave lying 
between the lateral and medial edges. Anterior to 
the dorsal process, the platform turns ventrally to 
form the leading edge of the upper jaw. 
The dorsal processes in A telopus vary consider- 
ably in length and shape. However, in all species 
the dorsal processes diverge laterally and in profile 
are usually directed slightly anteriorly. In some 
instances this contributes to the projecting shape 
of the snout; in others the snout profile is deter- 
mined more by the anterior projection of the nasal 
cartilage or the sphenethmoid complex. The dorsal 
processes are approximately the same in Dendro- 
phryniscus except that when viewed from the front, 
they are nearly straight up and down rather than 
laterally directed. Also, in profile the dorsal proc- 
esses project more anteriorly in Dendrophryniscus 
than in Atelopus. 
In Melanophryniscus the dorsal processes are 
straight and broad from the front and directed 
slightly anteriorly in profile. In contrast to Atelopus 
and Dendrophryniscus the premaxilla contributes 
strongly to the nearly vertical profile of Melano- 
phryniscus. 
The dorsal process in Oreophrynella is generally 
the same as that in the other three genera without 
the strong lateral divergence. However, in pro- 
file the dorsal process is directed posteriorly and 
gives the snout a rounded profile. 
Septomaxilla.?-The small septomaxilla lies ven- 
tral to the antero-lateral margin of the nasal, usu- 
ally just above the point of articulation between the 
maxilla and premaxilla. Parker (1882:240) stated 
that there are no septomaxillae in three species of 
Atelopus that he examined. My examination of 
these three species, as well as several other species 
of Atelopus, indicates that Parker was incorrect. 
All species examined of Atelopus have a similar 
shaped septomaxilla. The lateral edge is broad, and 
1971 MORPHOLOGY AND EVOLUTION OF NEOTROPICAL FROGS 15 
usually there is a moderately sized, anteriorly di- 
rected nasal process. The bone turns medially and 
then dorsally forming a U-shaped trough. The me- 
dial end is directed posteriorly. There is some vari- 
ation in the shape and presence of a foramen in the 
septomaxilla; these characters were not evaluated 
at the specific level. A few specimens of some 
species of Atelopus have an osseous connection 
between the septomaxilla and the maxilla and be- 
tween the septomaxilla and the sphenethmoid. 
The lateral part of the septomaxilla in Melano- 
phryniscus is considerably reduced compared to 
that of Atelopus. There is an elongate, anteriorly 
directed nasal process. The trough of the bone is 
broad, as is the medial edge. The medial end is 
directed posteriorly. 
The septomaxilla of Dendrophryniscus is 
nearly the same as in Melanophryniscus. The only 
difference is the broader lateral edge in Dendro- 
phryniscus. The nasal process is elongate in D. 
brevipollicatus and reduced in D. minutus. The 
trough and medial portions are broad in both. In 
several respects it is nearly intermediate between 
Atelopus and Melanophryniscus. 
The septomaxilla of Oreophrynella is very re- 
duced. The nasal process is nearly vertical in a 
lateral aspect. The trough is closed in front and 
more scooped. The medial process is also reduced 
and lacks the posterior flare characteristic of other 
genera. 
Maxilla.? The maxilla or upper jaw bone is an 
elongate structure lying between the premaxilla 
and the quadratojugal lateral to the pterygoid. The 
maxilla is edentulous and forms the ventro-lateral 
margin of the skull. 
The maxilla of Atelopus is deepest anteriorly at 
a point nearest the nasal and palatine bones. Usu- 
ally, but not always, there is a nasal process. The 
anterior tip nearest the premaxilla is usually 
pointed; posteriorly the maxilla gradually is de- 
creased in width to a rounded point at the quad- 
ratojugal. The ventral edge is nearly straight along 
its entire length. In most species, the ventral edge 
near the premaxilla is raised at about a 45? angle 
and pointed. The dorsal margin is raised gradually 
from its posterior limit to near the nasal bone where 
it may be raised abruptly to form a nasal process. 
Anterior to this process the dorsal edge is rounded 
and sloped gradually or abruptly down to form the 
anterior point. In some instances, this anterior 
dorsal edge is not rounded but rather is irregular 
in outline forming several secondary processes di- 
rected towards the anterior lateral margin of the 
nasal bone. A horizontal maxillary shelf extends 
the entire length of the maxilla about half way 
down the medial side. Posteriorly this shaft sup- 
ports the antero-ventral edge of the pterygoid. The 
shelf is widest anteriorly and is directed slightly 
dorsalad along its free edge. This forms a V-shaped 
trough beneath the nasal capsule. 
The maxilla in Melanophryniscus is similar to 
that in Atelopus. The anterior end is rounded to 
nearly squared; the posterior end often widens at 
the pterygoid rather than narrowing gradually as 
is characteristic of Atelopus. The maxillary shelf is 
prominent and forms the characteristic V-shaped 
trough. 
In Dendrophryniscus the maxilla rises gradually 
to the nasal process, which is always well developed 
and elongate. Posteriorly, the maxilla is pointed, 
and anteriorly it is rounded to blunt. The maxillary 
shelf is well developed and slightly wider anteriorly. 
The maxilla of Oreophrynella is much wider and 
not nearly as elongate as it is in the three previous 
genera. The width at the pterygoid is almost equal 
to the nasal portion. There is a moderate to poorly 
developed nasal process. The anterior end is nearly 
square with a slightly pointed projection directed 
dorsally towards the septomaxilla. The posterior 
part of the maxillary is pointed along the ventral 
edge and rises rapidly at about a 45? angle to the 
pterygoid portion and thence nearly straight across 
to meet the nasal portion. The maxillary shelf is 
gradually reduced in width posteriorly. 
Palatine? The palatine bones are present in all 
species of Atelopus that were examined. They lie 
on the ventral aspect of the antorbital ridge. Gen- 
erally they are subarcuate; some are nearly straight. 
The maxillary half of the palatine is the widest, 
the medial end is usually drawn to a sharp point. 
In A. boulengeri the medial end of the palatine is 
widest. No attempt was made to derive specific re- 
lationships based on palatine shape. 
The palatines are greatly reduced in one speci- 
men of D. minutus and absent in a second specimen. 
If present, they have extensive lateral reduction 
and appear as slivers of bone. The palatines of D. 
brevipollicatus are slightly shorter than those of 
Atelopus; their anterior edge often is fused to the 
underlying chondrocranium and large nasal process 
of the maxilla. There is secondary deposition of 
calcified cartilage in this general region. Apparently 
this gives added support to the antero-lateral part 
of the skull at the nasal-sphenethmoid-maxillary 
connection. 
Palatine bones are present in Melanophryniscus 
tumifrons but lacking in moreirae, stelzneri, and 
rubriventris. When present the bones are partially 
fused to the underlying chondrocranium. The pos- 
sibility exists that the palatines are actually present 
16 BULLETIN OF LOS ANGELES COUNTY MUSEUM OF NATURAL HISTORY No. 12 
but fused to the orbitosphenoid in the latter three 
species. However, the general appearance of the 
antorbital region suggests that the palatine has been 
lost rather than fused. Badenhorst (1945) reported 
that palatines were absent from her sectioned mate- 
rial of Melanophryniscus moreirae. 
Oreophrynella possesses a relatively large pala- 
tine. In ventral view it extends from the antero- 
medial edge of the orbit, laterally and slightly 
posteriorly to the maxilla. There is a connection 
between the palatine and the maxilla on the inner 
surface of the latter above the maxillary shelf. 
Vomer.?The. vomers (prevomers) of Atelopus 
are widely separated medially and, like the pre- 
maxilla and maxilla, are edentulous. Interspecific 
variation in shape of the bone was noted, but no 
attempt was made to group the species by vomerine 
shape. In general the vomer is crescent shaped and 
possesses a medial wing (Fig. 3). The bone is situ- 
ated on the lateral edge of the sphenethmoid com- 
plex, anterior and medial to the opening of the 
internal nares, the choanae. The wings of the vomer 
are directed slightly ventrally to follow the dorsal 
curvature of the oral cavity formed by the sphen- 
ethmoid complex and the maxilla. The anterior 
wing is usually the longest and best developed, and 
is directed toward the anterior tip of the maxilla. 
The medial and posterior wings of the vomer 
form the anterior and medial margins of the 
choanae. The medial wing is pointed and extends 
towards the nasal portion of the maxilla. The pos- 
terior process is often more rounded than the 
medial process and directed parallel to the mid- 
line. The vomer of A. boulengeri is blade-shaped, 
broadest posteriorly and lacks a medial wing. 
Although the vomers of Dendrophryniscus also 
are crescent-shaped and in the same position as 
those of Atelopus, there are some differences. In 
D. brevipollicatus these bones are much larger 
(Fig. 4). The anterior wing extends forward to 
connect with the maxilla laterally and the pre- 
maxilla anteriorly at the point of juncture of the 
latter two. The medial wing is more elongate than 
in Atelopus and extends a greater distance towards 
the nasal portion of the maxilla. In some individ- 
uals the medial wing, together with the posterior 
wing, encloses between one-half and two-thirds of 
the choanae. In D. minutus the vomers are smaller 
than they are in brevipollicatus but the same size 
as in Atelopus. The anterior wing also resembles 
that of Atelopus, while the medial wing is reduced 
and the posterior wing sharply pointed. 
A slightly different vomer is characteristic of 
Melanophryniscus (Fig. 5). The posterior wing 
is greatly reduced and sometimes absent. In con- 
trast the anterior wing is usually quite extensive, 
reminiscent of the condition of the anterior wing 
obtained in some species of Atelopus and in Den- 
drophryniscus brevipollicatus. The extensive ante- 
rior wing is characteristic of M. tumifrons, M. 
stelzneri, and M. rubriventris. The vomers in M. 
tumifrons and M. stelzneri are fused to the under- 
lying sphenethmoid complex. The vomer in M. 
moreirae is much wider and lacks the elongate 
anterior wing. The medial and anterior wings are 
of about equal length and form a crescent between 
them. The vomers are distinct from the underlying 
chondrocranium in M. moreirae and M. rubriven- 
tris. The fusion of the vomer to the chondrocra- 
nium in M. stelzneri and M. tumifrons correlates 
with the fusion of other skull elements. 
The vomers of Oreophrynella quelchii are tri- 
angular and small (Fig. 6). They are widely sepa- 
rated medially and situated anteriorly and medial 
to each choana. Two corners of the vomer, proba- 
ble remnants of the medial and posterior wings, 
and the included side outline the antero-medial 
edges of the choanae. The top of the triangular 
bone is directed towards the tip of the snout. 
Parasphenoid.?The parasphenoid bone is 
roughly T-shaped and located ventrally on the 
posterior part of the skull. The top of the T lies 
ventral to the prootic while the upright extends 
anteriorly to contact the postero-ventral part of 
the sphenethmoid complex. 
In Atelopus the anterior projection of the para- 
sphenoid is as wide or wider and usually longer 
than the lateral arms. The anterior projection abuts 
against or slightly overlaps the posterior part of the 
sphenethmoid complex; its terminal end is rounded 
or slightly pointed. There are well-developed ridges 
medial to the point of juncture of the anterior 
projection and lateral wings of the parasphenoid. 
These ridges provide a suitable elevated area for 
attachment of muscles. The lateral wings extend 
beneath the prootic. The edges of the lateral wings 
usually parallel each other and are directed per- 
pendicular to the anterior projection of the T. In 
some species of Atelopus the leading edges of the 
lateral wing may be directed slightly posteriorly. 
The posterior edge is usually straight, but some 
species have a slight posterior projection along 
the midline towards the foramen magnum. 
The anterior projection of the parasphenoid is 
about two times the length of the lateral wings in 
Dendrophryniscus. In addition there is always a 
rounded posterior projection. In other details it is 
nearly the same as Atelopus. 
In Melanophryniscus the parasphenoid is fused 
with the underlying base of the skull. The outline 
1971 MORPHOLOGY AND EVOLUTION OF NEOTROPICAL FROGS 17 
of the parasphenoid, obvious in some individuals, 
indicates that the anterior projection is much 
broader and longer than the lateral arms. This gives 
an overall dagger-shape to the parasphenoid. When 
discernible, a posterior projection is well devel- 
oped. Because of the overall fusion of this bone 
with the underlying chondrocranium, it is impos- 
sible to determine the extent and shape of the 
lateral arms and posterior projection in certain 
specimens and species, especially M. stelzneri. 
The parasphenoid of Oreophrynella is basically 
the same as in the other three genera. However, 
the anterior projection is much narrower, covering 
only one-fourth of the interorbital width of the 
skull; in the other genera the width of the anterior 
projection is always more than one-third of the 
interorbital width. The lateral arms are directed 
slightly anteriorly. A posterior process is present. 
Squamosal--The squamosal consists of two 
parts, a shaft or stem which extends from the lat- 
eral margins of the prootic ventrally meeting the 
quadrate and quadratojugal at the point of articu- 
lation with the lower jaw, and a dorsal arm which 
bends over the lateral portion of the prootic and 
onto its dorsal surface. This dorsal section of bone 
has been referred to as the temporal plate (Sand- 
ers, 1953:38) and as the otic plate (Tihen, 1962a: 
160). Griffiths (1954b) discussed a bone which 
he referred to as an otic element. This structure 
will be considered in detail later. 
The squamosal is basically the same in those 
species of Atelopus that were available for exami- 
nation. The lower portion of the shaft is turned 
postero-medially. This rotation forms a prominent 
lateral ridge along the upper half of the shaft. In 
some species there is a small hook at the lower tip 
of this lateral ridge. The lower half of the shaft 
is usually flat and bladelike with the medial edge 
lying more posteriorly than the lateral edge. In 
many species there is a flange on the upper part 
of the squamosal shaft which extends slightly an- 
teriorly and bends medially. This flange, when 
present, gives the upper shaft of the squamosal 
an expanded appearance and provides greater area 
for attachment of some of the adductor muscles of 
the lower jaw. When this flange is restricted to the 
dorsal anterior corner of the shaft, it is referred 
to as an anterior projection of the dorsal arm. 
The dorsal arm of the squamosal is directed 
posteriorly and extends onto the dorsal surface of 
the crista parotica or the prootic. Often the dorsal 
arm and the upper part of the squamosal shaft are 
sculptured. There is considerable interspecific vari- 
ation in the angle formed by the shaft and posterior 
arm. The angle is nearly right in some and obtuse 
in other. In a few species of Atelopus the dorsal 
arm is bent towards the shaft. This connection is 
best described by two obtuse angles. Considerable 
variation was also noted in the position and direc- 
tion of the shaft in relation to the arm. In some 
species the shafts are straight and lie directly be- 
neath the arm. In others, the shafts may diverge 
ventrally or bow inward. 
The squamosal of Dendrophryniscus is similar 
to that of A telopus. The lower portion of the shaft 
is bladelike but is not rotated medially as much 
as it is in Atelopus. As a result, the lateral ridge 
is not formed. The dorsal arm has a well-developed 
anterior projection that bends ventrally. The dor- 
sal arm extends medially over the prootic in D. 
brevipollicatus; it barely overlays the prootic in 
D. minutus. In occipital aspect the anterior projec- 
tion extends slightly lateral to the body of the shaft. 
The shaft itself is bowed and slightly divergent 
ventrally. 
In Melanophryniscus the squamosal stem is 
bladelike; it turns medially in three species, but the 
broad face remains lateral in M. moreirae. There 
is no lateral ridge on the squamosal of the latter 
species. A lateral ridge is present in the species 
with the medially turned stem. An anterior pro- 
jection is lacking in all Melanophryniscus exam- 
ined. The dorsal arm is very narrow and does not 
overlap the prootic dorsally. Rather it abuts 
against the crista parotica of the prootic. The en- 
tire squamosal is situated more anteriorly than it 
is in either Atelopus or Dendrophryniscus. 
The squamosal shaft of Oreophrynella is nai 
row and laterally flattened. There is no lateral 
ridge or medial rotation. The dorsal arm extends 
over the lateral edge of the prootic and onto its 
dorsal surface. There is a large anterior process 
that is somewhat ventrally directed. In occipital 
view the shaft is straight and directly below the 
dorsal arm. 
Quadratojugal-The quadratojugal is a small 
bone located slightly medial and ventral to the 
squamosal and fused intimately to the quadrate. 
In most forms it extends anteriorly towards the 
posterior part of the maxilla. 
The quadratojugal is L-shaped and present in 
all species of Atelopus examined. Because of its 
close association with the quadrate, it is difficult 
to determine its posterior extent. The quadrato- 
jugal may or may not contact the ventral shaft of 
the squamosal; it rarely touches the pterygoid. In 
most species the anterior tip of the quadratojugal 
reaches to or overlaps with the maxillary. How- 
ever, there is a reduction of the quadratojugal in 
some species of Atelopus (e.g. ignescens, pachy- 
dermus). In these forms there appears to be a 
ligamentous or cartilaginous connection between 
18 BULLETIN OF LOS ANGELES COUNTY MUSEUM OF NATURAL HISTORY No. 12 
the posterior tip of the maxilla and the quadrato- 
jugal. 
In Dendrophryniscus brevipollicatus the quad- 
ratojugal is reduced and is situated on the antero- 
ventral edge of the quadrate; in D. minutus it is 
reduced to only an ossified tip of the quadrate. 
The posterior tip of the maxillary and the quad- 
ratojugal are widely separated. 
The quadratojugal is small and restricted to the 
antero-ventral part of the quadrate in Melano- 
phryniscus. The anterior process, if present, is very 
short and widely separated from the maxilla. There 
appears to be a ligamentous connection between 
these two bones. In some specimens of M. moreirae 
the quadratojugal is absent. One specimen has a 
trace of the bone on one side only. Badenhorst 
(1945) reported that the quadratomaxillary 
(= quadratojugal) was small in her specimens of 
M. moreirae. Apparently there is a trend toward 
reduction and eventual loss of this bone in species 
of Melanophryniscus. It would be interesting to 
know whether the absence of this bone is restricted 
to specimens from a single locality or widespread 
throughout the species. 
The quadratojugal in Oreophrynella is absent 
from most specimens examined. One individual has 
a trace of ossification on one side. Whether this 
is actually a remnant of the quadratojugal or just 
a local calcium deposit is impossible to ascertain 
with the material available. 
Pterygoid.? The pterygoid lies medial to the 
maxilla and extends posteriorly to articulate with 
the otic capsule and the squamosal-quadrate com- 
plex. In Atelopus the pterygoid is closely associ- 
ated anteriorly with the posterior portion of the 
medial shelf of the maxilla, upon which it rests. 
From this point the pterygoid extends forward 
along the medial shelf approaching the nasal-max- 
illa-palatine juncture. Posteriorly the dorsal edge 
of the pterygoid arches dorso-medially to contact 
the otic capsule. The pterygoid articulates with the 
prootic via two heads. The medial head is the 
largest and attaches to the ventral surface of 
the otic capsule just anterior to the lateral tips 
of the parasphenoid and the fenestra ovalis. The 
lateral head attaches to the ventro-lateral tip of 
the prootic and the crista parotica and lies medial 
to the upper portion of the shaft of the squamosal. 
The eustachian tube passes posteriorly between 
these two heads. The lateral head is poorly devel- 
oped or absent in those species of Atelopus that 
have middle ear bones. 
The flattened postero-ventral process of the 
pterygoid passes medial to the squamosal shaft 
and folds around the quadrate. The postero-ventral 
process of the pterygoid together with the shaft 
of the squamosal effectively enclose the quadrate. 
Ventrally the pterygoid forms an obtuse angle 
where it contacts the maxilla anterior to the pos- 
terior tip of the latter bone. 
The pterygoids of both Melanophryniscus and 
Dendrophryniscus are basically the same as in 
Atelopus. However, in Melanophryniscus there 
is no close association between the maxilla and 
the anterior process of the pterygoid. In this con- 
dition Dendrophryniscus is closer to Atelopus than 
to Melanophryniscus. In both Melanophryniscus 
and Dendrophryniscus the posterior processes are 
reduced. The dorsal process connects with the 
antero-ventral part of the otic capsule through a 
large cartilaginous complex formed from the crista 
parotica and the process pterygoideus. The postero- 
ventral process is flattened and greatly reduced. 
The eustachian tube passes medially and dorsally 
to the pterygoid through the extensive cartilage 
on the lateral edges of the prootic. 
The dorsal connection of the pterygoid to the 
otic capsule in Oreophrynella is similar to the con- 
dition obtained in the other three genera. The ven- 
tral process does not extend to the lower medial 
edge of the squamosal shaft, as it does in some 
Atelopus and Dendrophryniscus, but rather is lo- 
cated medial to the quadrate about half way up 
the squamosal shaft. There is a much more gradual 
slope from the dorsal process to the anterior tip 
of the pterygoid in Oreophrynella, Melanophry- 
niscus, and Dendrophryniscus than is found in 
A telopus. 
Sphenethmoid complex.?Because of the diffi- 
culty in determining the internal construction of 
the nasal region and because of the almost com- 
plete fusion of endochondral elements, including 
the ethmoid, with the nasal cartilages anterior to 
the orbitosphenoid fontanelle, the entire anterior 
portion of the cranium is referred to as the sphen- 
ethmoid complex. 
In all species of Atelopus (except A. boulengeri), 
Melanophryniscus, and Dendrophryniscus the 
sphenethmoid complex is extensively ossified and 
about the same shape. This ossification includes 
most of the endochondral structure of the olfac- 
tory capsule. None of the bufonids examined 
exhibit this extensive ossification. Dorsally the 
posterior portion of the nasals and the prefrontals 
anterior to the frontal fontanelle are supported by 
an ossified sphenethmoid complex. In most species 
of Atelopus and Dendrophryniscus the nasal por- 
tion of the sphenethmoid is ossified and extends 
anterior to the nasal bones. In some species the 
anterior portion of the sphenethmoid complex is 
cartilaginous and referred to as nasal cartilage. 
This anterior extension of the sphenethmoid (car- 
1971 MORPHOLOGY AND EVOLUTION OF NEOTROPICAL FROGS 19 
tilaginous or ossified) contributes to the acute 
profile of the snout in many species. In Melano- 
phryniscus the anterior extension of the nasal carti- 
lage is present only in the species rubriventris and 
tumifrons. The fusion of the dorsal dermal bones 
in Melanophryniscus makes it difficult to deter- 
mine the posterior and lateral extensions of the 
sphenethmoid unit. 
In ventral aspect there are several ossified 
processes of the sphenethmoid complex. A pala- 
tine process extends laterally toward the maxilla 
and forms a postorbital ridge that supports the 
palatine bones. The palatine process is usually ex- 
panded distally. Ventrally the anterior portion of 
the sphenethmoid complex is ossified beneath the 
vomers in Atelopus and Melanophryniscus but 
only under the posterior edge of the vomers in 
Dendrophryniscus. In the single large specimen of 
A. boulengeri available, the anterior half of the 
sphenethmoid complex is unossified. Both the pala- 
tines and vomers are supported by cartilaginous 
material rather than ossified portions of the sphe- 
nethmoid. Perhaps lack of ossification is correlated 
with the large size of this specimen. In specimens 
of Atelopus and Dendrophryniscus the tip of the 
parasphenoid slightly overlaps the sphenethmoid 
and together with the overlapping prefrontals on 
the dorsal surface gives added support between the 
anterior and posterior parts of the skull. In Mela- 
nophryniscus there is a more intimate connection 
between the two halves of the skull, especially on 
the ventral surface. The posterior extension of the 
sphenethmoid complex fuses with the anterior pro- 
jection of the prootic, thus reducing the extent of 
the orbitosphenoid fontanelle. As previously men- 
tioned, there is also dorsal fusion of the dermal 
elements of the nasal-frontoparietal, and ventral 
fusion of the parasphenoid-sphenethmoid complex. 
These fusions add rigidity to the skull. 
The nasal capsules appear to be widely sepa- 
rated in Atelopus. In Melanophryniscus and Den- 
drophryniscus they are only narrowly separated. 
The ossification of the sphenethmoid complex 
in Oreophrynella is restricted to the ethmoid or 
posterior section. The entire nasal capsule and 
anterior portion of the sphenethmoid complex are 
reduced and unossified. This condition is very simi- 
lar to that obtained in most species of New World 
Bufo (Tihen, 1962a). 
Orbitosphenoid?-The optic nerve passes from 
the brain into the orbit through the optic foramen 
in the orbitosphenoid fontanelle. This fontanelle 
is posterior to the sphenethmoid complex and an- 
terior to the prootic and is floored midventrally by 
the anterior arm of the parasphenoid. The orbito- 
sphenoid in most Amphibia is the thin cartilagi- 
nous covering of the orbitosphenoid fontanelle that 
surrounds the optic nerve. In most amphibians it 
is unossified. 
All species of Atelopus and Dendrophryniscus 
have an unossified orbitosphenoid. The fontanelle 
is irregular in shape, is continuous above the para- 
sphenoid, and separates the prootic from the 
sphenethmoid complex. Basically the same condi- 
tion is found in Oreophrynella. 
In Melanophryniscus the fontanelle is greatly 
reduced by the apparent ossification of the orbito- 
sphenoid and its fusion with the surrounding endo- 
chondral and dermal elements. Because of this 
extensive fusion, it is impossible to determine the 
limits of the orbitosphenoid in the species of Mela- 
nophryniscus examined. There is secondary ossi- 
fication of the orbitosphenoid above the parasphe- 
noid arm in M. tumifrons and M. rubriventris. 
In M. moreirae and M. stelzneri the ossification 
is more extensive. There is no indication of a 
supraparasphenoid connection between the two 
fontanelles. The extensive ossification in M. 
moreirae and M. stelzneri restricts the orbito- 
sphenoid fontanelle to two circular openings 
smaller than those characteristic of M. tumifrons 
and M. rubriventris. All four species exhibit 
greater ossification of the orbitosphenoid than do 
species of Atelopus, Dendrophryniscus, and Oreo- 
phrynella. 
Prootic--The prootic forms the posterior and 
lateral parts of the endocranium of the skull. The 
auditory capsule and elements of the inner ear are 
contained within the prootic. Laterally the prootic 
ends in a cartilaginous abutment, the crista parotica, 
to which the suspensorium and a portion of the 
depressor muscle of the lower jaw attach. Pos- 
teriorly the prootic terminates in two occipital 
condyles which articulate with the vertebral col- 
umn. The occipital condyles are the only part of 
the exoccipital which are distinct from the pro- 
otic. Because of the intimate fusion between these 
elements, the entire posterior section of the endo- 
cranium is referred to as the prootic. 
There is little difference in the overall appear- 
ance and shape of the prootic in any of the genera 
examined. The taenia tecti transversalis and taenia 
tecti medialis are present in all species available of 
the four genera. Badenhorst (1945:18) reported 
the absence of the taenia tecti transversalis in 
Melanophryniscus moreirae, but it is obvious in 
all of my material. 
Developmental series of A telopus ignescens and 
A. varius indicate that the antero-ventral portion 
of the prootic enclosing the otic capsule anteriorly 
and surrounding the foramina of the third and the 
fifth and seventh cranial nerves is the first to ossify. 
20 BULLETIN OF LOS ANGELES COUNTY MUSEUM OF NATURAL HISTORY No. 12 
Later, the exoccipital and posterior surfaces of the 
prootic ossify. The remainder of the prootic and 
the auditory capsule follows. Nothing is known of 
the order of ossification in the other genera, al- 
though small specimens of Dendrophryniscus indi- 
cate the same sequence. 
Suspensorium.?The quadrate, together with 
contributions from the pterygoid process, crista 
parotica and processes pseudobasalis, make up the 
suspensorium in all species of Atelopus, Mela- 
nophryniscus, Dendrophryniscus, and Oreophry- 
nella. The quadrate is restricted to the lower 
portion of the suspensorium and is cartilaginous 
except in some species where it ossifies in the im- 
mediate vicinity of the quadratojugal. The quad- 
rate articulates with the lower jaw via two rounded 
condyles. 
Mandibular Arch 
The lower jaw is formed by the union of two 
arches, each composed of four basic components. 
The prearticular is the bony posterior portion of 
the mandible. It is grooved along its lateral surface 
and overlaps with the dentary. The dentary lies 
lateral to the anterior half of the prearticular and 
extends anteriorly and laterally to the Meckel's 
cartilage towards the symphysis where it fuses with 
the mentomeckelian. The Meckel's cartilage extends 
the entire length of the mandible to the symphysis 
in the groove of the prearticular. The anterior end 
is ossified to form the mentomeckelian which is con- 
nected to the other mandible by a ligament. Pos- 
teriorly the Meckel's cartilage broadens and turns 
slightly upward to form the articular fossa for the 
quadrate. 
There is some interspecific variation in the rela- 
tionships of the dentary to the prearticular and in 
the respective length of each. The medially directed 
curve in the prearticular is found posterior to a 
midpoint in all forms. The curve is farthest pos- 
terior in Atelopus, followed by Dendrophryniscus, 
Melanophryniscus, and Oreophrynella, respec- 
tively. 
Hyoid Apparatus 
The most important comparative study of the 
anuran hyoid and larynx is that by Trewavas 
(1933). Her exhaustive survey demonstrated the 
value of the hyolaryngeal apparatus in interpreting 
phylogenetic relationships. Parker (1940) and 
Griffiths (1959) also considered the hyolaryngeal 
apparatus and generally confirmed its importance 
in reflecting phylogenetic affinity. Only the hyoid 
portion of the hyolaryngeal apparatus was studied 
in this project. I have used a modified version of 
the terminology of Trewavas (1933) and Peters 
(1964: fig. 27B). 
The hyoid and its associated muscles function 
primarily to move the tongue in and out of the 
mouth during feeding. The hyoid consists of a 
hyoid plate which lies beneath the tongue. The 
hyoid plate gives rise anteriorly to two narrower 
processes, the manubria, which run anteriorly for 
a short distance and then turn laterally and con- 
tinue posteriorly, paralleling the mandible as the 
ceratohyals. Just posterior to the point at which 
the mandible articulates with the quadrate, the 
ceratohyals turn dorsally and continue up to the 
skull where they attach to the lateral portion on 
the otic capsule near the fenestra ovalis. In some 
forms there is a well-developed anterior process or 
continuation of the manubrium. A hyoglossal sinus 
extends deep into the hyoid plate between the 
manubria. Hypobranchial 1 extends laterally from 
the hyoid plate posterior to the origin of the manu- 
brium. In most frogs (see Bufo, Fig. 7) a postero- 
lateral process, hypobranchial 2, extends from the 
lateral margins of the hyoid plate posterior to 
hypobranchial 1. The hyoid plate, manubria, cera- 
tohyals, and hypobranchials 1 and 2 are cartilagi- 
nous. The postero-medial processes, hypobran- 
chials 4, are thin bony shafts which extend dorsally 
and laterally from the posterior part of the hyoid 
plate. Hypobranchial 4 lies dorsal to the clavicle of 
the pectoral girdle and encloses the arytenoid and 
the cricoid. 
In all species of Atelopus the hyoid plate is elon- 
gate, about three times its width at the narrowest 
point near the middle (Fig. 7). The hyoglossal 
sinus is moderate in depth and extends to a level 
near the origin of hypobranchial 1. The manubria 
are moderate in width and give rise to a narrow 
ceratohyal. There is a short anterior process ex- 
tending from each manubrium. These anterior 
processes are one-half to one-third the length of the 
manubria and bend laterally near their tips. In 
some specimens the anterior processes frequently 
rejoin the manubria just past the point where the 
latter turn posteriorly. This loop formation is illus- 
trated by Trewavas (1933: fig. 46). Hypobranchial 
1 is present, has a narrow base, and exhibits an 
anterior flare and usually a smaller posterior flare. 
Hypobranchial 2 apparently is lacking in all Atelo- 
pus examined. Hypobranchial 4 is elongate, about 
equal in length to the hyoid plate, and bends 
sharply dorsad about midway along its length. 
The hyoid apparatus of Dendrophryniscus (Fig. 
7) is very similar to the hyoid apparatus of Atelo- 
pus. The hyoid plate is elongate, its length about 
three times its width. The manubria are short, equal 
to about one-third of the length of the hyoid plate. 
1971 MORPHOLOGY AND EVOLUTION OF NEOTROPICAL FROGS 21 
The ceratohyals are narrow. There are no anterior 
processes. Hypobranchial 1 is narrow at its base 
and terminates with an anterior flare and some- 
times a posterior flare. There is no hypobranchial 
2. Hypobranchial 4 is elongate, equal in length to 
the hyoid plate and bends sharply dorsad about 
midway along its length. 
The hyoid of Oreophrynella quelchii (Fig. 7) is 
similar to that of Atelopus in overall appearance, 
although the hyoid plate is not as elongate as it is 
in either Atelopus or Dendrophryniscus. It is only 
about twice as long as it is wide. The manubria are 
very long, approximately the same length as the 
hyoid plate. Accordingly, the hyoglossal sinus is 
ATELOPUS DENDROPHRYNISCUS 
BUFO 
rCERATOHYAL 
HYPOBRANCHIAL  I 
HYPOBRANCHIAL 2 
HYPOBRANCHIAL 4 
MELANOPHRYNISCUS OREOPHRYNELLA 
FIGURE 7. The cartilaginous and bony (stipple) portions of the hyoid apparatus of representa- 
tive species of Atelopus, Dendrophryniscus, Melanophryniscus, and Oreophrynella. The hyoid 
apparatus of a typical species of Bufo is illustrated for comparison. Not drawn to scale. 
22 BULLETIN OF LOS ANGELES COUNTY MUSEUM OF NATURAL HISTORY No. 12 
much deeper than in Atelopus. The anterior por- 
tions of the ceratohyals are slightly expanded. 
There are no anterior processes. Hypobranchial 1 
is about the same as in Atelopus and Dendrophry- 
niscus; it is not expanded at its base and has small 
terminal flares anteriorly and posteriorly. Hypo- 
branchial 2 is absent. Hypobranchial 4 is shorter 
than the hyoid plate. It extends laterally and 
slightly dorsad but does not have the sharp dorsal 
bend that is found in Atelopus and Dendrophry- 
niscus. 
All four species of Melanophryniscus have ap- 
proximately the same hyoid structure (Fig. 7). 
The length of the hyoid plate is two to three times 
its width. The manubria are moderately long and 
slightly expanded where they give rise to the cera- 
tohyals. There are no anterior processes. The base 
of hypobranchial 1 is broadly expanded and ex- 
tends between one-half and two-thirds the length 
of the hyoid plate. Each hypobranchial 1 expands 
slightly to form a smaller anterior flare and a larger 
posterior flare. Hypobranchial 2 has been lost. 
Hypobranchial 4 extends laterally and bends 
slightly dorsad. Trewavas (1933:459, fig. 43) illus- 
trated Melanophryniscus moreirae under the name 
Dendrophryniscus stelzneri. Badenhorst (1945) 
also illustrated the hyoid of M. moreirae. 
In her classic work on the hyoid and larynx, 
Trewavas (1933) discussed the similarities be- 
tween Bufo, Atelopus, Melanophryniscus, and 
Oreophrynella. These similarities include the ab- 
sence of a constrictor laryngis posterior and a pul- 
vinaria vocalia, the point of attachment of the 
hyoglossus muscle, the shape of the hyoid plate, 
the dorsal bending of hypobranchial 4, the well- 
developed lateral process of the cricoid, and the 
absence (except in Melanophryniscus moreirae) 
of a middle posterior petrohyoid muscle. 
Trewavas (1933:517) suggested that the simi- 
larities among the hyolaryngeal apparati would 
justify the retention of the family Bufonidae for 
the genera Bufo, Atelopus, Oreophrynella, and 
Melanophryniscus. My examinations of Dendro- 
phryniscus indicate that it also lacks the constrictor 
laryngis posterior and, in this character, as well as 
in those previously mentioned, it should be con- 
sidered with the genera Atelopus, Melanophrynis- 
cus, and Oreophrynella. 
Vertebral Column 
The nature of the vertebral centrum has been 
considered important in the major classification of 
frogs. Although some earlier workers (e.g. Cope, 
1865, 1866) recognized the importance of verte- 
bral characteristics, it was not until Nicholls (1916) 
divided the phaneroglossid Anura into four basic 
groups and Noble (1922, 1931) verified and elab- 
orated on Nicholls' work that the importance of 
vertebral characteristics was established. Several 
workers, especially Parker (1934), Griffiths 
(1963), Inger (1967), and Kluge and Farris 
(1969) discussed the exceptions and inadequacies 
in the Nicholls-Noble scheme. 
Regional differentiation--There is considerable 
intergeneric variation in the type and number of 
vertebrae in the forms studied. There is also some 
intrageneric variation in vertebral number. 
A single cervical vertebra, the atlas, occurs in all 
frogs. Primitively, it lacks transverse processes. The 
trunk vertebrae vary in number and shape, but 
always possess transverse processes. A single sacral 
vertebra is characteristic of most frogs. It connects 
to the ilia by means of the lateral diapophyses. The 
urostyle or coccyx articulates with the sacral ver- 
tebra and extends posteriorly and dorsally between 
the ilia. 
Cervical vertebra-There are two basic types of 
cervical vertebra in the genera studied. One, found 
in Melanophryniscus and Dendrophryniscus, is a 
typical atlas lacking transverse processes. The other 
type is characteristic of Atelopus and Oreophry- 
nella and consists of an atlas fused with the first 
trunk vertebra. This fusion results in a cervical 
vertebra with transverse processes, hereafter re- 
ferred to as an atlas complex. 
The primitive atlases have a pair of well- 
developed condylar articulating facets which meet 
the occipital condyles of the skull. These facets are 
slightly concave and located ventro-laterally on an- 
teriorly directed processes. They are usually widely 
separated on the ventral midline. 
The vertebrae of Melanophryniscus and Den- 
drophryniscus lack a neural ridge, although there 
are some irregular knobs and ridges on the dorsal 
surface for muscle attachment. There is a small 
knob on the posterior dorsal surface of the atlases 
of Melanophryniscus and Dendrophryniscus that 
comes in close association with an anteriorly di- 
rected process from the dorsal surface of the first 
trunk vertebra. This dorsal connection probably 
gives added rigidity to the anterior part of the 
vertebral column. 
The postzygapophyses of the atlas are poorly 
developed in Melanophryniscus; the faces of the 
articulating facets are strongly concave and directed 
laterally and slightly posteriorly so that the prezy- 
gapophyses of the first trunk vertebra fit into a 
groove and articulate on three surfaces, one dor- 
sally, one ventrally, and one medially. In Den- 
drophryniscus the postzygapophyses are well 
developed; the articulating surfaces are directed 
ventrally. 
1971 MORPHOLOGY AND EVOLUTION OF NEOTROPICAL FROGS 23 
The atlas is fused with the first trunk vertebra in 
all species of Atelopus. Noble (1922:Plate 3, fig. 
4) illustrated a ventral aspect of the vertebral 
column of Atelopus varius clearly showing the 
fusion of the atlas and the first trunk vertebra. The 
foramen for the spinal nerve is located just pos- 
terior and slightly ventral to the transverse process. 
In some species, including A. ignescens, and A. 
pachydermus, there are small slits on the dorsal 
surface of the atlas that indicate incomplete fusion 
between the atlas and the first trunk vertebra. 
Radiographs of A. carrikeri suggest very little 
fusion between the atlas and the first trunk verte- 
bra; confirmation must await skeletal preparations. 
An ontogenetic series of A. ignescens and A. varius 
clearly shows the progressive fusion of these two 
vertebrae. In most species of Atelopus the dorsal 
surface of the atlas is ornately ridged with a cross 
or triangular pattern. The arms of the cross or the 
base of the triangle are situated above the line of 
fusion between the two vertebrae; the posterior 
projection of the cross forms a dorsal connection 
with the second vertebra. Apparently this gives 
additional rigidity to the vertebral column and per- 
haps represents an early evolutionary stage in 
fusion of the vertebral column. 
The anterior facets that articulate with the occip- 
ital condyles are ventro-laterally located and sep- 
arated medially. The postzygapophyses of the atlas 
complex are typically directed ventrally. In young 
specimens of Atelopus prior to fusion, the post- 
zygapophysis of the atlas and the prezygapophysis 
of the first trunk vertebra are oriented ventrally 
and dorsally respectively. These apophyses are not 
obvious after fusion. 
The only indications of fusion of the atlas and 
the first cervical vertebra in Oreophrynella are the 
presence of transverse processes on the first verte- 
bra and of a foramen for a spinal nerve just an- 
terior to each transverse process. Noble (1926: 
fig. 4) illustrated the vertebral column of Oreo- 
phrynella and showed the fusion of vertebrae one 
and two. An ontogenetic series is not available, so 
nothing can be said concerning the nature of the 
fusion of the zygapophyses of the atlas and first 
vertebra. The postzygapophysis of the atlas com- 
plex is typical; its articulating surface is directed 
ventrally. Dorsal ornamentation is restricted to a 
low A on the posterior two-thirds of the complex. 
The posterior wings of the A project between the 
vertebrae as short flanges and roof the gap. Be- 
cause there is no connection with the next vertebra, 
these flanges apparently give intervertebral protec- 
tion to the spinal cord rather than adding rigidity 
to the vertebral column. 
Trunk vertebrae?All vertebrae lying between 
the cervical vertebra, either the atlas or atlas com- 
plex, and the sacral vertebra are considered to be 
trunk vertebrae. All species of Atelopus examined 
have six trunk vertebrae, although occasionally 
seven are found in A. ignescens and A. carrikeri. 
One specimen of A. longirostris has only five dis- 
crete vertebrae; here the first is fused with the atlas 
complex. There is extreme interspecific and intra- 
specific variation in the shape, length, and direc- 
tion of the transverse processes. The large trans- 
verse processes on the first trunk vertebra posterior 
to the atlas complex are rotated anteriorly and bent 
ventrally. There they form attachments for muscles 
from the scapula and suprascapula and give dorsal 
support to the pectoral girdle. The remaining trans- 
verse processes are directed laterally and generally 
are of about the same length or slightly longer 
posteriorly. In certain species the processes of 
some vertebrae are directed anteriorly, in some 
posteriorly, and in some laterally. The transverse 
processes on the second trunk vertebra are usually 
the broadest. The edges of each transverse process 
are even in some species, and scalloped in others. 
All postzygapophyses are directed ventrally; all 
prezygapophyses are directed dorsally. The dorsal 
surface of each neural arch is ornate and forms a 
cross, a diamond, or a broad triangle. Usually 
there is a dorsal connection or articulation between 
the ornamentation on the neural ridge of consecu- 
tive vertebrae. 
Individuals from some populations, especially of 
Atelopus varius and A. senex, have fusions of 
trunk vertebrae and irregular placement of the ulti- 
mate transverse process. Noble (1926:14) men- 
tioned abnormal specimens of A. varius. One 
specimen of A. senex has the fourth and fifth trunk 
vertebrae and their transverse processes fused; 
another has the third and fourth and the fifth and 
sixth fused with all their transverse processes sepa- 
rated except on the right side of the three-four 
fusion where the processes also are fused. In an- 
other specimen from this same locality, there is a 
transverse process typical of trunk vertebrae on 
the right side of the sacral vertebra; the preceding 
trunk vertebra has an expanded process charac- 
teristic of the sacral vertebra. 
There are only four trunk vertebrae in Oreo- 
phrynella quelchii. The first transverse process is 
the longest and broadest; it is rotated slightly an- 
teriorly and directed posteriorly and ventrally. The 
transverse processes on trunk vertebrae two, three, 
and four decrease in length posteriorly and extend 
laterally. The third and fourth are directed slightly 
anteriorly. The dorsal surface of each neural arch 
has an inverted Y design, the posterior arms of 
24 BULLETIN OF LOS ANGELES COUNTY MUSEUM OF NATURAL HISTORY No. 12 
which overhang the intervertebral space. The post- 
zygapophyses are well developed, and their facets 
are directed ventrally. The facets of the prezyga- 
pophyses are directed dorsally. 
Usually there are seven trunk vertebrae in 
Melanophryniscus, although eight were observed 
in a radiograph of a single specimen of M. stelz- 
neri. The extra vertebra appears to be between the 
first and second or second and third trunk verte- 
brae. The transverse processes on the first vertebra 
are directed anteriorly. Those on the second are 
the longest and broadest; they are rotated anteriorly 
and directed ventrally and slightly posteriorly. The 
transverse processes on vertebrae three through 
seven decrease in length and width posteriorly. 
The edges of the transverse processes are scalloped 
and irregular in M. tumifrons, rubriventris, and 
stelzneri; they are nearly smooth in M. moreirae. 
In M. stelzneri the last transverse process is re- 
duced to a nubbin. 
In M. tumifrons, rubriventris, and stelzneri there 
is considerable dorsal ornamentation on the neural 
arch. This ornamentation consists of an irregular 
elevated cross or triangle design. In the latter the 
edges may be straight or concave. The ornamenta- 
tion is best developed in the anterior four or five 
trunk vertebrae and serves for muscle and ligament 
attachment. The posterior extensions of the dorsal 
ornamentation overlap the intervertebral spaces 
and in one specimen of M. tumifrons actually form 
bony connections at paravertebral points between 
two vertebrae. The zygapophyses are well devel- 
oped in M. stelzneri, rubriventris and tumifrons. 
The articulating facets of the postzygapophyses are 
directed ventrally and medially; those of the 
prezygapophyses are directed dorsally and slightly 
laterally. This type of articulation restricts the 
lateral movement of the vertebrae and together 
with the dorsal ornamentation apparently gives 
added rigidity to the vertebral column. 
In Melanophryniscus moreirae there is very little 
to no dorsal ornamentation or overhang; this leaves 
a wide intervertebral gap. The pre- and postzyga- 
pophyses are well developed and have dorsally and 
ventrally directed articulating facets. This condi- 
tion is more similar to that found in Atelopus than 
to that in other species of Melanophryniscus. It 
apparently allows for a more flexible vertebral 
column than is found in the other three species of 
Melanophryniscus. 
There are normally seven trunk vertebrae in 
Dendrophryniscus, although several specimens 
have only six. In those specimens with six verte- 
brae, obvious fusion has occurred between trunk 
vertebrae one and two or between the sixth and 
seventh or between the seventh and the sacral 
vertebra. Some specimens show no sign of fusion; 
in these it is anticipated that the terminal trunk 
vertebra was incorporated into the sacral vertebra. 
There are transverse processes on all trunk verte- 
brae; none are scalloped. The first is directed 
anteriorly and slightly ventrally. The second is the 
broadest; it is rotated anteriorly and directed ven- 
trally and slightly posteriorly. The third is directed 
posteriorly. The fourth through the seventh are 
thinner and point laterally or slightly anteriorly in 
D. brevipollicatus. The fourth and fifth point pos- 
teriorly, while the sixth and seventh are directed 
laterally in D. minutus. There is very little dorsal 
ornamentation on the neural arch. Two small 
dorso-Iateral projections extend over the interverte- 
bral gap but never connect with the next vertebra. 
The postzygapophyses are large and articulate on 
their ventral face with the dorsally directed prezy- 
gapophyses. 
Sacral vertebra?-The sacral vertebra is single. 
Posteriorly it articulates with the urostyle, and 
laterally it articulates with the ilia of the pelvic 
girdle. 
The primary difference between the sacral ver- 
tebra and the trunk vertebrae is the condition of 
the sacral diapophyses. In Atelopus, Melanophry- 
niscus, Oreophrynella, and Dendrophryniscus the 
sacral diapophyses are flattened and greatly ex- 
panded distally. The ilia attach on the ventral sur- 
face of the distal edge of each diapophysis. If there 
is dorsal ornamentation on the neural arch of the 
trunk vertebrae, it is usually repeated on the sacral 
vertebra, although the ornamentation is often less 
extensive. 
The sacrum of Oreophrynella includes two trunk 
vertebrae that are fused with the normal sacral ver- 
tebra. This fusion is clearly shown by the presence 
of three spinal nerve foramina on the ventral part 
of the sacral vertebra (Noble, 1926: fig. 4). 
Centrum? In all forms studied the intervertebral 
body is fused to the anterior end of the centrum. 
This condition is typically procoelous. The pos- 
terior portion of each centrum is rounded and fits 
into a cup of the centrum of each successive verte- 
bra. The centrum is continuous ventrally in all 
forms. The centrum begins about midway along 
the ventral part of the cervical vertebra and is con- 
tinuous through the sacrum. The only divergence 
from the typical procoelous pattern is found in the 
nature of the articulation of the urostyle with the 
sacral vertebra. In all forms this is accomplished by 
a bicondylar (rather than unicondylar) articulation 
with the coccyx. In some specimens the sacrum 
and coccyx are fused. 
Development.?Mookerjee (1931, 1936) and 
Mookerjee and Das  (1939)  traced the develop- 
1971 MORPHOLOGY AND EVOLUTION OF NEOTROPICAL FROGS 25 
ment of the vertebral column in several species of 
Anura and recognized two different modes of ver- 
tebral development, perichordal and epichordal. 
Griffiths (1963:258) also outlined the develop- 
ment of the vertebral centrum in Amphibia. He 
presented a scheme deriving the three types of 
centra, stegochordal, ectochordal, and holochordal, 
through four different pathways. Griffiths fails to 
give any evidence to support his contention that 
there are two distinct pathways of stegochordal 
central development. Inger (1967) followed Grif- 
fiths' proposal of three centra types but questioned 
the primitiveness of the ectochordal type. Kluge 
and Farris (1969:21-25) reviewed the pertinent 
literature and presented an excellent discussion of 
vertebral development. They pointed out the incon- 
sistencies of Griffiths' proposal and concluded that 
the apparent intrafamilial variability and overlap 
among vertebral categories mitigates against the 
use of this suite of characters in defining inter- 
familial relationships. They suggested that detailed 
study of continuous developmental series of many 
species is necessary before the nature of vertebral 
ossification can be generally applied to anuran 
phylogeny. 
In this effort, I examined three specimens of 
Atelopus varius measuring 8.5, 12, and 15 mm 
and three specimens of A. ignescens measuring 12, 
18, and 20 mm to determine the sequence of ver- 
tebral ossification. Dorsally, with the exception of 
the first two vertebrae that are separate from each 
other and unossified on the midline, the neural arch 
of A. varius is narrow and complete in the smallest 
specimen. In the 12 mm specimen, each vertebra 
has closed dorsally, and the first two are nearly 
completely fused with each other. At 15 mm the 
first two vertebrae have fused to form the atlas 
complex. There is also a gradual widening of the 
neural arch from the smallest to the largest speci- 
mens. All other dorsal and lateral components of 
the vertebrae (zygapophyses, transverse processes), 
with the exception of dorsal ornamentation, are 
present in the smallest specimen. 
The centra are completely ossified ventrally in 
the 8.5 mm specimen but are not connected by 
ossification to the lateral portions of the vertebrae. 
In the 12 mm specimen ossification has begun to 
connect the four ventral corners of the centra to 
the lateral portions of the vertebrae. This ossifica- 
tion is nearly complete in the 15 mm specimen. In 
the larger specimens, there is a thin line of ossifica- 
tion that connects successive vertebrae in the middle 
of the centrum. This probably represents the noto- 
chord. Only in the largest specimen (15 mm) is 
there any indication of ossification and fusion of 
the intervertebral disc to form the procoelous ver- 
tebrae. The bicondylar articulation of the coccyx 
to the sacrum is obvious at the 12 mm stage. 
The three stages of A. ignescens are almost iden- 
tical in the sequence of vertebral ossification. I 
consider this sequence to represent perichordal de- 
velopment (= ectochordal of Griffiths, 1963). The 
ossification of the centra to the rest of the verte- 
brae occurs first in the posterior vertebrae in the 
18 mm stage. Distinct posterior procoelous central 
facets are apparent in the 20 mm stage. Unfortu- 
nately, earlier stages of Atelopus are not available, 
nor are developmental stages of the other genera. 
Kluge and Farris (1969:23) indicated that the 
Atelopodidae have both perichordal and epichordal 
vertebrae. This conclusion was based on an exami- 
nation of Dendrophryniscus minutus {Atelopus 
minutus) and Atelopus cruciger. They indicated 
that the former is "very likely epichordal" while 
the latter is almost certainly "perichordal." I can 
find no indication of epichordal vertebrae in Den- 
drophryniscus minutus. Accordingly I consider all 
four genera perichordal, until developmental series 
prove otherwise. 
Coccyjt.-Griffiths (1963:277-78) reviewed the 
literature concerned with morphogenesis of the 
coccyx (urostyle) and unequivocally demonstrated 
that this bone is formed by the fusion of caudal 
vertrebrae with a single, medial hypochord. This is 
the same conclusion reached by Mookerjee (1931). 
Griffiths goes on to discuss the evolutionary impli- 
cations and early history of the coccyx. 
The coccyx usually articulates via a bicondylar 
socket with the sacral vertebra in Atelopus. How- 
ever, it is fused to the sacrum in some specimens 
of A. spurrelli and A. varius from Panama, and 
in A. glyphus, A. oxyrhychus, and an undescribed 
form from Ecuador. Cope (1865:101) indicated 
that the coccygeal condyle is simple in A. flaves- 
cens. However, examination of A. flavescens shows 
the typical bicondylar condition. There is a well- 
developed lateral flange at the anterior end of the 
coccyx in most specimens of A. spurrelli, A. varius 
and related species. This flange is present but nar- 
row and extends nearly the entire length of the 
coccyx in A. flavescens, A. spumarius, and A. 
zeteki. The lateral flange is absent from A. boulen- 
geri, A. ignescens, A. ebenoides, A. carrikeri and 
other closely related species. Most forms have a 
dorsal ridge which may be slightly expanded on 
the top to form a cap. Normally there is a wide 
gap between the dorsal surface of the sacral verte- 
bra and the ridge of the coccyx. A dorsal ridge is 
reduced or absent in some species, including A. 
boulengeri, A. ignescens, A. oxyrhynchus and A. 
pachydermus. An expanded diapophysis of a trans- 
verse process of a presacral vertebra sometimes is 
26 BULLETIN OF LOS ANGELES COUNTY MUSEUM OF NATURAL HISTORY No. 12 
found on the coccyx just posterior to its articula- 
tion with the sacrum. 
Only in one specimen of Melanophryniscus 
stehneri was the coccyx fused to the sacrum; all 
other specimens of the genus had the normal artic- 
ulation. All four species have a dorsal ridge and a 
slightly expanded flange anteriorly. 
The coccyx is fused to the sacral vertebrae in 
four cleared specimens of Dendrophryniscus brevi- 
pollicatus and one cleared specimen of D. minutus. 
Radiographs of a few specimens indicate an articu- 
lation, but confirmation must await additional skel- 
etal preparations. There is no lateral flange or dorsal 
ridge. Izecksohn (1968) reported that most of the 
specimens of D. leucomystax studied had bicondy- 
lar articulation. However, in three specimens the 
coccyx and sacral vertebra were fused. 
The coccyx of Oreophrynella is typically fused 
to the sacrum and has a greatly expanded lateral 
flange which extends from near the posterior distal 
tip of the sacral diapophyses postero-medially al- 
most to the tip of the coccyx. There is a low, but 
well-defined, dorsal ridge on the coccyx. 
Appendicular skeleton 
Pectoral girdle.-Cope (1864, 1865, 1866, 1867) 
recognized the importance of the structure of the 
pectoral girdle to salientian classification. Follow- 
ing the classical works by Parker (1868) and oth- 
ers, and the modifications by Boulenger (1882), 
the Cope scheme became generally accepted. Noble 
(1922, 1926) pointed to certain problems of a 
classification based on the condition of the pectoral 
girdle and considered the girdle condition useful 
only as secondary evidence in showing relation- 
ships. Griffiths (1963) argued that most Anura 
can be divided into two phylogenetic groups based 
on the presence or absence of epicoracoidal horns. 
Inger (1967) pointed out that several groups of 
frogs apparently achieved the firmisternal (fused) 
condition independently. Inger also argued that 
the arciferal (overlapping) condition probably is 
primitive. Kluge and Farris (1969) also concluded 
that the arciferal condition is primitive. The arcif- 
eral nature of the girdle is characteristic of most 
salamanders (Noble, 1931; Eaton, 1959) and oc- 
curs in developmental stages of many firmisternal 
forms (Noble, 1926; Griffiths, 1963). These data 
support the assumption that the arciferal condition 
of the pectoral girdle is primitive. 
The clavicles form the anterior portion of the 
girdle; the coracoids form the posterior part. Medi- 
ally, epicoracoidal cartilage connects the clavicles 
and coracoids to each other and their respective 
halves. Anteriorly, there is a continuous connec- 
tion between the clavicles; this is characteristically 
a firmisternal union and occurs in all forms exam- 
ined. The coracoids maintain this firmisternal con- 
nection on the posterior midline only in Atelopus. 
The epicoracoidal cartilages between each coracoid 
overlap in Dendrophryniscus, Melanophryniscus, 
and Oreophrynella, thus obtaining the arciferal 
condition posteriorly. Noble (1926) illustrated the 
nature of the overlap in Melanophryniscus stehneri 
and M. moreirae and Dendrophryniscus brevipol- 
licatus. He referred to this condition as arcifero- 
firmisternal and considered it to be an intermedi- 
ate step between the arciferal and the firmisternal 
conditions. Apparently Atelopus represents the end 
point in this shift to firmisterny. 
All species of Atelopus, Melanophryniscus, 
Dendrophryniscus, and Oreophrynella have basi- 
cally the same shaped pectoral girdle, whether the 
condition be one of overlap or fusion (Fig. 8). 
However, Oreophrynella has a much deeper cla- 
vicular area than the other forms, and Dendro- 
phryniscus minutus has a more rectangular shaped 
girdle. The medial border of each coracoid fora- 
men is formed of cartilage in all four genera. Also 
the clavicles and coracoids are fused laterally in 
all Atelopus (except a large specimen of A. bou- 
lengeri) and Melanophryniscus but are separate 
in most specimens of Dendrophryniscus and 
Oreophrynella. The lateral fusion of these two 
bones is considered to be a derived state, appar- 
ently correlated with the shift towards a more 
firmisternal condition. 
Griffiths (1963) focused attention on the devel- 
opment of epicoracoidal horns as a systematic 
character potentially of greater use than the arcif- 
eral or firmisternal nature of the pectoral girdle. 
His argument centered on the taxonomic useful- 
ness of a character, whose states are based only on 
degree of fusion or overlap. Griffiths contended 
that those species which have posteriorly directed 
epicoracoidal horns are generally the arciferal 
families, and those that lack well-developed epi- 
coricoidal horns are generally the firmisternal fam- 
ilies. Based on Griffiths taxonomic survey and in 
the absence of contrary data, Kluge and Farris 
(1969) accepted these criteria as useful in defining 
families of frogs. 
In developing the above outlined argument, 
Griffiths (1963:270) discussed the trend towards 
firmisterny in the atelopodid genera using an illus- 
tration (264, Fig. 8) of a proposed morphological 
continuum starting with the arciferal girdle of 
Bufo, continuing through Dendrophryniscus 
(probably Melanophryniscus!), Oreophrynella and 
a hypothetical condition, to the firmisternal Ate- 
lopus and Brachycephalus (discussed elsewhere, 
MS). This trend purportedly is characterized by an 
1971 MORPHOLOGY AND EVOLUTION OF NEOTROPICAL FROGS 27 
FIGURE 8. Ventral view of the pectoral girdle of (upper left) Atelopus varius LACM 64440, 
(lower left) Melanophryniscus rubriventris LACM 64441, (upper right) Oreophrynella quelchii 
LACM 64443, and (lower right) Dendrophryniscus brevipollicatus LACM 64444. Lines equal 
5 mm. 
28 BULLETIN OF LOS ANGELES COUNTY MUSEUM OF NATURAL HISTORY No. 12 
increase in epicoracoidal horn development and a 
corresponding suppression or loss of the sternal 
element. He argued that with the increase in depth 
of the precoracoidal bridge, the sternal primor- 
dium is separated later and the migrating anlagen 
is reduced while that of the epicoracoidal horns is 
increased. Griffiths stated that "Atelopus has car- 
ried the trend farther: zonally, epicoracoid fusion 
is complete but, behind [posterior to] the coracoids, 
they [the epicoracoids] overlap extensively, the 
sternum being completely suppressed" (italicized 
words are mine). 
Melanophryniscus, Dendrophryniscus, Oreo- 
phrynella, and all species of Atelopus which I ex- 
amined have a discrete sternal element (Fig. 8). 
This element probably represents the mesosternum. 
It shows some variation in the degree of ossifica- 
tion but in all cases it is ossified. In some species 
of Melanophryniscus and Atelopus the posterior 
portion of this sternal element is cartilaginous and 
may represent the xiphisternum. Griffiths (1963: 
270) mentioned occasionally encountering "incipi- 
ent abdominal cartilages" in some atelopodids. He 
gave no further information concerning the distri- 
bution, position or function of these cartilages. 
I am convinced that the element which I am calling 
a sternum is homologous among the four genera 
and is the same element called a sternum by many 
anuran biologists (e.g. Noble, 1931; Goin and 
Goin, 1962), and illustrated by Griffiths in Bufo. 
Further, I could find no indication of the extensive 
epicoracoidal horn development in either Atelopus, 
Melanophryniscus, Dendrophryniscus, or Oreo- 
phrynella nor the extensive posterior overlap of the 
epicoracoids in A telopus that were repeatedly men- 
tioned (Griffiths, 1959:459; 1963:264-270, fig. 8). 
Pelvic girdle?-The pelvic girdle of all five genera 
has the typical V-shape. The pubis and ishia fuse 
to form the posterior parts of the girdle and aceta- 
bulum; the ilia form the arms of the V and fuse 
together posteriorly with the ishia and pubis to 
complete the girdle and form the anterior lip of 
the acetabulum. The ilia are nearly round in cross 
section and articulate with the ventral surface of 
the expanded sacral diapophyses in Atelopus, 
Melanophryniscus, Dendrophryniscus, and Oreo- 
phrynella. 
Limb elements?-The relative length of the limbs 
is extremely variable and generally conforms to 
the relationships discussed below with respect to 
habitus of each form. The differences in limb 
length, especially among species of Atelopus, re- 
flect interspecific variation and in certain cases may 
be attributed to ontogenetic, sexual, or geographic 
variation. There are no distinct differences in shape 
of the forelimbs; all forms have a humeral ridge. 
The hindlimbs are typical of most frogs. There is a 
slight bend in the shaft of the femur of most of 
the forms; the bend becomes weakly S-shaped in 
some of the long-legged species ofAtelopus. 
Mesopodial elements?Because of the inherent 
difficulty in determining homologous elements and 
because of the current confusion in terminology, 
no attempt has been made to standardize names 
of the mesopodial elements. My terminology is a 
composite from several sources that seems to most 
adequately describe the situation in the genera 
studied. 
The carpal elements vary in number from six 
to four. All genera have two proximal elements, 
a radiale and an ulnare. Apparently the interme- 
dium has fused with one of these elements. Ritland 
(1955a) indicated that the intermedium fused with 
the ulnare in Ascaphus. 
The central carpals or centralia are represented 
by two elements in Melanophryniscus and Oreo- 
phrynella but only a single element in Atelopus 
and Dendrophryniscus. The proximal centrale, 
which probably represents a fusion of centralia II 
and III, is smaller in Oreophrynella than in the 
other four genera. Another central carpal element, 
perhaps centrale I, is located distally to centrale 
II-III and medial to distal carpal I. Centrale I 
articulates with the medial edge of metacarpal I. 
A distinct bony element is located distal to and 
articulates with centrale I. This element is con- 
sidered to be a prepollex. However, the possibility 
exists that it represents centrale I and that the other 
central carpal elements would have to be numbered 
accordingly; centrale I would become II, and cen- 
trale II-III would become III. Eaton (1959) and 
Romer (1962) suggested that the fourth central 
carpal is fused with the radiale in modern anurans. 
The large size of centrale II-III in Melanophrynis- 
cus might be the result of a fusion of central ele- 
ments II-IH-IV. The carpal bone that lies distal 
to this centrale and medial to distal carpal I would 
then be homologous to the element referred to as 
a prepollex in Oreophrynella. However, its posi- 
tion suggests that the prepollex has been lost in 
Melanophryniscus and that the distal element is I 
and the proximal element is centrale II-III. 
In Atelopus and Dendrophryniscus the first cen- 
trale has fused with centrale II-III, and together 
they form a single central carpal element, centrale 
I-II-III, that lies between the radiale and meta- 
carpal I. 
The distal elements in the manus are referred 
to as distal carpals. They are numbered according 
to the digit which they support. All forms have 
two distinct distal carpal elements. The first sup- 
ports digit  1; the second, third, and fourth are 
1971 MORPHOLOGY AND EVOLUTION OF NEOTROPICAL FROGS 29 
fused into a single unit and support digits 2, 3 
and 4. Distal carpals II, III and IV are fused into 
a single element as is characteristic of the other 
four genera. 
The tarsal elements, including the tibiale and 
fibulare (astragulus and calcaneum, respectively) 
number four or five. Digits 4 and 5 articulate with 
the distal head of the fibulare, suggesting a fusion 
of distal tarsal elements IV and V with the fibulare. 
Oreophrynella has distal tarsals I, II, and III at 
the base of digits 1, 2, 3. This same arrangement 
is characteristic of most Melanophryniscus. A sin- 
gle specimen of M. tumifrons has tarsals I and II 
fused. Dendrophryniscus also usually has three, 
but a single specimen of D. brevipollicatus has a 
fusion between distal tarsals II and III. All speci- 
mens of Atelopus possess two distal tarsal ele- 
ments; the second and third are fused. A single 
specimen of Atelopus varius has a fusion between 
distal tarsal I and the centrale that forms the base 
for the prepollex. 
The metacarpal elements are consistent among 
the genera. There are four metacarpals, and they 
generally increase in size in the following order: 
1, 2, 4, 3. The metatarsals of Atelopus, Melano- 
phryniscus, and Dendrohryniscus have the same 
general size relationship, listed in order of increas- 
ing length: 1, 2, 3, 5, 4. In Oreophrynella metatar- 
sals 1, 2, and 3 are of about equal length and 
slightly more than two-thirds the length of meta- 
tarsals 4 and 5, which are of about equal length. 
A distinct prepollex is found in Oreophrynella; 
it may consist of a single or two distinct elements. 
Apparently the prepollex is absent in Melanophry- 
niscus and Dendrophryniscus, although both have 
a small lateral, irregular spine or ridge that might 
be a remnant of a prepollex that fused to the side 
of metacarpal 1. This is better developed in males 
than in females. In Atelopus there is a single or 
double element that is usually fused to the lateral 
edge of the metacarpal 1. Frequently, it forms a 
lateral ridge on the metacarpal. This element may 
articulate with centrale I-II-III or be separated 
proximally and fused to the metacarpal. In some 
specimens, it extends nearly the entire length of 
the metacarpal; in others it may be restricted to the 
middle half of the metacarpal. In a few specimens, 
this ridge is free from the metacarpal; usually it is 
partially or completely fused to it as least at its 
proximal end. Laurent (1942) showed it com- 
pletely separate from the metacarpal in Atelopus 
varius. This irregular structure is better developed 
in males than in females. Because of the wide 
range of both inter- and intraspecific variation in 
Atelopus, there are two possible interpretations 
concerning the nature of this element. First, it may 
be derived from metacarpal 1, in connection with 
the male holding of the female during amplexus, 
or it may be a remnant of the prepollex that fused 
to the first metacarpal unit. I consider it to be a 
remnant of the prepollex until contrary evidence 
becomes available. 
A prehallux is present in all forms and may con- 
sist of one or two fused elements or three discrete 
elements. The structure is best developed in some 
Atelopus and poorly developed in Dendrophrynis- 
cus. 
Digits?All forms typically have four digits on 
the front foot and five digits on the hind foot, 
although in some species of Atelopus the fifth digit 
of the foot is greatly reduced and incorporated 
into the sole. 
The common phalangeal formula for the hand 
is 2-2-3-3 and is found in Oreophrynella and Mela- 
nophryniscus. Each element is short and robust 
in Oreophrynella, and the terminal element is stout 
and only slightly expanded. The condition in 
Melanophryniscus is similar, but each phalanx is 
slightly more elongate, and the terminal tips are 
very irregular though slightly expanded. There is 
a reduction in the number and length of the pha- 
langes in the first digit in Atelopus. Several species 
exhibit the typical 2-2-3-3 formula, but specimens 
of several species have only a single unit in digit 1 
giving a formula of 1-2-3-3. In some specimens 
the formula is 1-2-3-3 in one hand and 2-2-3-3 
in the other. Apparently there is a trend towards 
reduction in length and ultimate loss of the pha- 
langes in the first digit in some species of Atelopus. 
The tips of the terminal phalanges are generally 
expanded and slightly rounded. 
Specimens of Dendrophryniscus minutus have 
a 2-2-3-3 phalangeal formula and slightly ex- 
panded tips while D. brevipollicatus has lost the 
terminal phalanx in digit 1 and has a formula of 
1-2-3-3. In addition the terminal tips of D. brevi- 
pollicatus are expanded and T-shaped. 
The phalangeal formula for the foot is consist- 
ently 2-2-3-4-3 for Atelopus, Melanophryniscus, 
Dendrophryniscus, and Oreophrynella. The length 
and width of the phalanges in digit one in Atelopus 
are greatly reduced. A single specimen of A. spu- 
marius has a phalangeal formula of 1-2-3-4-3. In 
each of the species, the terminal tips of the feet 
have the same general shape that is characteristic 
of the hands. 
AUDITORY APPARATUS 
The value of the middle ear as a taxonomic 
character in frogs has been recognized for a con- 
siderable period of time. The perfect or imperfect 
condition of the ear was considered by Giinther 
30 BULLETIN OF LOS ANGELES COUNTY MUSEUM OF NATURAL HISTORY No. 12 
(1858a, b) and Mivart (1869) to be of primary 
importance in the classification of anurous Batra- 
chia. More recent authors (Noble, 1931; Parker, 
1934; Griffiths, 1954b, 1959) have demonstrated 
that the absence of a middle ear apparatus is char- 
acteristic of several unrelated species of frogs and 
questioned its value as a higher taxonomic char- 
acter. 
Parker (1932) implied that the absence of a com- 
plete ear apparatus in so many different groups of 
Anura is the result of convergent evolution through 
loss. Later (1934) he reversed himself and sug- 
gested that the absence of a middle ear structure 
was indeed a primitive character, an interpretation 
formulated following the report by Du Toit and 
de Villiers (1932) that the developing ear of young 
Hyperolius horstocki passes through an incomplete 
ear stage, almost exactly the same as the condition 
in earless frogs, before attaining the typical com- 
plete ear of the adult. 
Griffiths (1954b) interpreted the absence of an 
ear apparatus in the African frog Didynamipus 
sjostedti and the neotropical species of Dendro- 
phryniscus and Atelopus as indicating relation- 
ships and suggested that the African and neo- 
tropical frogs were evolved from a common stock 
in Africa; the neotropical group subsequently dis- 
persed. Later (Griffiths, 1959) he changed his 
mind and decided that the absence of a complete 
ear in the African and South American species was 
the result of parallel loss in the two groups. Each 
group was evolved independently from an ances- 
tral bufonid stock residing in the same area as 
their respective descendants. He considered the 
absence of the complete ear as a paedomorphic 
character of relatively common occurrence in 
Salientia. 
Ramaswami (1939) reviewed much of the ear- 
lier work on the middle ear of frogs. Eiselt (1941) 
made an exhaustive study of the opercular muscle 
and the middle ear in Anura. Tumarkin (1955) 
and Sedra and Michael (1959) discussed the cur- 
rent ideas concerning the ontogenesis of the anuran 
auditory mechanism. In general, their terminolo- 
gies are incorporated into this study. 
The middle ear of frogs consists of a tympanum, 
a middle ear cavity, and ear bones. The typmanum 
is a thin membrane which covers the external 
opening of the middle ear. It may be covered 
by a thin differentiated portion of skin and be 
apparent externally; this is referred to as an ex- 
ternal tympanum. It may be covered by undiffer- 
entiated skin and not be visible externally; this is 
referred to as an internal tympanum. Some frogs 
have various degrees of intermediacy between an 
external and internal tympanum.  The tympanic 
annulus surrounds the tympanic membrane. It is 
a cartilaginous ring derived from the quadrate. 
The middle ear cavity surrounds the ear bones 
and lies between the tympanum and the inner ear. 
The Eustachian tube connects the middle ear cav- 
ity with the pharyngeal cavity. The openings of 
the Eustachian tubes in the pharyngeal region of 
the mouth are called the ostia pharyngea. The ear 
bone in frogs is called a pseudocolumella and is 
composed of three parts: 1) a cartilaginous foot- 
plate (pars interna plectri or interstapedial) which 
is derived from and abuts against the otic capsule 
in the fenestra ovalis; 2) a bony stylus (pars media 
plectri or mediostapedial) which also is derived 
from the otic capsule and connects to 3) a carti- 
laginous extracolumella (pars externa plectri or 
extrastapedial) lying across the inner surface of 
the tympanic membrane. The extrastapedial appar- 
ently is derived from the quadrate. The fenestra 
ovalis (oval window) is the membranous area on 
the ventro-lateral wall of the otic capsule that 
separates the middle ear from the inner ear. 
In frogs with a tympanic-middle ear apparatus, 
hearing in adults is accomplished in at least two 
ways. Airborne sounds, picked up by the tym- 
panum, are passed down the ear ossicles to the 
fenestra ovalis and then to the inner ear. In frogs 
that lack a tympanic-middle ear apparatus appar- 
ently "hearing" is accomplished by a muscular 
connection between the suprascapula and the otic 
capsule. Posterior and medial to the footplate of 
the pseudocolumella and lying in the fenestra 
ovalis is a bony element called the operculum. The 
movable operculum is derived from the otic cap- 
sule and connects to the suprascapula via the oper- 
cular muscle. The opercular muscle is a specialized 
derivative of the levator muscles of the scapula. 
In a terrestrial environment vibrations from the 
ground are transmitted through the pectoral girdle 
to the operculum and then to the middle ear via 
the tensed opercular muscle. This pathway is avail- 
able to frogs with a typical tympanic-middle ear 
apparatus. 
In addition to examining the auditory apparatus, 
the ability of these frogs to call was investigated 
with reference to the presence of vocal slits and 
the nature of the vocal sac. The terminology used 
in their descriptions follows Peters (1964). 
Middle ear?A tympanum, annulus tympanicus 
and ear ossicles are absent from all species of 
Melanophryniscus, Dendrophryniscus, and Oreo- 
phrynella examined. All species of Atelopus also 
lack these structures with the exception of A. 
flavescens, A. spumarius, and an undescribed spe- 
cies, each of which lacks an external tympanum 
but possesses a moderate to well-developed internal 
1971 MORPHOLOGY AND EVOLUTION OF NEOTROPICAL FROGS 31 
tympanum, annulus tympanicus and ear ossicles. 
Opercular structure?All specimens of Atelopus, 
Melanophryniscus, Dendrophryniscus, and Oreo- 
phrynella have a well-developed operculum. Simi- 
larly, they all have a well-developed opercularis 
muscle which arises on the operculum and inserts 
on the ventral surface of the anterior distal part 
of the suprascapula. This condition, as well as the 
typical middle ear structure, is present in the three 
species that have the middle ear. 
Ostia pharyngea.?No specimens of Dendro- 
phryniscus have ostia pharyngea. Oreophrynella 
quelchii has small indentations that appear to be 
remnants of the pharyngeal opening of the Eusta- 
chian tubes, but there are no ostia pharyngea. All 
species of Melanophryniscus have ostia pharyngea. 
There is some variation in the size of the opening. 
M. stelzneri has the largest opening, and M. 
moreirae has the smallest. Species of Atelopus 
show as much variation among themselves as is 
shown by the other genera. Specimens of A. 
flavescens always have very large ostia pharyngea. 
Some species of Atelopus include individuals with 
ostia pharyngea and individuals without them. 
Ostia pharyngea are lacking completely in all 
specimens of some species. 
Vocal apparatus.?There are no vocal slits in 
available specimens of Dendrophryniscus or Oreo- 
phrynella. All specimens of the four species of 
Melanophryniscus examined have vocal slits. The 
vocal slits may be on one side or on both sides, 
but their positions are constant for each species. 
Males of M. stelzneri have a well-developed ex- 
ternal vocal pouch. External vocal pouches are not 
evident in the other species. In species of Atelopus, 
vocal slits may be absent or present. When present, 
they may be on the right side only, the left side 
only, or on both sides. External vocal sacs were 
not observed in specimens of Atelopus, Oreo- 
phrynella, or Dendrophryniscus. 
EXTERNAL MORPHOLOGY 
Habitus 
Frogs of the genus Atelopus are considerably 
different from most frogs in overall appearance and 
are referred to as atelopodid-like: the head is 
usually longer (snout to end of lower jaw) than it 
is wide (at the point of jaw articulation); the snout 
is strongly acuminate from a dorsal aspect and 
typically acute from a lateral profile; round nares 
open laterally and generally are situated closer to 
the tip of the snout than to the eye; the moderately 
sized eyes may be longer than, equal to, or shorter 
than the distance between the eye and the naris; 
the pupil is horizontally elliptical; there is no exter- 
nal tympanum, although there may be an elevated 
(supratympanic) area dorsal to where a tympanum 
is normally located. The body length (snout to 
vent) of adult males is generally greater than three 
times the body width (at sacrum). The limbs are 
usually long and spindly. Many species are brightly 
colored with vivid shades of red, yellow, or green, 
often contrasted in irregular patterns with dark 
colors, especially black. 
Dendrophryniscus brevipollicatus and D. minu- 
tus are smaller than most Atelopus species but very 
similar in head proportions (head length longer 
than or equal to head width) and snout shape. The 
round nares are closer to the tip of the snout than 
to the eye; the eye is proportionally smaller than 
in Atelopus, less than or equal to the distance from 
the eye to the naris. An external tympanum is 
absent; there is no indication of a supratympanic 
elevation. The species of Dendrophryniscus are 
slightly more robust than Atelopus; the body length 
ranges from two and one-half to three and one- 
quarter times the body width. The limbs are 
moderately long and thin. Dendrophryniscus is not 
brightly colored but possesses a cryptic pattern of 
brown or tan, often barred with darker markings 
dorsally. 
The species of the two remaining genera, Melan- 
ophryniscus and Oreophrynella, are different from 
both Atelopus and Dendrophryniscus in habitus. 
They generally resemble toads in overall appear- 
ance and are referred to as bufonidlike: heads are 
wider than they are long; from a dorsal aspect the 
snout is acuminate or slightly rounded; in profile 
the snout is rounded or vertical; the laterally open- 
ing nares are round and close to the tip of the 
snout; the eye is moderate in size and equal to or 
slightly less than the distance from the eye to the 
naris; the pupil of the eye is horizontally elliptical; 
an external tympanum is lacking. A supratympanic 
elevation is absent in Melanophryniscus and poorly 
developed in Oreophrynella. These genera are rep- 
resented by more robust species than are found in 
either Atelopus or Dendrophryniscus. Dorsally the 
species of Melanophryniscus and Oreophrynella 
are usually uniform dark brown or black often 
spotted or marked with yellow. The bottoms of the 
hands and feet are bright orange, yellow, or red. 
The bright colors on the ventral surface of the 
body are marbled with black except in M. rubri- 
ventris which is uniformly red beneath. 
Integument 
The skin of anurans has certain specific pecu- 
liarities which may be valuable in appraising rela- 
tionships. Of primary concern in this study are the 
texture of the epidermis and the nature and ar- 
rangement of the glands as they contribute to the 
32 BULLETIN OF LOS ANGELES COUNTY MUSEUM OF NATURAL HISTORY No. 12 
general appearance of the frog. Wherever possible, 
I have attempted to utilize the nomenclature pro- 
posed by Elias and Shapiro (1957) to describe the 
glandular development. 
The skin of most species of Atelopus is smooth 
to slightly rugose; in some forms (cruciger and 
oxyrhynchus) there is a dorso-lateral row of 
rounded warts (verrucae) along the body. Later- 
ally on certain other species {carrikeri, ignescens, 
and walkeri) there are warts with a thickened 
stratum corneum at their apex. These warts are 
different from those forming the dorso-lateral row 
and are referred to as coni apicali. No species have 
parotoid glands, although in some, including A. 
ignescens, there is glandular development in the 
parotoid region. Cursory examination of these 
glands in A. ignescens indicates that they are of 
the same type as those found on the dorso-lateral 
regions of the head, above the eyes, in scattered 
areas on the back, sides, and dorsal surfaces of the 
humerus and femur, and over the entire dorsal sur- 
face of the distal sections of each limb, onto the 
hands and feet. Microscopic examination of a thin 
section of glandular tissue indicates that the glands 
are strikingly different in size and structure from 
those that form the parotoid glands of most species 
of Bufo. 
Dendrophryniscus has a minutely glandular skin 
with small scattered warts, some of which are 
topped with low coni apicali. There is no glandu- 
lar development in the parotoid area. Low eleva- 
tion in this area results from the dorso-lateral 
extensions of the posterior arm of the squamosal 
and the lateral part of the prootic. 
All specimens of Melanophryniscus that were 
examined possess a rugose skin covered with ver- 
rucae. These warts are rounded in M. moreirae and 
rubriventris; in tumifrons the warts are tipped with 
one to several coni apicali, usually lighter in color 
than the verrucae. M. stelzneri has an intermediate 
condition. These coni are best developed in the 
lateral areas of the head and neck between the eye 
and the foreleg insertion. In all Melanophryniscus 
examined the warts on the dorsal surfaces are more 
abundant and better developed than those on the 
ventral surfaces. There are no discrete parotoid 
glands in any species. However, M. tumifrons has 
two nearly confluent large verrucae on the snout. 
The skin of Oreophrynella quelchii is rugose and 
covered with many small rounded warts. In over- 
all appearance it is very similar to M. moreirae. 
There are no distinct or enlarged warts in the paro- 
toid region. 
Other integumentary characters are the presence 
and configuration of a ridge along the upper eyelid. 
This ridge is found in some species of Atelopus 
and is absent from others. When the ridge is 
present, it forms a thickened brow. In Dendrophry- 
niscus the ridge is formed by about 12 small warts; 
in Melanophryniscus it is thickened in all species. 
In M. moreirae this thickening is edged with four 
or five warts that are rounded; in tumifrons the 
warts are pointed. 
Hands and Feet 
Several characters of varying significance which 
pertain to the hands and the feet have been used by 
anuran systematists in the past. The presence of a 
tarsal fold, the number of digits on the hands and 
feet, the shape of the terminal ends of the digits, 
the presence and configuration of tubercles on the 
palms and soles of the hands and feet, and the 
condition of the interdigital webbing are con- 
sidered in this study. 
A tarsal fold has been reported for some species 
of Atelopus (Rivero, 1963:115) but is absent from 
most species which I have examined. There is no 
tarsal fold in species of the other four genera. 
All species of Atelopus have four fingers and 
five toes, although the first digit of the hand may 
be reduced in length, fully webbed, and therefore, 
not immediately apparent. The digital tips of all 
species examined vary from slightly to not ex- 
panded. Four fingers and five toes are also charac- 
teristic of all species of Dendrophryniscus, Melan- 
ophryniscus and Oreophrynella. The inner two 
digits on the hind foot of Oreophrynella quelchii 
are opposable; the first toe is longer than the 
second and gives the foot an overall broader ap- 
pearance, quite distinct from species in the other 
genera. The digital tips are expanded in Dendro- 
phryniscus brevipollicatus but not in D. minutis. In 
contrast, the tips of the digits in Melanophryniscus 
are slightly thickened, and in Oreophrynella they 
are markedly thickened; the digits are not expanded 
in species of either genus. 
Many species of Atelopus have marked inter- 
specific variation in the condition of the tubercles 
on the hands and feet. Some species have moder- 
ately developed palmar tubercles. Subarticular 
tubercles, if present, are usually small and indis- 
tinct. The species of Dendrophryniscus, Melan- 
ophryniscus and Oreophrynella are consistent in 
having tubercles on both the hands and the feet. 
Dendrophryniscus has a large outer palmar tuber- 
cle and some subarticular tubercles on the hands 
and an outer metatarsal tubercle and several sub- 
articular tubercles on the feet. Species of Melan- 
ophryniscus have a moderate to well developed 
outer metacarpal tubercle and inner and outer 
metatarsal tubercles. Additional metacarpal, meta- 
tarsal, and subarticular tubercles are also present. 
1971 MORPHOLOGY AND EVOLUTION OF NEOTROPICAL FROGS 33 
All tubercles are well developed in specimens of 
M. tumifrons and stelzneri and poorly to moder- 
ately developed in M. moreirae and rubriventris. 
In Oreophrynella quelchii there are two flat meta- 
carpal tubercles, an inner and an outer, and an 
inner metatarsal tubercle. There are no distinct 
subarticular tubercles. Rather, there are three to 
five smaller tubercles on the ventral surfaces of the 
hands and feet near the bases of the digits. The 
rest of the palmar and soler surfaces are irregularly 
wrinkled and folded and covered with many small 
tubercles. 
The interdigital webbing in the genera studied 
differs from that observed in other frogs, such as 
hylids or ranids. It is not a membrane as denned by 
Peters (1964) but rather a thickened integumen- 
tary connection between digits, similar to the web- 
bing encountered in many of the more terrestrial 
anurans, such as toads of the genus Bufo. 
The condition of the interdigital webbing shows 
a great deal of variation in species of Atelopus. 
The fingers generally are not, or are only slightly, 
webbed, with the greatest development between the 
first and second digit. The toes vary from a slightly 
to fully webbed condition. Dendrophryniscus has 
no, or very rudimentary, webbing on the hands and 
rudimentary webbing on the feet. The specimens 
of Melanophryniscus and Oreophrynella show no 
webbing on the hands. The digits of the feet are 
joined at their bases by a fleshy extension of thick 
skin from the sole. While this has been interpreted 
as "webbing" by some authors (Cochran, 1955:3), 
it is different from the webbing characteristic of 
some species of Atelopus and Dendrophryniscus. 
REPRODUCTIVE BIOLOGY 
Several aspects of the reproductive biology of 
Anura provide information valuable to an under- 
standing of generic relationships (Jameson, 1957). 
Because of the lack of adequate knowledge con- 
cerning the life histories and reproductive behavior 
of the majority of the species studied, most of these 
characters concern morphology. Therefore, it is 
difficult to determine relationships based on simi- 
larities or differences in morphological characteris- 
tics alone. 
Urogenital System 
Uterus? Bhaduri (1953) made an exhaustive 
survey of the urogenital system in Anura. In 
general he concluded that the urogenital apparatus 
has no great significance for defining major taxo- 
nomic groups. He pointed out that the uterine 
condition generally is constant within a genus and 
sometimes within a family, although the extremes 
of uterine modification are found within the limits 
of some families. Bhaduri (1953:64) described the 
uterus of Bufo and referred to this condition as 
uterus septatus. In this condition there is a parti- 
tion of varying length between the uteri except 
posteriorly where a common uterus forms. This 
condition is intermediate between those frogs 
which have two uteri separate along their entire 
length (uterus separatus) and those frogs which 
lack any partition, having the uteri confluent, form- 
ing an unpaired common uterus (uterus corn- 
munis') . 
All specimens of the five genera that I examined 
seem to fall into Bhaduri's uterus septatus group. 
Some species of the bufonid Nectophrynoides ap- 
parently have this condition as well. 
Bidder's Organ.?The anterior portion of the 
progonad retains its ovarian nature in males of 
some frogs and is referred to as the Bidder's organ. 
This structure has been subjected to intensive study 
for nearly 200 years. Witschi (1933) presented a 
review of this structure and discussed its mor- 
phology with reference to sex determination and 
differentiation in several species of toads. He dem- 
onstrated that the cortical layer of the progonad 
supplies the germ cells of the Bidder's organ and 
postulated that the ovarian nature of this structure 
results from the withdrawal of medullary or testes- 
forming tissue from the anterior part of the progo- 
nad. He suggested that the ancestral form of the 
bufonids must have possessed a long gonad. In 
living species during the differentiation of the 
gonad, the functional testes or ovaries are restricted 
to the posterior segment, while the anterior portion 
of the gonatomes give rise to Bidder's organ and 
fat bodies. 
Davis (1936) surveyed the distribution of Bid- 
der's organs in Anura and found that this peculiar 
structure was restricted to bufonid toads. He re- 
ported the presence of Bidder's organs in several 
species of Bufo, Nectophrynoides, Pedostibes, and 
Pseudobufo. He reported their absence from 
Cacophryne borbonica and Nectophryne afra. 
Davis considered the bufonids to be derivatives 
from a leptodactylid stock and favored the hypoth- 
esis that the Bidder's organ originated with the 
Bufonidae. He pointed out that certain leptodacty- 
lids have elongate testes. This fact lends support 
to Witschi's proposition that the ancestral bufonid 
possessed long gonads. 
Dubois (1947) reported the presence of a Bid- 
der's organ in a juvenile Cacophryne borbonica 
and an adult male C. cruentata. He concluded that 
Bidder's organs are found in all genera of Bufonids, 
including Bufo, Pseudobufo, Pedostibes, Peloph- 
ryne, Nectophryne, Nectophrynoides, and Cacoph- 
34 BULLETIN OF LOS ANGELES COUNTY MUSEUM OF NATURAL HISTORY No. 12 
ryne. He considered it a characteristic of the 
Bufonidae lacking in other anuran families. 
Dubois (1947) mentioned the presence of a 
Bidder's organ in the Australian leptodactylid 
Pseudophryne bibronii and in Rhinophrynus. The 
report of a Bidder's organ in Pseudophryne is 
based on Stohler's work (1932). Rau and Gatenby 
(1923) and Ponse (1924) reported this organ 
lacking from Pseudophryne bibronii. Dubois does 
not mention the source of his information concern- 
ing Rhinophrynus. Rau and Gatenby (1923) ex- 
amined several specimens of Rhinophrynus dor- 
salis and could find no trace of a Bidder's organ. 
I also examined several specimens of Rhinophry- 
nus and found no indication of a Bidder's organ. 
Griffiths (1954b, 1959) confirmed the presence 
of a Bidder's organ in Bufo, Nectophrynoides, 
Pedostibes, and Pseudobufo and reported this 
structure from several species of Atelopus, Melan- 
ophryniscus stelzneri, Oreophrynella quelchii, 
Cacophryne borbonica, and Nectophryne afra. 
Inger (1954) reported the presence of Bidder's or- 
gans in Ansonia and Pelophryne. Griffiths (1954b) 
indicated that the Bidder's organ is better devel- 
oped in young individuals than in old individuals 
and pointed out that the structure may be degen- 
erate in fully adult individuals, especially in genera 
other than Bufo. Witschi (1933) and Davis (1936) 
also reported that the Bidder's organ is better de- 
veloped in the juveniles of some species of Bufo. 
Griffiths maintained that the presence of a Bid- 
der's organ in a species is proof of bufonid affiini- 
ties, but that its absence need not, on its own merit, 
establish that such affinities do not exist. He further 
contended that the degeneration or disappearance 
of the organ might be expected either ontogeneti- 
cally or phylogenetically. Later Griffiths (1959: 
477) stated that the presence of Bidder's organs 
in Atelopus, Dendrophryniscus, Oreophrynella, 
Didynamipus, and Cacophryne makes their origin 
from the Bufonidae indisputable. 
A Bidder's organ has been reported from several 
specimens of Atelopus, Melanophryniscus, Oreo- 
phrynella quelchii, and Dendrophryniscus brevi- 
pollicatus. 
Gonads?There is very little information avail- 
able concerning the evolutionary importance of 
differences in size, shape, and coloration of the 
gonads of frogs. Witschi (1933) and Davis (1936) 
mentioned that the ancestral forms of bufonids 
possessed long gonads. Davis (1935) reported that 
Cacophryne borbonica lacks the lengthened testes 
characteristic of other toads. Inger (1966:74) 
indicated that the snout-vent length to testes length 
ratios of C. borbonica and C. cruentata range be- 
tween four and five to one. In Bufo biporcatus the 
ratio was eight to one. 
The testes of some specimens of A. varius from 
near La Hondura, Costa Rica, and of some speci- 
mens of A. longirostris and A. senex have a black 
testicular peritoneum. In all the other species of 
Atelopus, the testes are white; apparently the testi- 
cular peritoneum is colorless. It is possible that the 
black coloration of testicular peritoneum is cor- 
related with the breeding condition of males of 
some species and may shield the testes from ultra- 
violet light. Lack of adequate knowledge concern- 
ing the breeding cycles of Atelopus prevents fur- 
ther discussion at this time. 
In most species of A telopus the ratios of snout- 
vent lengths to testes lengths for mature males 
range from about seven to one in A. ignescens to 
ten to one in A. flavescens. These data are difficult 
to evaluate because of the probable increase in size 
of the testes during the breeding season. 
No males of Dendrophryniscus minutus were 
available; in males of D. brevipollicatus the testes 
are covered with a black peritoneum. The testes 
are not elongated. Ratios of snout-vent length to 
testes length range between 13 and 14 to 1. The 
testicular peritoneum of Melanophryniscus more- 
irae is unpigmented; in M. stelzneri, rubriventris, 
and tumifrons the peritoneum is moderately to 
heavily pigmented black. Snout-vent length to 
testes length ratios vary between seven and eight 
to one. The testes of Oreophrynella quelchii are 
elongate and appear white. Snout-vent length to 
testes length ratios average less than five to one. 
The only information gathered concerning the 
ovaries was the number, size, and pigmentation of 
the eggs they contained. Only animals with well- 
developed eggs were examined. 
Noble (1927) indicated that the eggs of Atelo- 
pus are very small and pigmented. Apparently 
Noble was referring to eggs of a species of Melano- 
phryniscus. In 1926 he listed M. stelzneri and 
moreirae as species of Atelopus. The ovarian eggs 
of several species of A telopus are moderate in size 
and lack pigment. The number of eggs in both 
ovaries varies among the species. As few as 350 
and 372 eggs were found in A. gracilis and A. 
ignescens; 910 eggs were taken from a female 
A. varius. Mature ovarian eggs average about 0.8 
mm in a female A. gracilis (36 mm snout-vent 
length), about 1.3 mm in a female A. ignescens 
(45 mm snout-vent length) and about 1.0 mm in a 
female A. varius (40 mm snout-vent length). There 
is no apparent correlation between the size of the 
female and the number of eggs laid, although only 
three  specimens  were  examined  in  detail.  Ap- 
1971 MORPHOLOGY AND EVOLUTION OF NEOTROPICAL FROGS 35 
parently there is a correlation between the size of 
the eggs and the size of the animal; the smaller 
eggs are found in smaller animals. There is a sug- 
gestion that the size of eggs increases as the num- 
ber of eggs decreases (comparison of A. ignescens 
and A. varius), but probably this size increase is 
subject to limitations imposed by the size of the 
females (A. gracilis and A. ignescens). 
The only report of eggs of Atelopus refers to 
A. varius from Costa Rica. Starrett (1967) ob- 
served egg deposition in the laboratory and stated 
that the average diameter of the eggs is 1.6 mm. 
She reported that the eggs are uniformly white and 
laid in strings. The longest string contained 23 eggs 
and measured 35 mm; her measurements of the 
eggs and string may have been affected by handling 
and the artificial conditions in the laboratory. I sus- 
pect that the eggs are deposited in continuous 
strings. 
Carvalho (1949) described the eggs and tad- 
poles of Dendrophryniscus brevipollicatus. Ap- 
parently the eggs are relatively large and deposited 
in groups of five or six. The eggs are yellow and 
measure about 2 mm in diameter. 
Two females of Dendrophryniscus brevipollica- 
tus contained 52 and 65 eggs in the ovaries. The 
female with 52 eggs measured 25 mm snout-vent 
length and had 23 eggs in the left ovary and 29 
eggs in the right. Most of the eggs are about equal 
in size; the largest measures 1.6 mm. The second 
female was approximately the same size but 
showed a definite differential maturation of the 
eggs. This frog contained eight large and 30 
smaller eggs on the left and seven large and 20 
smaller eggs on the right. Unfortunately the eggs 
were not measured before the frog was skeleton- 
ized. In all cases the eggs had a black animal pole 
and a light vegetal pole. Carvalho (1949) indi- 
cated that Dendrophryniscus brevipollicatus lays 
yellow eggs and deposits them at intervals. My in- 
formation from one female seems to support his 
field observations concerning the spacing between 
egg deposition. It is difficult to explain the dis- 
crepancy in egg coloration. Perhaps there is a dis- 
persal of dark pigment following fertilization, and 
the resulting zygote, because of the large amount 
of yolk, appears yellow. 
Duellman and Lynch (1969) described the eggs 
and tadpoles of Dendrophryniscus minutus from 
Ecuadorian specimens. Two females laid gela- 
tinous strands of eggs measuring 245 and 285 mm 
and containing 245 and 291 eggs respectively. The 
eggs have a dark brown animal pole. They are 
not partitioned from each other as was described 
for Atelopus (Starrett, 1967). The strands of the 
two females measure 5.0 and 7.5 mm in diameter. 
The eggs measure about 0.8 mm in diameter. 
Two females of D. minutus measuring 18 and 
25 mm snout-vent length contained about 175 and 
420 eggs respectively. AH the eggs are black on 
the animal pole and measure about 0.8 mm. The 
eggs completely filled the body cavity from just 
behind the heart to the cloaca. 
The eggs of D. leucomystax are smaller, more 
numerous than those of D. brevipollicatus (Car- 
valho, 1949), and have a black animal pole (Izeck- 
sohn, 1968). 
A female of Melanophryniscus moreirae (26 
mm snout-vent length) contained 78 eggs in the 
ovaries. The eggs, about 1.7 mm average diameter, 
are black with a light vegetal pole. Starrett (1967) 
reported 70 eggs laid in the laboratory by one indi- 
vidual. Eggs are laid singly but adhere to each 
other in small clumps. She stated that the average 
diameter of the eggs is 2.0 mm. Bokermann (1967) 
examined 20 females and found 33 to 60 eggs 
(x = 48). He suggested that females deposited 
eggs over a period of several days. 
Two female M. stelzneri, 24 and 28 mm snout- 
vent lengths, contained 206 and 237 eggs respec- 
tively. Each egg measures about 1.2 mm in diam- 
eter. Their eggs are black with a light vegetal pole. 
Females of M. stelzneri lay more eggs than female 
M. moreirae of the same size. Apparently this in- 
crease in egg number corresponds to the decrease 
in egg size. Fernandez (1927) reported 201, 351, 
and 91 eggs each from three pair of M. stelzneri. 
Budgett (1899) and Fernandez (1927) reported 
a very rapid rate of development, 48 hours or less, 
from egg to tadpole in M. stelzneri. The eggs of 
Melanophryniscus moreirae hatch in about 100 
hours (Starrett, 1967). The more rapid rate of 
development of M. stelzneri, as compared to M. 
moreirae, may be reflected in the higher number 
of eggs characteristic of M. stelzneri. Less yolk 
material is required for early development of each 
egg, and therefore more eggs of smaller size can 
be produced with the same amount of material and 
can fit in the same space. The eggs of M. rubri- 
ventris also have a dark animal pole (McDiarmid, 
1971). No female specimens of M. tumifrons were 
available for examination. 
Two females of Oreophrynella quelchii were ex- 
amined. The first measured 22 mm snout-vent 
length and contained four eggs in the left ovary 
and six eggs in the right. All the eggs are large, 
uniformly light (probably yellow) and contain 
much yolk. The four eggs on the left average 1.9 
mm in diameter; five of those on the right measure 
about 2.8 mm and a sixth measures 1.4 mm in 
diameter. This difference in egg size suggests that 
36 BULLETIN OF LOS ANGELES COUNTY MUSEUM OF NATURAL HISTORY No. 12 
O. quelchii may lay two clutches of eggs. A sec- 
ond female contained 15 large unpigmented eggs. 
There were six eggs in the left ovary and nine in 
the right. Unfortunately these eggs were not mea- 
sured before the specimen was skeletonized. 
Breeding Biology 
Mating call.? Bogert (1960) classified frog calls 
into several categories and characterized the mat- 
ing call as the vocalization of adult frogs most 
commonly heard in breeding aggregations. Duell- 
man (1967) described the different types of mat- 
ing calls based on the social organization of 
anurans. 
Males of Atelopus varius utter a weak call in the 
field and often in the laboratory. Our lack of 
knowledge concerning the breeding behavior of 
these frogs makes it impossible to categorically 
state that this vocalization is a mating call and 
serves to attract frogs of the same species. Sexton 
(1958) found resident males of A. cruciger calling 
throughout the summer, and yet only a single indi- 
vidual was known to have mated. Starrett (1967) 
mentioned a weak call emitted by a male after it 
separated from a female following amplexus. She 
illustrated a sound spectrogram of a more typical 
call of A. varius. My field experience with Atelo- 
pus leads me to suspect that this call is not a mat- 
ing call but may represent a form of male release 
call or possibly function as a territorial call (see 
Bogert, 1960, for discussion). I heard this type of 
call from a male Atelopus varius that was several 
hundred meters from the nearest breeding site. 
Some forms of Atelopus emit a soft "chirp" when 
they come into contact with each other in collect- 
ing sacks (Campbell, personal communication); 
others apparently do not. Dunn (1933) also re- 
ported this chirping call. 
I suspect that the effectiveness of a weak mating 
call for a frog that breeds along fast moving, noisy 
streams is limited, especially since Atelopus varius 
lacks a tympanum and middle ear. However, only 
extensive field work using tape recordings of male 
calls to determine their effect on other males as 
well as on females will clarify the nature of this 
supposed mating call. 
Lutz (1932) briefly reported on the breeding 
biology of Dendrophryniscus brevipollicatus. Car- 
valho (1949) discussed the breeding habits of this 
species in more detail and described the call. Ap- 
parently this frog spends most of its time in bro- 
meliads on trees and emits a weak "tic... tic", 
presumably its mating call. Nothing is known con- 
cerning the mating call of D. minutus. Izecksohn 
(1968) described the call of D. leucomystax as 
consisting of weak peeps. 
Much more information is available concerning 
the mating calls of species of Melanophryniscus. 
Budgett (1899) described the mating call of Mela- 
nophryniscus stelzneri from the Paraguayan Chaco 
as consisting of two clear musical "pings," fol- 
lowed by a long descending "trill." He mentioned 
that both the males and females have a call. Fer- 
nandez (1927) reported mating calls of M. stelz- 
neri from the Sierra de Cordoba, Argentina. 
Barrio (1964) described the mating call of Mela- 
nophryniscus stelzneri by means of sound specto- 
grams of tape recorded calls. Starrett (1967) and 
Bokermann (1967) reported mating calls for M. 
moreirae. These reports, together with information 
obtained from Werner C. A. Bokermann (personal 
communication) and Jay M. Savage (personal 
communication) indicate that mating calls are 
characteristic of and important in the breeding 
biology of Melanophryniscus. 
Nothing is known about the mating calls, breed- 
ing habits, or behavior of Oreophrynella. 
Breeding behavior?All references to breeding 
behavior in Atelopus are based on the reports by 
Starrett (1967), Duellman and Lynch (1969), my 
own field experience with A telopus varius in Costa 
Rica, and the report by Sexton (1958) on the life 
history of Atelopus cruciger in Venezuela. As far 
as is known, Atelopus utilizes streamside habitats 
for breeding. Apparently males spend most of their 
time along the streams while the females are often 
encountered on the hillsides and some distance 
from the streams. The breeding season apparently 
extends throughout the year as amplexing pairs 
have been collected in March, July, August, and 
November. Whether females move down to the 
streams in response to some environmental cue, 
such as rainfall, or whether they are attracted by 
the vocalization of males, is unknown. Whatever 
the reason, the frogs locate each other, perhaps by 
visual recognition, and pair by means of axillary 
amplexus. The pairs remain in amplexus for a long 
period, a minimum of 19 days in the field (Sexton, 
1958) and a minimum of 20 days in the labora- 
tory (Starrett, 1967), during which time the fe- 
male carries the male on her back. Apparently, the 
female searches for a place to deposit her eggs. 
During these movements, the pair seldom moves 
more than a meter from the stream and often is 
found sitting on rocks along the edge or in the 
shallows. Starrett (1967) suggested that these frogs 
probably deposit their eggs beneath or around 
rocks in the stream. I suspect that the eggs are 
attached to rocks or other objects in the stream 
to avoid being washed away. The eggs hatch in 
about six days. Tadpoles of A. varius have been 
taken in March, June, and September in Costa 
1971 MORPHOLOGY AND EVOLUTION OF NEOTROPICAL FROGS 37 
Rica; those of A. certus in May in Panama; those 
of A. ignescens in Ecuador in August; and those of 
A. spumarius in July in Ecuador. Recently meta- 
morphosed specimens of Atelopus varius have been 
collected in Costa Rica in August and October. 
Specimens of A. ignescens and A. spurrelli at a 
similar stage were taken in Colombia in July and 
April respectively. 
Dendrophryniscus brevipollicatus spends most 
of its time in bromeliads. Apparently males call 
from these sites during the rainy parts of the year. 
The eggs are deposited at intervals in small clumps 
in the axils of the plant. The tadpoles spend their 
entire lives in the bromeliads (Carvalho, 1949). 
Melin (1941) reported an amplexing pair of 
Dendrophryniscus minutus in February or March 
in Brazil. Duellman and Lynch (1969) collected 
two amplexing pairs of D. minutus at the edge of 
a quiet, murky swamp at the edge of the forest in 
July in Ecuador. 
The species of Melanophryniscus apparently 
breed in temporary, rain filled ponds. Melanophry- 
niscus stelzneri breeds in January in Argentina; 
M. moreirae breeds in October and November in 
Brazil. Budgett (1899) stated that the eggs of 
stelzneri are laid in separate globules of jelly which 
float on the water. Fernandez (1927) and Barrio 
(1964) indicated that breeding commences in 
ponds formed by heavy rains. Similar breeding be- 
havior was described for M. stelzneri by Fernan- 
dez (1927), and by Starrett (1967) and Boker- 
mann (1967) for M. moreirae. Females of both 
species deposit eggs singly or in small groups. 
Amplexus is axillary in Atelopus, Dendrophry- 
niscus, and Melanophryniscus. 
Larvae 
Goin (1960) indicated that all members of the 
family Atelopodidae have indirect development, 
that is, they lay eggs in water and pass through 
a tadpole stage before transformation. This has 
been confirmed for Atelopus, Dendrophryniscus, 
and Melanophryniscus. Starrett (1967) described 
and illustrated the tadpoles of Atelopus varius and 
Duellman and Lynch (1969) described and 
illustrated the larvae of A. certus, A. ignescens, 
and A. spumarius. The larva of Dendrophryniscus 
brevipollicatus was described and illustrated by 
Carvalho (1949); that of D. minutus was de- 
scribed and illustrated by Duellman and Lynch 
(1969). 
The larva of Melanophryniscus stelzneri was 
described and illustrated by Fernandez (1927). 
Ahl (1938) and later Starrett (1967) and Boker- 
mann (1967) described and illustrated the larvae 
of M. moreirae. The characteristics of larvae of 
known species are summarized in Table 1. 
Starrett (1967) and Duellman and Lynch (1969) 
discussed the similarities between the tadpoles of 
Atelopus, Melanophryniscus, Dendrophryniscus, 
and Bufo. They indicated their close relationships 
with other forms (e. g. Ansonia) of the family 
Bufonidae. All have the typical 2/3 tooth row pat- 
tern and a sinistral spiracle. The anus is median 
in Bufo and Atelopus; it is dextral in Melanophry- 
niscus and Dendrophryniscus. The lateral margins 
of the mouth have an indented row of papillae in 
Bufo and Melanophryniscus stelzneri; in M. mo- 
reirae the papillae are restricted to the lateral mar- 
gins but do not exhibit the indentation. Dendro- 
phryniscus brevipollicatus has nearly a complete 
TABLE 1 
Summary of Larval Characteristics 
Spiracle Anal Tube Tooth Rows 
Labial 
Papillae Mouth 
Atelopus1 single, 
sinistral 
median % single row, 
interrupted on 
lower lip 
large, ventral, 
large suctorial 
disc 
Dendrophryniscus2- single, 
sinistral 
dextral % single row, 
interrupted on 
upper lip; or 
single row, 
lateral 
small, antero- 
ventral, 
no suctorial disc 
Melanophryniscus3 single, 
sinistral 
dextral % single row, 
lateral 
moderate, 
antero-ventral, 
no suctorial disc 
1
 ? Atelopus certus, A. ignescens, A. spumarius, A. varius 
2
 = Dendrophryniscus brevipollicatus, D. minutus 
3
 = Melanophryniscus moreirae, M. stelzneri 
38 BULLETIN OF LOS ANGELES COUNTY MUSEUM OF NATURAL HISTORY No. 12 
row of papillae with a slight anterior interruption; 
in D. minutus the papillae are restricted to the lat- 
eral margins. Larvae of Dendrophryniscus do not 
have lateral indentations. Larvae of Atelopus have 
a row of papillae complete except posteriorly where 
the large ventral sucking disc extends well back 
onto the body. There is no lateral indentation in the 
row of papillae in Atelopus larvae. 
ECOLOGY AND DISTRIBUTION 
The majority of species of Atelopus are found 
from Costa Rica southward through Panama, 
along the Andean parts of South America to Bo- 
livia, and in the mountainous areas of northern 
Venezuela (Fig. 9). Two species occur at inter- 
mediate elevations in the Guiana area and adjacent 
parts of northeastern Brazil. Another species has 
been recorded from the mountainous areas of Per- 
nambuco, Brazil. Apparently species of Atelopus 
generally are absent from the basins of the Ama- 
zon and Orinoco rivers. Recently, A. flavescens 
was collected southeast of Santarem, Para, Brazil 
(Martha Crump, personal communication). 
Species of Atelopus occupy a wide range of hab- 
itats. Some species occur near sea level in the wet 
lowland tropical forests of the Darien in Panama 
and the Choco in Colombia. Others are known 
from the cold, unforested paramos above 3000 
meters in the Ecuadorean Andes and above 4000 
meters in the Santa Marta Massif in northwestern 
Colombia (Rivero, 1963). 
As now understood, all species of Atelopus are 
diurnal, and most are found along streams (Ruth- 
ven, 1922; Dunn, 1933; Sexton, 1958; Starrett, 
1967). Atelopus appears to be primarily terrestrial 
although individuals are capable of climbing. Dunn 
(1933) and Taylor (1952) reported individuals on 
leaves at night. In Costa Rica I have collected frogs 
at night on leaves and branches up to a meter 
above the ground. Occasionally frogs are encoun- 
tered during the day climbing around on rocks and 
logs near the streams, but rarely are they found 
on leaves during the day. Sexton (1958) made 
some interesting observations on the population 
structure and ecology of Atelopus cruciger in 
Venezuela. 
Dendrophryniscus brevipollicatus occurs in the 
mountainous regions of southeastern Brazil in the 
states of Rio de Janeiro and Sao Paulo and in 
the Districto Federal. This species is arboreal and 
spends much of its time in bromeliads, where its 
eggs and larvae have been found. Dendrophrynis- 
cus leucomystax, also from the state of Rio de 
Janeiro, Brazil, has been collected on ginger leaves 
around the margin of forest marshes. Two speci- 
mens were found beneath fallen tree trunks. None 
were taken in bromeliads (Izecksohn, 1968). Den- 
drophryniscus minutus is known from scattered 
localities in the upper Amazon Basin in Peru, 
Ecuador, Brazil, and Guyana. It is a diurnal spe- 
cies that is commonly encountered among dry 
leaves on the forest floor (Melin, 1941; Lutz and 
Kloss, 1952) or around quiet murky swamps at 
the edge of tropical rainforest in Ecuador (Duell- 
man and Lynch, 1969). Atelopus proboscideus, 
probably referable to this genus, is known only 
from the type locality at Bahia, Brazil (Fig. 9). 
There are six species of Melanophryniscus in 
South America (Fig. 10). A disjunct population of 
M. moreirae occurs near Obidos, Para, Brazil 
(Cochran, 1948). The next closest records for the 
genus are from the state of Rio de Janeiro, Brazil. 
From here it ranges south and eastward into Para- 
guay, Uruguay and across the Chaco of Argentina 
to the southeastern slopes of the Andes in Jujuy, 
Salta, and La Rioja Provinces of Argentina (Gal- 
lardo, 1961a). Bokermann (1966) suggested that 
the type locality, "Pernambuco", for M. tumifrons 
was probably incorrect. I concur, since the species 
has not been recorded from the area subsequently. 
The species of Melanophryniscus are diurnal and 
found from the low dry regions of the Argentinian 
Chaco and from patches of savanna in the Amazon 
Basin to the high, cold slopes of Mount Itatiaia, 
Rio de Janeiro, Brazil at 2400 meters. Apparently 
the species are most abundant following heavy 
rains. 
Oreophrynella is restricted to the immediate vi- 
cinity of Mount Roraima situated on the Guyanan, 
Venezuelan, and Brazilian borders (Fig. 10). Spe- 
cimens of O. quelchii are known from the summit 
above 2400 meters. The type and only known 
specimen of O. macconnelli was taken from the 
base of Mt. Roraima at about 1050 meters. Speci- 
mens of both forms were collected in September 
and October (Lankester, 1900). 
SYSTEMATICS 
GENERIC DEFINITIONS 
In the preceding section the characters and their 
various states in the four genera and included 
species were discussed in detail. In this section only 
those characters and their associated states that are 
diagnostic and significantly contribute to an under- 
standing of the evolutionary relationships among 
the four genera and that are considered useful in 
elucidating the familial status of the Atelopodidae 
will be mentioned. Each character is described un- 
der the same number to facilitate comparisons 
among the four genera. 
Species for which specimens have been exam- 
ined in detail are included under each genus. Spe- 
1971 MORPHOLOGY AND EVOLUTION OF NEOTROPICAL FROGS 39 
ties for which specimens were not available for 
detailed examination are listed as referred forms 
based on their descriptions and available data. The 
general geographic range of each genus is briefly 
outlined. 
ATELOPUS Dumeril and Bibron 1841 
Definition: 1) tensor fasciae latae muscle strap- 
like and elongate, originating on anterior third of 
ilium and inserting on proximal third of cruralis 
muscle; 2) semitendinosus muscle a thin slip visible 
two-thirds of its length, inserting via a long tendon; 
3) adductor longus muscle absent; 4) depressor 
mandibulae muscle present in SQ or SQat condi- 
tion; 5) adductor mandibulae muscle present in S 
condition; 6) frontoparietal bone of normal ex- 
tent, separate or fused on midline; 7) occipital 
groove closed; 8) leading edge of premaxillary 
platform twice as wide as medial edge; 9) septo- 
maxilla U-shaped with trough; 10) palatines pres- 
ent; 11) vomers present; 12) parasphenoid pointed 
anteriorly, overlapping sphenethmoid complex 
medially; 13) squamosal stem rotated, anterior 
process usually present, dorsal arm extending over 
prootic; 14) quadratojugal well developed or re- 
duced; 15) pterygoid well developed and of normal 
extent; 16) sphenethmoid complex ossified; 17) 
orbitosphenoid unossified; 18) quadrate cartilagin- 
ous, restricted to lower part of suspensorium; 19) 
prearticular curve in mandible posterior to mid- 
point, Meckel's cartilage unossified; 20) anterior 
processes of ceratohyal very small and directed 
laterally; 21) hypobranchial 1 present with narrow 
base, hypobranchial 2 absent; 22) constrictor laryn- 
gis posterior muscle absent; 23) seven presacral 
vertebrae; 24) atlas fused with first trunk vertebra; 
25) sacrum usually separate from trunk vertebra; 
26) coccyx with bicondylar articulation or fused 
with sacrum; 27) sacral diapophyses greatly ex- 
panded, elliptically flattened in cross section; 28) 
pectoral girdle firmisternal; 29) sternum present; 
30) ilia, sacral diapophyses articulate in a flat joint; 
31) one centrale in hand; 32) two distal carpals; 
33) phalangeal formula of hand, 2-2-3-3 or 1-2-3-3; 
34) phalangeal formula of foot, 2-2-3-4-3; 35) 
middle ear apparatus absent, except in A. ftav- 
escens, A. spumarius and an undescribed species; 
36) ostia pharyngea present or absent; 37) vocal 
slits present or absent; 38) habitus atelopodiform, 
body length greater than three times body width; 
39) skin smooth or with rounded warts or coni 
apicali; 40) Bidder's organ present; 41) testicular 
peritoneum black or unpigmented; 42) eggs white, 
350-910 total counts; 43) larvae aquatic. 
Range:   tropical areas of central Costa  Rica 
southward through Panama, Colombia, Ecuador, 
into Peru and Bolivia and northeastward through 
northern Venezuela; disjunct populations in Guy- 
ana, Surinam, French Guiana, and northeastern 
Brazil (Fig. 9). 
Content: boulengeri Peracca 1904; carrikeri 
Ruthven 1916; certus Barbour, 1923; chiriquiensis 
Shreve 1936; cruciger (Lichtenstein and von Mar- 
tens 1856); ebenoides Rivero 1963; elegans (Bou- 
lenger 1882); exigua Boettger 1892; flavescens 
Dumeril and Bibron 1841; glyphus Dunn 1931; 
ignescens (Cornalia 1849); longirostris Cope 1868; 
oxyrhynchus Boulenger 1903; pachydermus 
(Schmidt 1857); senex Taylor 1952; spumarius 
Cope 1871; spurrelli Boulenger 1914; varius (Lich- 
tenstein and von Martens 1856); walkeri Rivero 
1963; zeteki Dunn 1933. 
Referred forms: bibronii (Schmidt 1857); bi- 
color Noble 1921; bufoniformis Peracca 1904; 
carinatus Andersson 1945; erythropus Boulenger 
1903; longibrachius Rivero 1963; nicefori Rivero 
1963; palmatus Andersson 1945; pedimarmoratus 
Rivero 1963; pernambucensis Bokermann 1962; 
planispinus Jimenez de la Espada 1875; rugulosus 
Noble 1921; seminiferus Cope 1874; tricolor Bou- 
lenger 1902; willimani Donoso Barros 1969. 
DENDROPHRYNISCUS Jimenez de la Espada 
1870 
Definition: 1) tensor fasciae latae muscle short 
and broad, originating on middle third of ilium 
and inserting on distal third of cruralis muscle; 2) 
semitendinosus muscle a thin slip not visible along 
its length, inserting via a long tendon; 3) adductor 
longus muscle absent; 4) depressor mandibulae 
muscle present in SQ or SQm condition; 5) adduc- 
tor mandibulae muscle present in S condition; 6) 
frontoparietal bone of normal extent, separate 
along midline except in largest adults, fused in 
posterior one-quarter; 7) occipital groove open; 
8) leading edge of premaxillary platform twice as 
wide as medial edge; 9) septomaxilla U-shaped 
with trough; 10) palatines present or absent; 11) 
vomers present; 12) parasphenoid pointed anteri- 
orly, overlapping sphenethmoid complex medially; 
13) squamosal stem slightly rotated, anterior 
process well developed, degree dorsal arm extends 
over prootic variable; 14) quadratojugal reduced; 
15) pterygoid well developed and of normal ex- 
tent; 16) sphenethmoid complex ossified; 17) or- 
bitosphenoid unossified; 18) quadrate cartilagin- 
ous, restricted to lower part of suspensorium; 19) 
prearticular curve in mandible posterior to mid- 
point, Meckel's cartilage unossified; 20) anterior 
processes of ceratohyal absent; 21) hypobranchial 
1 present with narrow base, hypobranchial 2 ab- 
sent; 22) constrictor laryngis posterior muscle ab- 
40 BULLETIN OF LOS ANGELES COUNTY MUSEUM OF NATURAL HISTORY No. 12 
sent; 23) eight presacral vertebrae; 24) atlas not 
fused with first trunk vertebra; 25) sacrum usually 
separate from trunk vertebra; 26) coccyx fused 
with sacrum; 27) sacral diapophyses greatly ex- 
panded, elliptically flattened in cross section; 28) 
pectoral girdle firmisternal anteriorly, arciferal 
posteriorly; 29) sternum present; 30) ilia, sacral 
diapophyses articulate in a flat joint; 31) one cen- 
trale in hand; 32) two distal carpals; 33) pha- 
langeal formula of hand 2-2-3-3 or 1-2-3-3; 34) 
phalangeal formula of foot, 2-2-3-4-3; 35) middle 
ear apparatus absent; 36) ostia pharyngea absent; 
37) vocal slits absent; 38) habitus atelopodiform, 
body length greater than two and one-half times 
width; 39) skin minutely glandular with small scat- 
tered warts; 40) Bidder's organ present; 41) testic- 
ular peritoneum black; 42) eggs with black animal 
pole and light vegetal pole, 52-64 total counts for 
brevipollicatus, 175-420 total counts for minutus; 
43) larvae aquatic. 
Range: tropical lowland areas of the Amazon 
Basin in Brazil, Peru, Ecuador, Surinam, French 
Guiana, Guyana, and mountainous areas of south- 
eastern Brazil (Fig. 9). 
Content: brevipollicatus Jimenez de la Espada 
1870; leucomystax Izecksohn 1968; minutus (Melin 
1941) by present arrangement. 
Referred forms: proboscideus (Boulenger 1882) 
by present arrangement. 
MELANOPHRYNISCUS Gallardo 1961 
Definition: 1) tensor fasciae latae muscle short 
and broad, originating on middle third of ilium and 
inserting on distal half of cruralis muscle; 2) semi- 
ATELOPUS 
OREOPHRYNELLA 
FIGURE 9. Map depicting the known ranges of species of Atelopus and Oreophrynella. 
1971 MORPHOLOGY AND EVOLUTION OF NEOTROPICAL FROGS 41 
tendinosus muscle a thin slip visible two-thirds of 
its length, inserting via a long tendon; 3) adductor 
longus muscle absent; 4) depressor mandibulae 
muscle present in SQ condition; 5) adductor man- 
dibulae muscle present in S condition; 6) fronto- 
parietal bone of normal extent, anterior half fused 
or separate on midline, posterior half always fused; 
7) occipital groove open except posteriorly; 8) 
leading edge of premaxillary platform twice as 
wide as medial edge; 9) septomaxilla U-shaped 
with trough; 10) palatines present or absent; 11) 
vomers present; 12) parasphenoid pointed anteri- 
orly, overlapping sphenethmoid complex medially; 
13) squamosal stem rotated in three species, an- 
terior process lacking, dorsal arm not extending 
over prootic; 14) quadratojugal reduced or absent; 
15) pterygoid well developed and of normal extent; 
16) sphenethmoid complex ossified; 17) orbito- 
sphenoid ossified; 18) quadrate cartilaginous, re- 
stricted to lower part of suspensorium; 19) pre- 
articular curve in mandible posterior to midpoint, 
Meckel's cartilage unossified; 20) anterior processes 
of ceratohyal absent; 21) hypobranchial 1 present 
with wide base, hypobranchial 2 absent; 22) con- 
strictor laryngis posterior muscle absent; 23) eight 
presacral vertebrae; 24) atlas not fused with first 
trunk vertebra; 25) sacrum usually separate from 
trunk vertebra; 26) coccyx with bicondylar articu- 
lation with sacrum; 27) sacral diapophyses greatly 
expanded, elliptically flattened in cross section; 28) 
pectoral girdle firmisternal anteriorly, arciferal pos- 
teriorly; 29) sternum present; 30) ilia, sacral dia- 
pophyses articulate in a flat joint; 31) two centralia 
in hand; 32) two distal carpals; 33) phalangeal for- 
mula of hand 2-2-3-3; 34) phalangeal formula of 
foot 2-2-3-4-3; 35) middle ear apparatus absent; 
36) ostia pharyngea present; 37) vocal slits pres- 
ent; 38) habitus bufoniform, body length never ex- 
ceeds three times width; 39) skin rugose, covered 
with rounded warts or coni apicali; 40) Bidder's 
organ present; 41) testicular peritoneum black or 
unpigmented; 42) eggs with black animal pole and 
light vegetal pole, 78-237 total counts; 43) larvae 
aquatic. 
Range: southern temperate and tropical areas 
of Argentina, Paraguay, Uruguay, and southeast- 
ern Brazil; an allopatric population of one species 
described from the Amazon Basin in Para, Brazil 
(Fig. 10). 
Content: moreirae (Miranda-Ribeiro 1920); 
rubriventris (Vellard 1947) by present arrange- 
ment; stelzneri (Weyenbergh 1875); tumifrons 
(Boulenger 1905). 
Referred forms: devincenzii Klappenbach 1968; 
sanmartini Klappenbach 1968. 
OREOPHRYNELLA Boulenger 1895 
Definition: 1) tensor fasciae latae muscle short 
and broad, originating on middle third of ilium 
and inserting midway along cruralis muscle; 2) 
semitendinosus muscle a thin slip visible two-thirds 
of its length, inserting via a long tendon; 3) ad- 
ductor longus muscle absent; 4) depressor man- 
dibulae muscle present in SQ condtion; 5) adduc- 
tor mandibulae muscle present in E condition; 6) 
frontoparietal bone greatly reduced anteriorly ex- 
posing frontal fontanelle, fused in midline pos- 
terior to parietal fontanelles; 7) occipital groove 
open; 8) leading edge of premaxillary platform 
twice as wide as medial edge; 9) septomaxilla 
U-shaped with trough; 10) palatines present; 11) 
vomers present; 12) parasphenoid pointed anteri- 
orly, overlapping sphenethmoid complex medi- 
ally; 13) squamosal stem not rotated, large anterior 
process, dorsal arm extending over prootic; 14) 
quadratojugal absent; 15) pterygoid well devel- 
oped and of normal extent; 16) ethmoid portion 
only of sphenethmoid complex ossified; 17) or- 
bitosphenoid unossified; 18) quadrate cartilagin- 
ous, restricted to lower part of suspensorium; 19) 
prearticular curve in mandible posterior to mid- 
point, Meckel's cartilage unossified; 20) anterior 
processes of ceratohyal absent; 21) hypobranchial 
1 present with narrow base, hypobranchial 2 ab- 
sent; 22) constrictor laryngis posterior muscle ab- 
sent; 23) five presacral vertebrae; 24) atlas fused 
with first trunk vertebra; 25) sacrum fused with 
trunk vertebrae; 26) coccyx fused with sacrum; 
27) sacral diapophyses greatly expanded, ellipti- 
cally flattened in cross section; 28) pectoral girdle 
firmisternal anteriorly, arciferal posteriorly; 29) 
sternum present; 30) ilia, sacral diapophyses ar- 
ticulate in a flat joint; 31) two centralia in hand; 
32) two distal carpals; 33) phalangeal formula 
of hand, 2-2-3-3; 34) phalangeal formula of foot, 
2-2-3-4-3; 35) middle ear apparatus absent; 36) 
ostia pharyngea vestigial; 37.) vocal slits absent; 
38) habitus bufoniform, body length never exceeds 
three times width; 39) skin rugose, covered with 
rounded warts; 40) Bidder's organ present; 41) 
testicular peritoneum unpigmented; 42) eggs yel- 
low, 10-15 total counts; 43) larvae probably 
aquatic. 
Range: restricted to the immediate vicinity of 
Mount Roraima on the Guayanan, Venezuelan, 
and Brazilian borders (Fig. 10). 
Content: quelchii (Boulenger 1895a). 
Referred form: macconnelli Boulenger 1900. 
42 BULLETIN OF LOS ANGELES COUNTY MUSEUM OF NATURAL HISTORY No. 12 
DENDROPHRYNISCUS 
MELANOPHRYNISCUS BUtl! 
FIGURE  10.   Map depicting the known ranges of species of Dendrophryniscus and Melano- 
phryniscus. 
1971 MORPHOLOGY AND EVOLUTION OF NEOTROPICAL FROGS 43 
EVOLUTION 
At the outset of this project several basic as- 
sumptions were deemed essential to completion of 
the study. One of these assumptions was that the 
genera included in the Atelopodidae by Noble 
(1922, 1926) and Davis (1935) were a natural 
group. Even though Griffiths (1959) removed 
Oreophrynella and Dendrophryniscus (=Mela- 
nophryniscus) from the Atelopodidae and put 
them in Bufonidae, I thought that their inclusion 
in this study was necessary. Griffiths shifted these 
two genera from one family to another on the basis 
of the nature of their epicoracoidal fusion and 
arbitrarily defined the family Atelopodidae. In ad- 
dition, he did not examine Dendrophryniscus 
brevipollicatus; therefore the position of this genus 
in his scheme is unknown. 
EVALUATION OF CHARACTER STATES 
To determine the evolutionary patterns among 
the four genera, the characters utilized in the 
generic definitions must be evaluated as to their 
primitive (ancestral) or advanced (derived) states. 
This kind of analysis assumes that those taxa hav- 
ing the greater numbers of primitive character 
states are nearer the ancestral stock than are those 
with a small number of primitive character states. 
The evolutionary changes in a character from a 
primitive to an advanced state have been evaluated 
in accord with the following postulates: 
1) character states of modern species which also 
are found in fossils of presumed ancestors or 
of forms considered closely related are consid- 
ered primitive; 
2) character states based on structures that show 
a reduction in number of parts or a simplifica- 
tion of existing parts are considered advance- 
ments. This postulate is referred to as Willis- 
ton's Rule. It states that the number of parts 
of an organism tends to be reduced, with the 
fewer parts greatly specialized in function 
(Simpson, 1949, 1953; Gregory, 1951). This 
postulate does not assume irreversibility of 
characters; 
3) character states that are universal or frequent 
in other modern groups or widespread within 
the taxonomic group being studied are con- 
sidered to be primitive. This postulate may be 
derived from the Rule of Parsimony (Kluge, 
1967; Kluge and Farris, 1969) and has been 
discussed by Throckmorton (1968). 
INTERGENERIC RELATIONSHIPS 
Several characters are the same among the four 
genera and are used in the generic definitions only 
to elucidate the differences between these four gen- 
era and other procoelous Anura (e.g. Brachycepha- 
lus). These characters are not discussed in this 
section unless they are important to the later treat- 
ment of familial relationships. To facilitate com- 
parison among the genera, each character is 
numbered as it was in the generic definitions. Each 
state is designated primitive or advanced accord- 
ing to one of the above postulates. Where neces- 
sary, a brief discussion is presented to support the 
designation. It is impossible to designate certain 
character states as either primitive (P) or ad- 
vanced (A); these are not rated but simply num- 
bered. 
1) tensor  fasciae   latae  muscle?two  character 
states: 
P. short, origin on middle third of ilia?primi- 
tive. 
A. elongate, origin on anterior third of ilia- 
advanced. 
Dunlap (1960) and Tihen (1960) indicated 
that the elongate tensor condition is only 
found in four other genera of frogs. An 
anterior origin occurs also in the bufonids 
Ansonia and Mertensophryne and in the lep- 
todactylids Limnodynastes and Eleutherodac- 
tylus. I suggest that the elongation of this 
muscle in Atelopus reflects a change in pat- 
terns of locomotion from hopping to walk- 
ing. Since most frogs exhibit a short condition, 
and because all evidence indicates that the 
earliest known frogs were "hoppers", I have 
followed the Rule of Parsimony in determin- 
ing the evolutionary stage. 
2) semitendinosus muscle?two character states: 
1. visible two-thirds of length, inserting via 
long tendon. 
2. invisible, inserting via long tendon. 
3) adductor longus muscle?one character state: 
A. absent?advanced. 
Inger (1967) pointed out that there has been 
a progressive trend towards fragmentation 
and specialization of musculature in the series 
from fish to mammal. Tihen (1965) also sug- 
gested that there has been a general trend to- 
wards increasing complexity of musculature. 
Noble (1922) indicated that the adductor 
longus muscle is a separate slip of the pec- 
tineous muscle. He supported the undifferenti- 
ated nature (absence) of the muscle as the 
primitive condition. However, Kluge and Far- 
ris (1969) pointed out that absence of the 
adductor longus muscle in some primitive 
frogs might be the result of a secondary loss 
44 BULLETIN OF LOS ANGELES COUNTY MUSEUM OF NATURAL HISTORY No. 12 
to increase locomotor efficiency. Tihen 
(1960) argued that the muscle probably was 
present in early salientians and has been lost 
in some forms. Until more information is 
available and in accord with Williston's Rule, 
I consider the absence of the adductor longus 
muscle in the forms studied an advanced state. 
4) depressor mandibulae?three character states: 
P. SQat?primitive. 
A2. SQ?advanced. 
A3. SQm?advanced. 
The attachment of a slip of the depressor 
mandibulae muscle to the annulus tympanicus 
must be considered primitive. The presence 
of an annulus tympanicus is associated with 
the presence of a middle ear, which, as indi- 
cated by the fossil record, is a primitive con- 
dition. When the middle ear is lost, the an- 
nulus tympanicus and its associated depressor 
slip are also lost. Starrett (1968) has shown 
that the SQ condition has been independently 
derived in several families. Postulates one and 
two are applicable. 
5) adductor mandibulae muscle?two character 
states: 
A,. S?advanced. 
A2. E?advanced. 
The S and E conditions are independent de- 
rivatives of an S and E ancestral condition 
(Starrett, 1968). Once a muscle has been 
lost, it is not reacquired (Tihen, 1965 and 
Williston's Rule). 
6) frontoparietal bone?two character states: 
P.   normal size, separate or fused on midline 
?primitive. 
A. greatly reduced exposing frontal and pa- 
rietal fontanelles?advanced. 
The advanced condition is attributed to ar- 
rested development. The primitive condition 
is implied from postulates one and three. 
7) occipital groove?three character states: 
1. open. 
2. open except posteriorly. 
3. closed. 
10) palatine bone?two character states: 
P. present?primitive. 
A. absent?advanced. 
The primitive condition was designated ac- 
cording to Williston's Rule. 
13) squamosal?three character states: 
1. stem rotated, anterior process present, dor- 
sal arm over prootic. 
2. stem rotated, no anterior process, no dor- 
sal overlap of prootic. 
3. stem not rotated, anterior process present, 
dorsal arm over prootic. 
14)  quadratojugal?three character states: 
P-t. well developed?primitive. 
P2. present but reduced?primitive. 
A. absent?advanced. 
The primitive states were designated accord- 
ing to Williston's Rule. It was impossible to 
determine whether the well developed state or 
the reduced state was ancestral; therefore 
both were designated primitive. 
16) sphenethmoid complex?two character states: 
P.   only ethmoid portion ossified?primitive. 
A. completely ossified?advanced. 
Designation of the primitive state was based 
on the Rule of Parsimony. 
17) orbitosphenoid?two character states: 
P.   unossified?primitive. 
A. ossified?advanced. 
The primitive state was determined according 
to the Rule of Parsimony. 
20) anterior process of ceratohyal?two character 
states: 
P.   absent?primitive. 
A. small and directed laterally?advanced. 
Since none of the bufonid (closely related 
modern group) genera have an anterior pro- 
cess of the ceratohyal (Trewavas, 1933), the 
Rule of Parsimony would indicate that the 
absence of an anterior process is primitive. In 
addition, in both Melanophryniscus and Den- 
drophryniscus, the anterior part of the cera- 
tohyal is expanded; I consider the process in 
Atelopus to be the culmination of a trend 
toward an increase in the cartilaginous extent 
of this part of the ceratohyal. This treatment 
follows that part of Williston's Rule referring 
to specialization of parts. 
21) hypobranchial 1?two character states: 
1. narrow base. 
2. wide base. 
hypobranchial 2?one character state: 
A. absent?advanced. 
The advanced state of hypobranchial 2 was 
designated according to Williston's Rule and 
the Rule of Parsimony. 
23)  number of presacral vertebrae?three charac- 
ter states: 
P.   eight?primitive. 
A2. seven?advanced. 
A3. five?advanced. 
Griffiths (1963) indicated that higher num- 
bers of presacral vertebrae are primitive. 
Tihen (1965) pointed out that the number of 
presacral vertebrae may decrease but does not 
1971 MORPHOLOGY AND EVOLUTION OF NEOTROPICAL FROGS 45 
increase. Both Inger (1967) and Kluge and 
Farris (1969) agreed that high numbers of 
presacral vertebrae in frogs appear to be more 
primitive. Fossil evidence also suggests that 
higher numbers are primitive. Postulates one, 
two, and three are applicable. 
24) atlas?two character states: 
P. separate from first trunk vertebra?primi- 
tive. 
A. fused to first trunk vertebra?advanced. 
Bones, once fused, rarely become independent 
elements again (Tihen, 1965). This is in ac- 
cord with Williston's Rule and also with the 
Rule of Parsimony. 
25) sacrum?two character states: 
P.   separate from trunk vertebrae?primitive. 
A. fused to trunk vertebrae?advanced. 
The reasons given under character 24 also 
apply here. 
26) coccyx?two character states: 
P. articulates with sacrum?primitive. 
A. fused to sacrum?advanced. 
The reasons given under character 24 also 
apply here. 
28) pectoral girdle?two character states: 
Ax. anteriorly firmisternal, posteriorly arcif- 
eral?advanced. 
A2. entirely firmisternal?advanced. 
The arciferal condition occurs in most Bufoni- 
dae, in primitive amphibians and salamanders 
and in the developmental stages of firmisternal 
forms. Therefore, it must be primitive. The 
arcifero-firmisternal and firmisternal girdles 
represent stages in an evolutionary continuum 
from an overlapping to a fused condition. 
Postulates one and three are applicable here. 
31) centralia of hand?two character states: 
P.   two?primitive. 
A. one?advanced. 
The primitive state was determined according 
to Williston's Rule. 
33) phalangeal formula of hand?two character 
states: 
P.  2-2-3-3-primitive. 
A.  1-2-3-3?advanced. 
The primitive state was determined according 
to Williston's Rule. 
35) middle ear apparatus?two character states: 
P.   internal tympanum and middle ear present 
?primitive. 
A. absent?advanced. 
The fossil record clearly indicates that a 
middle ear was present in the primitive frog. 
Tihen (1965) also suggested that the middle 
ear was originally present, and once lost is not 
reacquired. 
36) ostia pharyngea?three character states: 
P.   present?primitive. 
A2. vestigial?advanced. 
A3. absent?advanced. 
The ostia pharyngea are associated with hear- 
ing and vocalization. When the middle ear 
degenerates, the ostia pharyngea also degen- 
erate. Their absence is, therefore, a derived 
condition. 
37) vocal slits?two character states: 
P.   present?primitive. 
A. absent?advanced. 
The presence of vocal slits is frequently but 
not invariably associated with the presence of 
a middle ear and ostia pharyngea. Once the 
ability to hear is lost, there is a degeneration 
of the vocal apparatus. As with characters 35 
and 36, the absence of vocal slits is derived. 
38) habitus?two character states: 
1. atelopodiform. 
2. bufoniform. 
39) skin texture?four character states: 
1. smooth. 
2. scattered warts. 
3. minutely glandular with small warts. 
4. rugose and covered with warts. 
40) Bidder's organ?one character state: 
A. present?advanced. 
The presence of a Bidder's organ only in the 
Bufonidae and the atelopodid genera indi- 
cates that it is an advanced character accord- 
ing to the Rule of Parsimony. 
41) color of testicular peritoneum?two character 
states: 
1. unpigmented. 
2. black. 
42) egg color and number?three character states: 
Pj.. unpigmented, 350 to 910 eggs?primitive. 
P2. black animal pole, light vegetal pole, 52 to 
420 eggs?primitive. 
A. unpigmented, ten to  15 eggs?advanced. 
Tihen (1965) suggested that numerous, fully 
aquatic eggs probably are primitive and that 
any departure from this trend is derived. It is 
difficult to ascertain whether a black or a 
white egg is primitive, therefore, the primitive 
allocation of the first two states is based on 
the large number of eggs and not the color. 
Eggs which are often exposed to sunlight are 
usually black. 
Twenty-nine characters and their 65 associated 
character states were used in this analysis. Two 
characters, 3 and 40, exhibit the same state in the 
46 BULLETIN OF LOS ANGELES COUNTY MUSEUM OF NATURAL HISTORY No. 12 
TABLE 2 
Distribution of 29 Characters among 
Atelopus, Dendrophryniscus, Melanophryniscus, and Oreophrynella 
Characters* 1 2 3 4 5 6 7 10 13 14 16 17 20 21 
Atelopus A* 1* A P,A, A, P 3 P 1 Pi.P. A P A 1 
Dendrophryniscus p* 2 A A2,A3 Ax P 1 P,A 1 P2 A P P 1 
Melanophryniscus P 1 A A2 Ax P 2 P,A 2 P2,A A A P 2 
Oreophrynella P 1 A A? A2 A 1 P 3 A P P P 1 
Characters 23 24 25 26 28 31 33 35 36 37 38 39 40 41 42 
Atelopus A2 A P P,A A2 A P,A P,A P,A3 P,A 1 1,2 A 1,2 Pi 
Dendrophryniscus P P P A Ax A P,A A A3 A 1 
^ A 2 P2 
Melanoph ryniscus P P P P A, P P A P P 2 4 A 1,2 P2 
Oreophrynella A3 A A A A, P P A A, A 2 4 A 1 A 
*Characters numbered as in generic definitions; A = advanced state, Al; AOJ etc. = different advanced states; 
1, 2, etc. = different neutral states; P ? primitive state, Pl; P2, etc. =: different primitive states. 
four genera but are included because of their im- 
portance in later comparisons. Atelopus has under- 
gone the greatest radiation in terms of component 
species and, hence, has the greatest number of 
character states, 37. Melanophryniscus and Den- 
drophryniscus have about the same number of 
states, with 31 and 32 respectively. Oreophrynella 
is represented by a single species and thus has only 
29 character states. 
In Table 2 the distribution of the 65 character 
states among the four genera is listed. Melanophry- 
niscus has the greatest number of the 20 primitive 
states with 14, followed by Atelopus with 13, Den- 
drophryniscus with 11 and Oreophrynella with 
only seven. This comparison would seem to indi- 
cate that Melanophryniscus is the most primitive 
and that Oreophrynella is the least primitive. While 
this analysis gives some indication of relative 
primitiveness among the four genera, it fails to 
consider the number and distribution of shared 
primitive characters. 
Ten of the 14 primitive character states found 
in Melanophryniscus are shared with Dendrophry- 
niscus. Five of these 10 are also shared with Ate- 
lopus (Table 2). Characters 10 and 14 actually 
are the same in some species of Dendrophryniscus, 
Melanophryniscus, and Atelopus. However, in 
Melanophryniscus and Dendrophryniscus, there 
are independent trends towards fusion and loss of 
the palatines and reduction and loss of the qua- 
dratojugal. These two characters, therefore, are 
indicative of greater similarities among the three 
genera than apparent in the original analysis. The 
three genera show similarities in numbers of eggs, 
but egg color clearly allies Dendrophryniscus with 
Melanophryniscus. 
Oreophrynella has seven primitive characters, 
five of which it shares with Melanophryniscus. 
Four of these five states are also shared with Den- 
drophryniscus, and two of them with Atelopus as 
well. A fourth character state (17) is found in 
Oreophrynella, Dendrophryniscus, and Atelopus. 
The absence of this primitive state in Melanophry- 
niscus indicates divergence in this character from 
the primitive stock which probably had an unossi- 
fied orbitosphenoid. Only in character 16 does 
Oreophrynella have a primitive state not found 
in the other three genera. 
With the exception of the state of character 17, 
all the primitive character states of Dendrophry- 
niscus are also found in Melanophryniscus. The 
sharing of 10 primitive character states clearly 
indicates that Dendrophryniscus was derived from 
the same stock that gave rise to Melanophryniscus. 
Atelopus has 13 primitive characters, two of 
which are common to all the genera. Characters 6, 
14, 25 and 42 are primitive in Atelopus, Melano- 
phryniscus, and Dendrophryniscus, although char- 
acters 14 and 42 have different primitive states in 
Atelopus. Atelopus, Dendrophryniscus, and Oreo- 
phrynella share only character state 17. Although 
this character is advanced in Melanophryniscus, 
the ancestral stock undoubtedly had the primitive 
state. It is noteworthy that no primitive characters 
are common only to Atelopus, Melanophryniscus 
and Oreophrynella. This again indicates the simi- 
larity between Melanophryniscus and Dendrophry- 
niscus. Nearly all character states present in the 
former are also present in the latter. 
Atelopus has two primitive character states not 
found in the other genera. The presence of a 
middle ear and the closely associated presence of 
1971 MORPHOLOGY AND EVOLUTION OF NEOTROPICAL FROGS 47 
a depressor slip to the annulus tympanicus are 
unique to three species of Atelopus. This condition 
can only be interpreted as the retention of a primi- 
tive character state. 
Three remaining primitive character states, 26, 
36, and 37, are shared only between Atelopus and 
Melanophryniscus. Two of these characters, 36 
and 37, are closely associated with hearing and are 
being lost. The bicondylar articulation of the 
coccyx in the two genera again suggests common 
ancestry. 
Analysis of shared primitive character states 
indicates that Melanophryniscus is more primitive 
than Dendrophryniscus though both probably were 
derived from the same ancestor. Atelopus probably 
was derived from this same ancestral stock, diverg- 
ing much earlier, however. Several primitive states 
were retained in the Atelopus line. Most of the 
primitive states retained in the Melanophryniscus 
line are also found in Dendrophryniscus. Oreo- 
phrynella apparently is closer to the Melanophry- 
niscus line than it is to the Atelopus line. The 
derivation of Oreophrynella and Melanophryniscus 
from a common ancestor could have occurred 
prior to the split of Melanophryniscus and Ate- 
lopus stocks or after the separation of the A telopus 
line. 
An analysis of relationships based on the group- 
ing of entities according to their shared primitive 
character states is very useful. However, this 
approach does not consider the advancements or 
evolutionary change of each unit. Once relation- 
ships have been established on the basis of similar 
primitive character states, the evolutionarily impor- 
tant characters of advancement must be incor- 
porated into the study. The patterns of relation- 
ships established on the basis of advanced charac- 
ter states may be compared to those established on 
degree of similarity in primitive states. 
There are 22 characters that have 27 different 
advanced character states (Table 2). Melanophry- 
niscus has the least number of advanced states with 
ten, followed by Dendrophryniscus with 14, Oreo- 
phrynella with 15, and Atelopus with 16. 
All advanced character states (10) found in 
Melanophryniscus, except that of characters 14 
and 17, are also found in Dendrophryniscus. This 
supports the close relationship between these two 
genera established from the primitive character 
states analysis. Melanophryniscus also shares six 
states with Oreophrynella, four of which are found 
in all the genera. The fifth character state, the con- 
dition of the pectoral girdle (28), is advanced in 
all four genera, but differs between Melanophry- 
niscus, Dendrophryniscus and Oreophrynella, and 
Atelopus. The advanced state in Atelopus repre- 
sents the culmination of an evolutionary trend to- 
wards fusion. Of the remaining five character 
states, two are common to Melanophryniscus, Den- 
drophryniscus, and Atelopus which clearly sets 
Oreophrynella off from these three genera. Charac- 
ter state 10, shared only by Melanophryniscus and 
Dendrophryniscus, reflects the independent loss of 
palatines in these genera. Only Melanophryniscus 
has an ossified orbitosphenoid; this state generally 
coincides with the increased skull ossification 
characteristic of the genus. The absence of the 
quadratojugal (14) in some species of Melano- 
phryniscus and Oreophrynella probably indicates 
independent losses. 
Dendrophryniscus has 14 advanced character 
states, only one of which is unique. The secondary 
modification of the depressor mandibulae distin- 
guishes Dendrophryniscus minutus from D. brevi- 
pollicatus and the other genera. Of the remaining 
13 states, eight are shared with Melanophryniscus, 
and most of these are also found in Atelopus. 
The remaining five character states of Dendro- 
phryniscus are shared with Atelopus or with Ate- 
lopus and Oreophrynella. The coccyx (character 
26) is fused with the sacrum in some Atelopus, in 
Dendrophryniscus and Oreophrynella. It is free 
and articulates with the sacrum in other species of 
Atelopus and all Melanophryniscus. Therefore, it 
seems likely that this advanced state has been in- 
dependently derived in each of the three lines. 
Character 37 has exactly the same distribution as 
character 26. In this instance the loss of vocal slits 
is obviously correlated with the loss of the middle 
ear (see discussion below). There is also a pro- 
gressive loss of the ostia pharyngea (character 36) 
associated with the loss of the middle ear. 
The two remaining character states are the loss 
of a digit of the hand (31), which may be the result 
of arrested development, and fusion of centralia 
in the hand (33). Apparently, both of these states 
were derived independently in Atelopus and Den- 
drophryniscus. Each may be the result of a shift 
in locomotory patterns. 
Most of the advanced character states of Oreo- 
phrynella are either common to all genera or re- 
stricted to Oreophrynella. Of its 15 advanced 
states, Oreophrynella shares six with Melano- 
phryniscus and seven each with Atelopus and Den- 
drophryniscus. Of the six shared with Melano- 
phryniscus, four are common to all genera. The 
fifth (28), as previously discussed, is part of an 
evolutionary continuum; the sixth (14) represents 
the parallel loss of the quadratojugal. 
The fusion of the atlas with the first trunk verte- 
48 BULLETIN OF LOS ANGELES COUNTY MUSEUM OF NATURAL HISTORY No. 12 
bra (24) and the fusion of the coccyx with the 
sacrum (26) are independent advancements 
shared between Oreophrynella and Atelopus in the 
former case and among Oreophrynella, Atelopus, 
and Dendrophryniscus in the latter case. The loss 
of vocal slits has also occurred in these three 
genera. 
In three (6, 25, 42) of the six remaining char- 
acters, Oreophrynella has an advanced state while 
the other three genera have primitive states. In the 
remaining three characters Oreophrynella is very 
different from the other genera in which the char- 
acters are either primitive or in a different ad- 
vanced state (5 and 23). The condition of the ostia 
pharyngea (36) is different in Oreophrynella but 
probably of less significance than are the other 
six states. 
Atelopus, with 16, has the greatest number of 
advanced states. As previously mentioned, four 
of these are found in all genera. Two states (5 and 
16) are shared with Melanophryniscus and Den- 
drophryniscus. Several (24, 26, 31, 33, 36 and 37) 
are shared with either Dendrophryniscus or Oreo- 
phrynella or both. These states were discussed 
above and represent parallel advancements. The 
four other advanced character states are restricted 
to Atelopus. Character states 1 and 20 are ad- 
vanced only in Atelopus; character state 28 is more 
advanced than the advanced states in the other 
genera. Character state 23 is also advanced in 
Oreophrynella though apparently it was achieved 
independently. 
The comparison of advanced character states 
indicates that Atelopus, Oreophrynella, and Den- 
drophryniscus are specialized in more characters 
than Melanophryniscus. Atelopus has the most 
advanced states, but only four of the 16 are re- 
stricted to Atelopus. Oreophrynella has 15 ad- 
vanced character states, but has six states which 
are unique. Dendrophryniscus and Melanophry- 
niscus have 14 and ten advanced states, respec- 
tively, but each has only a single unshared state. 
It appears that the primitive character state 
analysis and the advanced character state analysis 
show similar results. In general these same results 
are indicated in an examination of the neutral 
characters. Dendrophryniscus is most closely re- 
lated to Melanophryniscus in both analyses. Its 
similarity to Melanophryniscus in shared primitive 
states as well as the presence in Dendrophryniscus 
of all the advanaced states of Melanophryniscus 
except two suggests that Dendrophryniscus was 
derived from a Melanophryniscus ancestral stock. 
Similarities between Atelopus and Dendrophrynis- 
cus in advanced character states are attributed to 
parallel evolution probably the result of similar 
selective forces. 
Melanophryniscus has the greatest number of 
primitive and least number of advanced states 
among the genera. These data indicate that Me- 
lanophryniscus is most similar to the ancestral 
stock which gave rise to the four genera. 
Melanophryniscus shares more primitive and 
more advanced characters with Atelopus than it 
does with Oreophrynella. 
Oreophrynella has the least number of primitive 
characters and the most advanced states of any of 
the other genera. I interpret these data to mean that 
the Oreophrynella line and the Melanophryniscus- 
Atelopus-Dendrophryniscus line diverged from the 
ancestral stock at different times. 
The high number of advanced states in Oreo- 
phrynella indicates that it subsequently has become 
greatly specialized. Atelopus also has specialized in 
several characters but has retained some primitive 
states not found in Oreophrynella or Melanophry- 
niscus and its derivative Dendrophryniscus. That 
this condition has evolutionary advantage is indi- 
cated by the complex radiation of several different 
species groups in Atelopus. Oreophrynella, how- 
ever, is a much more specialized frog and as such 
has a restricted range and few species. The high 
number of primitive character states and the low 
number of advanced states characteristic of Me- 
lanophryniscus suggest that this genus is more 
generalized than either Atelopus or Dendrophry- 
niscus and probably resembles the ancestral stock 
more closely than do either of the latter two 
genera. 
My interpretation of the relationships among 
the four genera is shown by a dendrogram (Fig. 
11). This representation is based on two factors, 
the number of shared primitive character states 
and the number and sequence of advancements. 
The first establishes the basic relationships, while 
the second is useful in drawing the phylogeny. 
To elucidate the two lineages that gave rise to 
Atelopus, Melanophryniscus, and Dendrophrynis- 
cus on the one hand and Oreophrynella on the 
other, the major character states of the first group 
of genera are contrasted with those of the second 
lineage. The characteristics of Oreophrynella are 
placed in parentheses: adductor mandibulae 
muscle, S (E); frontoparietal bone normal, cover- 
ing most of frontal and all of parietal fontanelles 
(greatly reduced exposing entire frontal and parie- 
tal fontanelles); quadratojugal usually present (ab- 
sent); sphenethmoid complex ossified (only the 
ethmoid   portion   ossified);   presacral   vertebrae, 
1971 MORPHOLOGY AND EVOLUTION OF NEOTROPICAL FROGS 49 
FIGURE 11. Diagrammatic representation of the evolu- 
tionary relationship among Atelopus, Dendrophrynis- 
cus, Melanophryniscus, and Oreophrynella. 
seven or eight (five); sacrum usually separate from 
trunk vertebra (fused with trunk vertebra); and 
eggs unpigmented or with a pigmented animal pole, 
from 50 to 900 in number (unpigmented, 10 to 
15). 
The major character shifts that occurred in the 
evolution of Oreophrynella involved the loss of the 
adductor mandibulae posterior subexternus muscle; 
the differential metamorphosis of the frontoparie- 
tal bone; the loss of the quadratojugal; the reduc- 
tion of presacral vertebral elements from eight to 
five; the fusion of the atlas and first trunk verte- 
bra, the terminal trunk vertebra and the sacrum, and 
the sacrum and the coccyx. Oreophrynella retains 
the primitive condition in most of the thigh muscle 
characters; in the palatines; in the unossified ante- 
rior portion of the sphenethmoid; in the hyoid struc- 
ture; in the arcifero-firmisternal pectoral girdle; 
and in the higher number of centralia and higher 
phalangeal formula of the hand. 
The A telopus - Melanophryniscus - Dendrophry- 
niscus lineage retains the primitive condition in the 
nature of the frontoparietal, palatine, and qua- 
dratojugal; in the separate condition of the sacrum 
and trunk vertebrae; and in the higher number of 
eggs. These three genera lack the adductor externus 
muscle, have an ossified sphenethmoid, and possess 
seven or eight presacral vertebrae, all differences 
from Oreophrynella. 
Melanophryniscus retains most of the primitive 
character states. Dendrophryniscus was derived 
from Melanophryniscus but differs from it in the 
modified depressor mandibulae muscle condition, 
in the fusion of coccyx and sacrum, in the reduc- 
tion of centralia in the hand from two to one, and 
in the loss in one species of the distal element on 
the first digit of the hand. 
Atelopus has a few primitive character states 
which are not found in either Melanophryniscus or 
Dendrophryniscus. The more important of these are 
the presence in one species of a middle ear and its 
associated annulus tympanicus slip of the depres- 
sor mandibulae muscle and the presence of a well- 
developed quadratojugal. Atelopus has several 
character states that distinguish it from the other 
genera. These are advancements and are of major 
importance in the evolutionary history of the genus. 
They include an elongation of the tensor fasciae 
latae muscle; reduction to seven presacral verte- 
brae; fusion of the atlas and first trunk vertebra; 
attainment of a firmisternal condition in the pec- 
toral girdle; and the development of unpigmented, 
heavily yolk-laden eggs and specialized larvae 
associated with the mountain stream habitat. 
EVOLUTIONARY TRENDS 
The small size of the family (four genera and 
about 50 species) makes it difficult to evaluate in 
detail character trends which might be important 
in determining evolutionary changes. Nevertheless, 
several evolutionary trends became apparent dur- 
ing the course of this study. Some of these trends 
are important to an understanding of the history 
of the group and probably reflect major functional 
shifts in the biology of the organism. Other charac- 
ter trends are indicated but cannot be associated 
with any major shift, functional or otherwise. 
Locomotion 
There are several changes in character states 
which apparently are correlated with changes in 
basic locomotory patterns. Although observations 
on live material are limited to seven species of Ate- 
lopus and Melanophryniscus moreirae, some con- 
clusions concerning patterns of locomotion can be 
made. These frogs are walkers rather than hoppers. 
Although they are capable of jumping, they usually 
move each limb independently rather than moving 
the hind legs together in a coordinated jump. 
The shortening of the vertebral column by fusion 
of the atlas and first trunk vertebra and of the 
posterior trunk vertebra and sacrum suggests close 
association with walking locomotion. The trend 
50 BULLETIN OF LOS ANGELES COUNTY MUSEUM OF NATURAL HISTORY No. 12 
towards fusion at the atlas is indicated in Melano- 
phryniscus and Dendrophryniscus where there is 
close association between the first two vertebrae. 
Apparently the fusion in Atelopus represents a 
culmination of this trend. The fusion of trunk 
vertebrae with the sacrum occurs in some species 
of Atelopus, Melanophryniscus, and Dendrophryn- 
iscus but is present in all specimens of Oreophryn- 
ella quelchii. It is important to note that Oreo- 
phrynella has a reduced number of vertebral 
elements through both anterior fusion at the atlas 
and posterior fusion at the sacrum. 
Another feature probably correlated with the 
reduction in number of vertebral elements is the 
shift towards a firmisternal girdle. Even though a 
firmisternal pectoral girdle occurs in several un- 
related groups of frogs, there is little doubt that 
it is a derived condition. The functional aspects of 
a firmisternal as opposed to an arcifero-firmisternal 
or an arciferal condition are unknown. The shift 
from a hopping to a walking pattern of locomotion 
apparently has dictated a more rigid union between 
the two halves of the pectoral girdle to give added 
support to the limbs and perhaps has given rise to 
the firmisternal condition. The fusion apparently 
begins at the anterior end of the girdle and pro- 
ceeds posteriorly on the midline. Oreophrynella 
and Melanophryniscus have slightly more overlap 
than is found in Dendrophryniscus. 
It seems apparent that limb elongation as well 
as a change in habitus are results of the shift to 
walking. Atelopus and Dendrophryniscus have the 
atelopodid habitus and more elongate legs than the 
more primitive Melanophryniscus and Oreophry- 
nella which resemble the arciferal "hopping" Bufo. 
It is also assumed that the forward shift of the 
origin of the tensor fasciae latae muscle may reflect 
an advancement and, hence, specialization asso- 
ciated with the walking. With the walking type of 
locomotion, the hind legs often are directed much 
more posteriorly than in hopping. If the tensor 
arose on the posterior part of the ilium, as it does 
in most frogs, efficiency would be greatly decreased 
when the leg was directed posteriorly. Only by 
moving the origin anteriorly on the ilium is the 
efficiency of the muscle increased so that it can 
function to tense the adductor of the hind leg thus 
initiating motor response. 
The ancestral form must have had eight pre- 
sacral vertebrae, free vertebral elements, an arci- 
feral girdle, a posterior origin of the tensor muscle, 
and a hopping locomotion. Toads of the genus 
Bufo fit this description very well. The trends in 
the derived forms have been toward increased 
rigidity along the longitudinal axis of the body by 
reducing the number of functional vertebral ele- 
ments from eight to seven and five through fusion 
of the vertebrae and of the coccyx with the sacrum, 
to shift to an arcifero-firmisternal or firmisternal 
condition, to move the origin of the tensor muscle 
anteriorly on the ilium, and to shift to walking 
locomotion. 
Without much additional field experience with 
these frogs, it is difficult to postulate what environ- 
mental or biological factors contributed to and 
selected for the apparent shift in locomotory pat- 
tern. Several ideas are put forward concerning what 
may be important factors contributing to the loco- 
motory shift. These ideas are hypotheses and are 
proposed in the hope of stimulating additional work. 
These shifts may be in response to a shift from 
a terrestrial to an arboreal existence. Melano- 
phryniscus is entirely terrestrial; some species of 
Atelopus climb onto bushes at night; Dendrophry- 
niscus brevipollicatus is arboreal. The T-shaped 
terminal phalanges in D. brevipollicatus increase 
the surface area of the toe tip and are directly 
correlated with its climbing habits. Perhaps the 
apparent opposable digits of the feet in Oreophry- 
nella also reflect a partially arboreal existence. 
A second possible explanation is associated with 
the aposematic coloration and diurnal habits of 
most of these frogs. Diurnal frogs are more sus- 
ceptible to predators that hunt by sight than are 
nocturnal species. A slow, methodical pattern of 
locomotion such as walking may have a higher 
selective value than an erratic, abrupt pattern such 
as jumping. It also may be true that an aposematic 
device is more effective when the predator can 
easily discern the warning color pattern. If apose- 
matic coloration is effective in protecting the frog, 
then walking may be more efficient than jumping 
or hopping in terms of energy utilization. 
Finally, these frogs feed primarily on ants; there 
is no need for quick movements during feeding. 
These myrmecophagous habits also may have con- 
tributed to shifts in locomotory patterns. 
Hearing 
One of the most interesting evolutionary shifts 
found in several different lineages of frogs is the 
loss of the auditory apparatus. Probably the hear- 
ing apparatus of early Amphibia evolved in re- 
sponse to the shift from an aquatic to a terrestrial 
habitat and functioned primarily in environmental 
adjustment. It generally is agreed (Eaton, 1959) 
that the primitive frog probably had a well- 
developed tympanum and middle ear. Vocaliza- 
tion is characteristic of most frogs and could only 
have developed after the means for reception of 
airborne sound, the ear apparatus, was already 
1971 MORPHOLOGY AND EVOLUTION OF NEOTROPICAL FROGS 51 
present. Therefore it seems that the absence of an 
ear apparatus is secondarily achieved. 
Only a few species of Atelopus (e.g. flavescens, 
spumarius) have any remnant of an ear apparatus. 
The retention of an internal tympanum and middle 
ear in these forms is a primitive condition. The 
other species of Atelopus investigated, as well as the 
species of the other genera, lack an ear apparatus. 
Although the biological factors important in 
initiating the loss of the external tympanum and 
middle ear are unknown, it is clear that once the 
ear starts to degenerate, the vocal apparatus also 
begins to degenerate. Species of Melanophryniscus 
are the only forms studied which always have ex- 
ternal vocal pouches, visible ostia pharyngea, and 
vocal slits. Mating calls are common in the breed- 
ing biology of these frogs. Oreophrynella and 
Dendrophryniscus lack an ear appartus; these frogs 
also lack ostia pharyngea and vocal slits. Many 
species of Atelopus lack ostia pharyngea and vocal 
slits. Only in the species which still have an internal 
tympanum and middle ear are the ostia pharyngea 
well developed. Thus it seems that the ancestral 
stock of these forms had a well-developed hearing 
apparatus. It also is likely that vocalization was an 
important component of their breeding biology. 
Vocalization is still prominent in the breeding be- 
havior of the more primitive Melanophryniscus, 
but I suggest that because of their apparent in- 
ability to hear, vocalization is becoming less im- 
portant in their reproductive behavior and will 
eventually be lost. The same seems to be true of 
several species of Atelopus, some of which may 
already have abandoned the vocal portions of their 
reproductive behavior. 
Earlessness is more prevalent in frogs which 
occur at very high elevations, especially in the 
families Bufonidae and Leptodactylidae. Most of 
these earless frogs also occur in tropical areas and 
probably were derived from ancestors that lived in 
the warmer lowlands. Possibly a shift to diurnal 
activity patterns which allows the frogs to take 
advantage of warmer daytime temperatures was 
necessary for successful colonization of colder 
montane habitats. Once the frogs moved into this 
new environment, their reproductive behavior was 
modified in accord with the shift to diurnal activity. 
Frogs in which breeding calls or choruses formed 
an integral part of the reproductive behavior would 
be subject to strong selection by predators that 
were attracted by the calls. In fact, a shift to diur- 
nal patterns for any reason would subject the frogs 
to strong selection by sound hunting predators 
during the reproductive season if vocalization was 
important in their breeding behavior. 
These hypotheses could be invoked to explain 
the loss of the ear in those species of Atelopus 
which apparently do not vocalize. The latter hy- 
pothesis, correlating earlessness with high elevation, 
seems more reasonable, because those species with 
middle ears are found at low to intermediate eleva- 
tions today. 
Once the middle ear apparatus has been lost, 
apparently the ability to receive airborne sounds 
also is lost. What effect this has on species that 
relied on vocalization as an important factor in 
their reproductive behavior is unknown. 
Aposematic Coloration 
There is overwhelming morphological evidence 
of a trend towards the loss of the middle ear 
apparatus and its associated structures and hence 
the loss of the ability to detect airborne sounds. 
There also is evidence that the degeneration of the 
vocal apparatus is associated with the loss of the 
ear and that vocalization is becoming less impor- 
tant in the reproductive behavior of these frogs. 
With the loss of vocalization as a behavioral means 
of recognition, there apparently has evolved a 
secondary means of recognition in these frogs. I 
suggest that specific modifications of the basic apo- 
sematic coloration and subsequent visual recogni- 
tion by most species, at least of Atelopus, replaces 
the role of vocal recognition and is of primary im- 
portance in mate selection. 
Aposematic coloration is a well-known feature 
of many tropical frogs (Cott, 1957). Many of 
these brightly colored species are diurnal and thus 
subjected to sight hunting predators (e.g., birds, 
snakes). However, most of these bright diurnal 
frogs, including Melanophryniscus and Atelopus, 
possess poisonous glandular skin secretions, an 
effective defense against predators. Daly and 
Myers (1967) discussed the biological and chemi- 
cal aspects of the toxins of poison frogs of the 
family Dendrobatidae. Fuhrman, Fuhrman and 
Mosher (1969) discussed the toxic qualities of the 
skin secretions of Atelopus zeteki and commented 
on its distinctive properties. 
Cott (1957:292) made reference to the dis- 
criminatory feeding of a Crested Seriema (Cari- 
ama cr is tat a) which refused specimens of Melano- 
phryniscus stelzneri. Budgett (1899:310) de- 
scribed a grass snake which fed only on a specimen 
of Physalaemus cuvieri although several specimens 
of Melanophryniscus stelzneri were available. 
Cochran (1961) reported the unken reflex in 
Melanophryniscus stelzneri. This peculiar behav- 
ioral response is part of an aposematic display to 
reveal the bright colors of the ventral surface. The 
52 BULLETIN OF LOS ANGELES COUNTY MUSEUM OF NATURAL HISTORY No. 12 
skin secretion of this frog is reported to be very 
poisonous; increases in amount of secretion have 
been recorded during this display. All these reports 
confirm the aposematic nature of the bright color 
in Melanophryniscus. The same is suggested for 
Atelopus. 
There is little doubt that the aposematic colora- 
tion would have evolved after the frogs acquired 
the diurnal habit and probably after the skin de- 
veloped the toxic secretory glands. Also there is 
little doubt that the original function of the bright 
coloration was to act as a warning device for pred- 
ators. 
Cott (1957) stated that conspicuous coloration 
is known to have little, if any, sexual significance 
in frogs. Muntz (1964), however, has shown that 
some frogs have color vision. Daly and Myers 
(1967) suggested that bright coloration might 
serve for intraspecific recognition in some popula- 
tions of Dendrobates. Savage (1967) indicated 
that bright coloration and sexual dichromism in 
Bufo periglenes was intimately related to their 
breeding behavior and mate selection. I suspect 
that the aposematic coloration of several species of 
Atelopus and Melanophryniscus has secondarily 
become extremely important in their reproductive 
behavior and that mate selection is accomplished 
primarily through visual recognition. Once the 
middle ear and vocal apparatus started to degen- 
erate, the preadaptation of aposematic coloration 
would be favored by selection in terms of repro- 
duction as well as subject to selective forces from 
predators. The additive selection by two very 
different forces might well account for the tremen- 
dous amount of inter- and intrapopulational varia- 
tion found in forms of Atelopus varius, Atelopus 
ignescens, and Melanophryniscus stelzneri. Sexual 
dichromism is well marked in many populations 
of Atelopus in Costa Rica, Panama, and Colombia. 
This fact lends additional support to the hypothesis 
that mate selection depends on visual rather than 
vocal recognition. The realization that visual cues 
might be important in reproductive biology and 
that color pattern alone could serve as the pre- 
mating isolating mechanism introduces a new di- 
mension into anuran systematics. 
Paedomorphosis 
Wake (1966) has demonstrated the importance 
of paedomorphosis in the evolution of piethodontid 
salamanders. It appears that paedomorphosis has 
played a minor role in the evolution of atelopodid 
frogs probably because of their terrestrial habits. 
There is clearly a paedomorphic character in 
Oreophrynella quelchii. In this species the fronto- 
parietal fails to complete metamorphosis and in 
the adult appears as two separate elements that fuse 
only posteriorly. Thus, the frontal and parietal fon- 
tanelles are exposed. The functional or evolu- 
tionary significance of this differential metamor- 
phosis is unknown. 
Griffiths (1954a) has shown the developmental 
patterns of metamorphosis of the frontoparietal 
bone in Bufo marinus. Examination of sequential 
stages of two species of Atelopus and Bufo marinus 
confirms his findings and indicates that the same 
basic metamorphic patterns of frontoparietal de- 
velopment also occur in the atelopodid genera. It 
is assumed that the ancestral form of Oreophryn- 
ella exhibited a similar pattern of metamorphosis 
and that the adult had a well-developed fronto- 
parietal bone. In Oreophrynella, however, a pae- 
domorphic trait has been substituted for the an- 
cestral adult condition by means of a modification 
of the ontogency of the ancestor in the descendant. 
The absence of palatine bones and quadrato- 
jugals and the reduction of the terminal phalanges 
in several species of Atelopus and Dendrophryn- 
iscus could be attributed to arrested development. 
The absence of palatine bones in some species of 
Melanophryniscus may be due to fusion and sub- 
sequent incorporation into the underlying chondro- 
cranium rather than failure to develop. 
There is a trend towards increasing the rigidity 
of the skull by either fusion of dermal bones, such 
as the nasals, frontoparietals and parasphenoid, to 
each other or to the underlying chondrocranium, or 
by secondary dermal ossification. This trend is espe- 
cially noticeable in Melanophryniscus and may re- 
flect a tendency of these frogs to be slightly more 
fossorial than are other species. The strengthening 
of the skull by fusion of elements is a common 
occurrence in certain fossorial forms. 
FAMILIAL RELATIONSHIPS 
Removal of Brachycephalus ephippium from the 
Atelopodidae (McDiarmid, MS) leaves Griffiths' 
family with only a single genus, Atelopus. I have 
shown the close relationship between Atelopus, 
Melanophryniscus, Dendrophryniscus, and Oreo- 
phrynella, although the latter genus apparently is 
representative of a slightly different lineage. To 
evaluate the evolutionary status of the Atelopodi- 
dae, these four genera must be considered as a unit. 
The four genera have eight or fewer procoelous 
presacral vertebrae; an SQ depressor condition; 
separate sartorius and semitendinosus muscles; 
the tendon of the semitendinosus muscle lying 
ventral to the tendon of the gracilis muscle; larvae 
with   horny  beaks   and  denticles   and   a   single 
1971 MORPHOLOGY AND EVOLUTION OF NEOTROPICAL FROGS 53 
sinistral spiracle; and pectoral (axillary) amplexus. 
The genera lack ribs and a parahyoid. This com- 
bination of features characterizes the four genera 
as members of the suborder Procoela. They differ 
from the Hylidae and Centrolenidae by lacking 
teeth, toe discs, and intercalary cartilages and by 
having arcifero-firmisternal or firmisternal pectoral 
girdles. They are distinguished from the Lepto- 
dactylidae (sensu strictu) by lacking an omoster- 
num and teeth and from the Dendrobatidae by the 
absence of teeth, an omosternum and hypobran- 
chial 2, and by the presence of expanded sacral 
diapophyses. These genera differ from the Pseudi- 
dae by the presence in the latter of an omoster- 
num, teeth on the maxilla, premaxilla and vomer, 
and an accessory phalanx in each digit. They differ 
from all these families in their constant SQ de- 
pressor condition and the presence of a Bidder's 
organ. Most of their characters are found, how- 
ever, in the remaining procoelous family, Bufoni- 
dae. 
Before discussing the familial status, I should 
like to consider the ancestral stock and those 
characteristics which must have been present in 
the form that gave rise to the Atelopus-Melano- 
phryniscus-Dendrophryniscus line and to the Oreo- 
phrynella line. The primitive states of all the 
characters used in the previous analysis are as- 
sumed to be in the composite ancestral line. This 
hypothetical atelopodid is compared to the genus 
Bufo in Table 3. The comparison shows that the 
hypothetical ancestor shares most of its 23 charac- 
ters with Bufo. The only major difference between 
the two is in character 40. The absence of a 
Bidder's organ is considered primitive and so pos- 
tulated for the ancestor. However, Bufo, like the 
four atelopodid genera, also has a Bidder's organ. 
Therefore, the immediate ancestral form probably 
also had a Bidder's organ. 
The striking similarities in shared primitive 
characters between the hypothetical ancestral ate- 
lopodid and Bufo suggest that a Bufo-like frog is 
the ancestor of the atelopodid genera. However, 
there are several characters which distinguish mem- 
bers of the genus Bufo from the four atelopodid 
genera. As was pointed out earlier, phylogenies are 
based on evolutionary change and character ad- 
vancement and not on shared character states 
alone. Therefore, although the ancestral form had 
many characters found in contemporary Bufo, I do 
not think that it was a Bufo as we know the genus 
today. 
Before continuing this comparison, it is neces- 
sary to clarify an osteological relationship which 
has been used to show close relationship between 
Atelopus and Bufo. Griffiths (1954b) described an 
otic element in atelopodid and bufonid species and 
stated that it was not found in members of any 
other families. In an effort to verify Griffiths' work, 
a series of eight metamorphosing Bufo marinus, 
including tadpoles measuring 24.6 mm total length 
(hind legs only) to metamorphosed individuals 
measuring 11.9 mm snout-vent length, and a 
recently metamorphosed series of Atelopus ignes- 
cens and A. varius were prepared in the same 
manner as was Griffiths' material. Griffiths stated 
that the otic element appears over the crista paro- 
tica just after the squamosal passes the posterior 
border of the eye. He presented a semi-diagramatic 
illustration and a photomicrograph showing this 
condition in A. ignescens (Griffiths, 1954b: Plate 
2, figs. 5, 6). According to Griffiths, after the otic 
element has fused with the squamosal, the resulting 
compound bone stops its backward migration and 
begins to rotate postero-laterally through 70 de- 
grees. Thus the otic component of the squamosal 
passes over the lateral portion of the auditory cap- 
sule coming to lie apposed to the lateral edge of 
the ear capsule in adults. Griffiths stated (op. 
cit. :39) that the rotation is completed well after 
the tail has disappeared. 
I could find no indication of a separate otic ele- 
ment in the specimens of Atelopus examined. A 
separate center of ossification was found in two 
metamorphosing specimens of Bufo marinus. In 
Bufo this element fuses with the squamosal and 
then with the lateral (otic) portion of the fronto- 
parietal. Griffiths claimed that the dorsal arm or 
otic portion of the squamosal (his latero-otic) of 
Atelopus persisted because it continued to function 
as the origin of the depressor muscle. It has been 
shown previously that the depressor muscle 
attaches to the dorsal arm of the squamosal in all 
four atelopodid genera. 
I interpret the otic element, as described ini- 
tially by Sedra (1949) and later by Griffiths 
(1954b), to be a second center of ossification of 
the squamosal bone. This second center over the 
crista parotica may have developed in response to 
environmental pressure correlated with a particu- 
lar behavior (fossorial existence) or in those forms 
in which dermal ornamentation (e.g., cranial crests 
of bufonids) is characteristic. Two centers of ossi- 
fication do not necessarily indicate two distinct 
bones, as claimed by Griffiths. Two centers of ossi- 
fication are known in other bony elements and it 
is not unreasonable to postulate their existence in 
the squamosal bone considering the tremendous 
reorganization of the cranium during metamor- 
phosis. This proposal would be particularly ap- 
54 BULLETIN OF LOS ANGELES COUNTY MUSEUM OF NATURAL HISTORY 
TABLE 3 
Comparison of Characters of the 
Hypothetical Atelopodid Ancestor and the Genus Bufo 
No. 12 
Character* 
1. tensor fasciae latae 
3. adductor longus 
4. depressor mandibulae 
5. adductor mandibulae 
6. frontoparietal 
10. palatine 
14. quadratojugal 
16. sphenethmoid complex 
17. orbitosphenoid 
18. ceratohyal 
21. hypobranchials 1 and 2 
23. presacral vertebrae 
24. atlas 
25. sacrum 
26. coccyx 
28. pectoral girdle 
31. centralia of hand 
33. phalangeal formula of hand 
35. middle ear apparatus 
36. ostia pharyngea 
37. vocal slits 
40. Bidder's organ 
42. eggs 
Hypothetical Ancestor 
short, origin on middle 
third of ilia 
present 
SQat 
SandE 
normal size, separate 
or fused on midline 
present 
present 
only ethmoid portion ossified 
unossified 
anterior process absent 
present 
eight 
separate from first 
trunk vertebra 
separate from trunk 
vertebra 
articulates with sacrum 
arciferal 
two 
2-2-3-3 
present 
present 
present 
absent 
black and white or 
white, 52-910 
Bufo 
short, origin on middle 
third of ilia 
present or absent 
SQ 
SandE 
normal size, separate 
or fused on midline 
present 
present 
only ethmoid portion ossified 
unossified 
anterior process absent 
present 
eight 
separate from first 
trunk vertebra 
separate from trunk 
vertebra 
articulates with sacrum 
arciferal 
two 
2-2-3-3 
present or absent 
present or absent 
present or absent 
present 
black and white, 31-30,000 
* Characters numbered as in generic definitions. 
plicable to those elements of the skull which are 
important in feeding, such as the squamosal. 
The use of this character to indicate close rela- 
tionship between groups of frogs is unwarranted 
until we know more about the distribution and 
variability of the element in other frog families. 
Reinbach (1939) discussed the otic element in the 
leptodactylid Calyptocephalella gayi. Examina- 
tion of a metamorphosing tadpole of this species 
(FMNH 23829, 126 mm total length, 52.5 mm 
snout-vent length) confirms his findings con- 
cerning the presence and development of the 
element in the Leptodactylidae. Thus, Griffiths' 
arguments (1954b: 45) concerning the homology 
of the otic element in Calyptocephalella are at 
least partially unfounded. Until we know more 
about morphogenesis of bony elements in the 
skull of many more species of frogs, attempts 
to homologize this element, even if it is a dis- 
crete bone, are fruitless. This is especially true 
with reference to homologies between bones in 
modern frogs and fossil amphibians. Jarvic (1968) 
used Griffiths' work with the otic element to 
support his contention of an origin of the Anura 
independent of other modern Amphibia. Taken 
to its logical conclusion, this would place the Anura 
in closer relationship to reptiles than to Urodeles. 
Jarvic based this interpretation of anuran relation- 
1971 MORPHOLOGY AND EVOLUTION OF NEOTROPICAL FROGS 55 
ships in part upon the homology of the otic ele- 
ment with the intertemporal of osteolepiforms 
because the otic element "arises at a place corre- 
sponding to the centre of radiation of the inter- 
temporal ..." 
Griffiths (1954b) did not find a discrete otic 
element in all species of Bufo which he examined. 
Because the otic element is present in some mem- 
bers of the Leptodactylidae, as well as in the Bufo- 
nidae and the Atelopodidae, and because the ele- 
ment is variable in form or absent in some members 
of the latter two families, Griffiths' claim that the 
otic element, as a separate bone, is diagnostic of 
and restricted to the Bufonidae and Atelopodidae 
is rejected. 
Since Bufo possesses many of the characters 
postulated for the ancestral atelopodid, it seems 
reasonable to compare the four atelopodid genera 
with the Bufonidae. This comparison will allow for 
an evaluation of their familial status. 
For this comparison I examined specimens of 
several species of Bufo, the neotropical bufonid 
Crepidophryne epioticus, representatives of several 
Asian bufonid genera including Ansonia, Caco- 
phryne, Pelophryne, and Pedostibes, and the 
African  bufonid  Nectophrynoides vivipara.  For 
information concerning the African genera Wol- 
terstorffina, Mertensophryne, Laurentophryne, 
Nectophryne, and the Asian genus Pseudobufo, I 
referred to Tihen's paper (1960). Additional 
material was obtained from Tihen's papers on Bufo 
(1962a, b) and from Savage and Kluge's paper on 
Crepidophryne (1961). Very little has been written 
about Didynamipus and Werneria. In fact the 
proper familial allocation of the latter genus is 
questionable. 
The characterization of the family Bufonidae 
is based on a combination of data obtained by 
examinations of the previously mentioned genera, 
of information contained in the literature, and of 
the characteristics used by Noble (1922, 1931) 
and Griffiths (1963) to define the family. Major 
characters common to the Bufonidae and the Ate- 
lopodidae are listed for comparison in Table 4. 
A comparison shows the similarities that exist 
between the two groups. Ten character states are 
the same in both families, while six characters 
have different states. Each of these six different 
states is clearly a stage in an evolutionary con- 
tinuum, and as such, should not be used to differ- 
entiate these two units at the familial level. 
TABLE 4 
Comparison of 16 Major Characters 
in the Bufonidae and the Atelopodidae 
Character 
vertebrae 
coccyx 
presacral vertebrae 
pectoral girdle 
sacral diapophyses 
tensor fasciae latae 
adductor longus 
tendon of semitendinosus 
depressor mandibulae 
adductor mandibulae 
Bidder's organ 
posterior laryngeal constrictor 
and pulvinaria vocale 
hypobranchials 1 and 2 
maxillary and vomerine teeth 
omosternum 
sternum 
Bufonidae 
holochordal, procoelous 
bicondylar articulation 
or fusion with sacrum 
seven or eight 
arciferal or arciferal- 
firmisternal 
greatly dilated 
short or elongate 
present or absent 
ventral to gracilis 
SQ 
S and E or E 
present 
absent 
both present 
absent 
present or absent 
present 
Atelopodidae 
holochordal, procoelous 
bicondylar articulation 
or fusion with sacrum 
five, seven or eight 
arciferal-firmisternal 
or firmisternal 
greatly dilated 
short or elongate 
absent 
ventral to gracilis 
SQ 
SorE 
present 
absent 
1 present, 2 absent 
absent 
absent 
present 
56 BULLETIN OF LOS ANGELES COUNTY MUSEUM OF NATURAL HISTORY No. 12 
The reduction in number of vertebral elements 
has been discussed previously. The trend towards 
reduction is well documented in the atelopodids 
and also is characteristic of the Asian bufonid 
Pelophryne and the African bufonids Merten- 
sophryne, Laurentophryne, and Nectophryne 
(Tihen, 1960). The primitive number of eight is 
reduced to a low of five in Oreophrynella. 
The shift from an arciferal to a firmisternal 
girdle must have passed through an arcifero- 
firmisternal condition. This apparently is what is 
found in three of the atelopodid genera as well as 
in the Asian bufonid Cacophryne and the African 
Didynamipus. A firmisternal girdle is found only 
in Atelopus. Griffiths (1959) suggested that Caco- 
phryne had its origin in the genus Pedostibes. 
Parker (1932) indicated that Didynamipus orig- 
inated independently from an ancestor which also 
gave rise to Nectophryne. The shift from arciferal 
to firmisternal apparently has occurred or is still 
occurring in at least three different lines within 
the two major groups. 
Tihen (1960) argued that the presence of an 
adductor longus muscle was a primitive condition 
and indicated that it was present in all members 
of the genus Bufo. However, my observations and 
those of Limeses (1964) indicate the absence of 
this muscle from some species of Bufo. Its absence 
from the other bufonid genera and the atelopodid 
genera as well suggests a closer relationship be- 
tween the two groups than originally indicated. 
The differences in the adductor mandibulae 
muscles are the result of at least two and probably 
more independent losses of either the externus or 
the subexternus muscles. The only condition from 
which both an S and an E condition could be de- 
rived is the primitive S and E condition which 
characterizes Bufo. All of the Asian and African 
genera which have been examined {Ansonia, 
Pedostibes, Nectophrynoides) to date, the Neo- 
tropical bufonid Crepidophryne, and the atelo- 
podid Oreophrynella show an E condition. The 
three other atelopodid genera, Melanophryniscus, 
Dendrophryniscus, and Atelopus, have an S condi- 
tion which probably arose only once. The E con- 
dition, however, probably arose at least twice, once 
in the Neotropics and once in the Old World 
Tropics. 
The absence of hypobranchial 2 suggests that 
the atelopodid genera were derived from an an- 
cestor which possessed both hypobranchial 1 and 
2. The presence of hypobranchial 2 in Bufo, Nec- 
tophrynoides, Cacophryne, and the Neotropical 
Crepidophryne is considered the primitive state. 
This character together with several others indi- 
cates that Crepidophryne is part of a Neotropical 
lineage more closely related to Bufo than to the 
atelopodid genera. 
An omosternum is present in Nectophrynoides 
and absent from the other bufonid genera. Its 
absence from the atelopodid genera does not con- 
stitute any greater a difference than exists between 
Nectophrynoides and the other bufonid genera. 
Of the 16 characters used in the comparison, the 
atelopodid genera are distinct from all members 
of the Bufonidae in only two characters, the S con- 
dition of the adductor mandibulae muscle (except 
in Oreophrynella) and the absence of hypobran- 
chial 2. 
Because of the close similarities between the 
Bufonidae and the four genera previously con- 
sidered in the Atelopodidae, because the two char- 
acter states which distinguish these four genera 
from the Bufonidae obviously are derived, and 
because the family Bufonidae, especially Bufo, is 
most closely related to the ancestral stock of the 
atelopodid genera, I consider the genera Atelopus, 
Melanophryniscus, Dendrophryniscus, and Oreo- 
phrynella to be members of the family Bufonidae. 
These four atelopodid genera together with Crep- 
idophryne and Bufo represent the Neotropical 
segment of the wide-ranging family Bufonidae 
which apparently has undergone several major, 
independent radiations in the three tropical regions 
of the world. 
The Bufonidae may be distinguished from other 
frog families by the following combination of 
characters. 
Bufonidae Gray 1825 
Definition: 1) eight, seven or five presacral ver- 
tebrae; 2) vertebrae holochordal, procoelous; 3) 
coccyx with bicondylar articulation or fused with 
sacrum; 4) pectoral girdle arciferal, arcifero- 
firmisternal, or firmisternal; 5) sacral diapophyses 
greatly dilated; 6) tensor fasciae latae muscle 
short or elongate; 7) tendon of semitendinosus 
muscle lying ventral to gracilis muscle; 8) adduc- 
tor longus muscle usually absent (present in most 
species of Bufo); 9) depressor mandibulae muscle 
in SQ; 10) adductor mandibulae muscle in S and 
E, S, or E; 11) Bidder's organ present; 12) hypo- 
branchial 1 present, hypobranchial 2 present or 
absent; 13) posterior laryngeal constrictor muscle 
and pulvinaria vocale absent; 14) maxillary and 
vomerine teeth absent; 15) omosternum usually 
absent (present but cartilaginous in Nectophry- 
noides; 16) sternum present; 17) aquatic larva or 
viviparity; 18) karyotype, 2N = 20 or 22. 
1971 MORPHOLOGY AND EVOLUTION OF NEOTROPICAL FROGS 57 
HISTORICAL PERSPECTIVE 
It was difficult to relate ideas concerning the 
phylogeny of the Neotropical bufonid genera to 
the paleogeographic development of South Amer- 
ica, primarily because so little is known about 
the geological history of most of South America. 
Harrington (1962) and several other workers 
(Childs and Beebe, 1963) have summarized our 
geological knowledge. Most works have dealt with 
the Andean area and represent efforts of petro- 
leum geologists in their search for oil deposits. 
While useful, these works rarely are pertinent to 
an interpretation of the early history of the South 
American frog fauna. 
A second difficulty is the virtual absence of an 
adequate fossil record. Hecht (1962, 1963) has 
summarized what little is known about the early 
history of frogs. He concluded that the major 
adaptive radiation of frogs was completed by the 
Early or Middle Jurassic and that by Cretaceous 
the higher groups of frogs had evolved. Most, if 
not all, of the procoelous families probably were 
distinct by the beginning of Tertiary. Tihen 
(1962b) reviewed the New World fossil bufonids 
and assigned the Argentinian Oligocene fossil de- 
scribed as Neoprocoela edentata by Schaeffer 
(1949) to the genus Bufo. Tihen referred this form 
to the calamita group of Bufo, that today is re- 
stricted to the Old World. Estes and Wassersug 
(1963) described a Miocene fossil from Colombia 
which they referred to the modern species Bufo 
marinus. Thus, the genus Bufo was in South Amer- 
ica as early as the Oligocene. Such evidence indi- 
cates that Blair's hypothesis (1963) that Bufo 
marinus invaded South America during the Ter- 
tiary isolation of that continent from North Amer- 
ica is unsatisfactory. Cole, Lowe, and Wright 
(1968) indicated that the marinus species group of 
Bufo is karyotypically distinct from North Amer- 
ican species groups of Bufo and suggested a long 
term isolation and independent evolutionary di- 
vergence of the marinus group in South America. 
These data and the South American fossils suggest 
that Bufo reached South America prior to the dis- 
ruption of the Panamanian land connection as 
early as the Late Cretaceous or Paleocene. 
Darlington (1957) postulated that the bufonids 
and the genus Bufo probably originated in the Old 
World Tropics and that they migrated to the New 
World across the Bering Land Bridge early in the 
Tertiary. However, the fossil record neither sup- 
ports nor denies Darlington's postulate. In fact, 
Hecht (1963) pointed out that the present distri- 
butions of modern frogs offer very few clues to 
the center of origin of frogs much less to the origin 
of the family Bufonidae. The present distribution 
of the bufonids, as well as that of certain other 
circum-tropical families, may be explained better 
by invoking a pre-Jurassic contact of the Old and 
New World continents. 
Another problem is the lack of agreement among 
anuran systematists about a classification of higher 
frogs. A brief comparison of the phylogenies and 
classifications proposed by Noble (1931), Reig 
(1958), Griffiths (1963), Inger (1967), and 
Kluge and Farris (1969) adequately illustrates 
this disagreement. Hecht (1963) summarized the 
current patterns of classification and proposed his 
own. While some of Hecht's proposals suffer from 
lack of detailed knowledge of many procoelous 
genera, his phyletic interpretation placed on a time 
axis is extremely useful in appraising the relation- 
ships among modern families in light of their fossil 
record. 
I have shown that the New World bufonid 
genera, with the possible exception of Bufo, were 
derived from an ancestral form that probably was 
present in South America by the beginning of the 
Tertiary. The overall similiarities among all species 
of Bufo, whether African, European, Asian, North 
American or South American, suggest that Bufo 
is an extremely successful frog which has been in 
the three major tropical areas since the Early Ter- 
tiary and remained about the same, or which 
evolved in a single area and subsequently dispersed 
into the rest of the world. The first hypothesis 
assumes that the primitive bufonid from which 
most of the modern bufonid genera were derived, 
was very similar to the modern genus Bufo. The ab- 
sence of bufonids from Madagascar and Australia 
probably precludes their differentiation in Africa 
prior to the Cretaceous and in the Indo-Australian 
area prior to the Upper Cretaceous. These assump- 
tions are not in disagreement with the fossil record 
nor with current ideas concerning the age of most 
procoelous anuran families. 
The second hypothesis assumes either a South 
American origin for the genus and dispersal into 
the Old World prior to the Paleocene or an Old 
World origin, probably Africa, for Bufo and sub- 
sequent dispersals into the New World. Tihen 
(1962a, b) favored an Old World origin for the 
genus and indicated that every major line of diver- 
sification within the family is represented among 
the various African forms. The primitive condition 
of Nectophrynoid.es vivipara (Tihen, 1960) and 
recent karyological data (Bogart, 1968) seem to 
support an African ancestry. The original location 
of the ancestral bufonid is not critical to this dis- 
cussion because, despite the place of origin, it or 
58 BULLETIN OF LOS ANGELES COUNTY MUSEUM OF NATURAL HISTORY No. 12 
its descendants had to have been in South America 
by the beginning of the Cenozoic. 
I am assuming that the Neotropical genera, with 
the possible exception of Bufo, were derived from 
a single ancestral group that was in South Amer- 
ica prior to the opening of the Panamanian portal. 
The ancestral form from which Melanophryniscus 
and Atelopus were derived probably occupied 
savanna like habitats in areas that today form the 
Paranaiba, San Francisco, and Parana Basins of 
Brazil. Harrington (1962) showed continental 
facies in these areas in the Late Cretaceous and 
throughout most of the Tertiary. 
Melanophryniscus has retained many of the 
generalized characteristics of the ancestral form 
and generally has remained in seasonal areas of 
subtropical savanna, pampean grassland, and de- 
ciduous forest (Smith and Johnson, 1945). The 
presence of a population of Melanophryniscus 
moreirae in the Amazon Basin suggests that the 
genus ranged further north, probably sometime 
during the Late Tertiary. Today it is found there 
only in relict patches of savanna. Haffer (1967, 
1969) discussed the isolated remnants of savannas 
in the forests of the lower Amazon. Gallardo 
(1966) suggested that a majority of the Chacoan 
amphibian species, including species of Melano- 
phryniscus, may be considered Guiano-Brazilian 
in origin. Haffer (1969) and Vanzolini and Wil- 
liams (1970) offered convincing arguments sup- 
porting climatic cycles of wet-dry periods in the 
Amazon Basin in the Pleistocene. Dispersal into 
the Amazon Basin from southern Brazil during a 
dry period also may account for the disjunct 
occurrence of M. moreirae near Obidos, Para, 
Brazil. I suspect that Melanophryniscus secondarily 
invaded the tropical and subtropical rain forests 
in southeastern Brazil and the subtropical forests 
on the eastern slope of the southern Andes in 
northeastern Argentina. 
Dendrophryniscus was derived from the ances- 
tral Melanophryniscus stock and has adapted to 
wet tropical forests. In southeastern Brazil one 
species has become a bromeliad inhabitant, while 
another and the species in the Amazon Basin are 
found on the forest floor. The ancestral form 
probably originated in the forests of the ancient 
Coastal Brazilian Shield. The striking differences 
among the species examined suggest that they have 
been isolated for a considerable period of time. 
The presence of another species of Dendrophryn- 
iscus in northeastern Brazil indicates a former 
connection between the eastern species along the 
coastal portions of the Brazilian Shield. They were 
probably connected to the form in the Amazon 
Basin sometime in the Tertiary. Subsequent cli- 
matic shifts, similar to those discussed by Vanzo- 
lini and Williams (1970) and geologic changes 
have fragmented Dendrophryniscus into three or 
four isolated populations. 
It appears that one of the major factors initially 
contributing to differentiation of the Atelopus line 
was the ability of the ancestors to utilize streams 
for breeding. As soon as the ancestral forms moved 
out of the lowlands and away from the primitive 
pond and still-water breeding sites and into 
the mountains and irregular terrain, they adapted 
to fast moving stream habitats. Thus an en- 
tire new adaptive zone became available. Behav- 
ioral modifications, including placing the eggs 
beneath rocks, and structural changes followed. 
Increased yolk content of each egg allowed for a 
longer developmental period of the tadpole within 
the egg but reduced the number of eggs per spawn. 
The larvae, however, were larger and stronger 
upon hatching and better able to maintain them- 
selves in the fast moving water. In addition the 
larger, more advanced larvae were able to feed 
immediately on hatching and were not dependent 
on yolk material for food early in the free-swim- 
ming stage. Specialized larval adaptations, espe- 
cially the sucking disc, were evolved in this fast 
stream habitat. 
With the initial major uplifts of the Andes in 
Peru and Ecuador in the Late Cretaceous (Ham 
and Herrera, 1963) and in Colombia in Late Cre- 
taceous and Paleocene (Jacobs, Biirgl and Conley, 
1963) and of the mountains of northern Vene- 
zuela in the Eocene (Mencher, 1963), entire new 
areas became available to and were exploited by 
Atelopus. The ancestral stock probably came from 
southeastern portions of the Guiana Shield where 
the more primitive Atelopus flavescens is found 
today. However, it is possible that Atelopus ini- 
tially invaded the montane habitat in the southern 
Andes and moved north along their eastern slope. 
It appears also that the higher elevations have 
been invaded several times. The presence of Ate- 
lopus in Central America is attributed to a north- 
ward movement following emergence of the Pana- 
manian isthmus in the Pliocene (Lloyd, 1963). 
The apparent close association of species of 
Atelopus with mountain streams has restricted spe- 
cies to particular drainage systems. The resulting 
isolation (drainage system and mountain top) 
probably accounts for the striking differences 
among many populations and probably has been 
an important factor contributing to speciation. 
Very little can be said about the history of 
Oreophrynella.  Apparently it was derived from 
1971 MORPHOLOGY AND EVOLUTION OF NEOTROPICAL FROGS 59 
the same stock that gave rise to the other three 
genera but probably at a different time. It is a 
very specialized frog which is restricted to an 
ancient part of the Guiana Shield. Harrington 
(1962) indicated that this area has been a rela- 
tively stable land mass since the Triassic. 
Mayr and Phelps (1967) discussed the relation- 
ships among the bird faunas of the tepuis of Vene- 
zuela. They mentioned that many authors attribute 
the avian distributional patterns to fragmentation 
of a once widespread avifauna in conjunction with 
the erosion and dissection of a former vast sand- 
stone tableland, the Roraima formation, which 
covered more than a million square kilometers 
of the Guiana Shield. Mayr and Phelps pointed 
out that this plateau is much more ancient than 
its bird fauna and accordingly discredited the 
theory. However, this theory may explain the dis- 
stribution of Oreophrynella, whose origin may 
extend well into the Early Tertiary and probably 
into the Cretaceous. 
While this study has clarified many of the prob- 
lems that previously existed, it has created, at least 
in my mind, several new problems and posed 
numerous questions. Many of these problems are 
concerned with the reproductive biology and the 
effectiveness of certain alleged premating isolating 
mechanisms. Some of my interpretations as well as 
possible explanations to several of the new prob- 
lems suffer from a lack of knowledge of the biol- 
ogy of these frogs. Solutions and answers must 
wait until more information becomes available 
through extensive field work. 
RESUMEN 
Ranas Atelopodid son uno de los grupos mas 
interesantes y diversos de la anura neotropical. 
El estado sistematico y la historia evolucionista 
de los generos de esta familia han sido pobremente 
intendidos. La meta de este estudio era analizar 
los conocimientos desponibles de la morfologta y 
biologia de estas ranas para clarificar sus rela- 
ciones evolucionistas y su historia. El grupo incluye 
aproximadamente 40 especies de cuatro generos: 
Atelopus, Dendrophryniscus, Melanophryniscus, 
y Oreophrynella. Los datos indican que Brachyce- 
phalus, un genero originalmente incluido dentro 
la familia, no esta estrictamente relacionado con 
los demas cuatro generos. Sus relaciones seran 
discutidos en otra parte. 
Especimenes representantes de todos los generos 
y de la mayoria de las especies fueron estudiados. 
Datos respecto de la miologia, la osteologia, y 
morphologia reproductiva fueron acumulados. 
Todo literatura desponible fue revisada y la infor- 
mation pertinente fue asimilado como parte de 
este informe. 
Una description detallada de la musculatura del 
muslo y de la mandibula y la osteologia de cada 
especie es presentado. Los craneos, las cinturas 
pelvianas, y los aparatos hioideos son describidos 
y ilustrados. Componentes del oido, algunas 
aspectos de la morfologia externa, la biologia 
reproductiva, y la ecologia son descritos. 
Cada genero es definido par 43 caracteres. Sus 
distributions geograficas son presentados en resu- 
men y sus especies incluidas y referidas son men- 
cionadas. Atelopus minutus Melin y Atelopus 
proboscideus Boulenger son puestos en el genero 
Dendrophryniscus. Atelopus rubriventris Vellard 
es puesto en el genero Melanophryniscus. 
Los cuatro generos son discutidos y sus estados 
caracteristicos comparados. Melanophryniscus 
tiene el major numero de caracteristicas primi- 
tivas y el menor numero de caracteristicas avan- 
zados y es probablamente el mas semejante a la 
cepa ancestral. Atelopus, tambien, tiene muchos 
caracteristicas primitivas, pero tiene el mayor 
numero de caracteristicas avanzadas. Atelopus y 
Melanophryniscus fueron derivados del mismo 
linaje, pero Atelopus ha experimentado una radia- 
tion significativa al nivel especifico y muestra 
varios adelantos no encontrados en los demas 
generos. Dendrophryniscus fue derivado de la 
Hnea de Melanophryniscus y muestra una evolu- 
tion paralela en algunas estados caracteristicos 
con Atelopus. Oreophrynella tiene mas estados 
avanzados caracteristicos que Melanophryniscus 
o Dendrophryniscus y solamente uno menos que 
Atelopus. Sin embargo, Oreophrynella tiene el 
mayor numero de estados unicos y el menor 
numero de estados primitivos. Aparentamente 
Oreophrynella fue derivado desde la cepa ancestral 
a un tiempo distincto de la linea Melanophryn- 
iscus-Atelopus-Dendrophryniscus y posteriormente 
se ha hecho muy especializado. 
Los mayores direcciones evolucionarias y los 
cambios de caracteristicas morfologicos aparenta- 
mente son asociados con cambios en el mecanismo 
de locomotion; otras son el resulto de metamor- 
fosis diferencial. Modificaciones biologicas aso- 
ciados con la perdida del aparato del oido medio 
y el desarrollo de coloration aposematico son im- 
portantes. El estato familiar del Atelopodidae es 
discutido y rechazado. Los generos Atelopus, Den- 
drophryniscus, Melanophryniscus, y Oreophryn- 
ella son puestos en la familia Bufonidae, lo que 
es redifinido. 
La cepa ancestral de la cual los cuatro generos 
fueron derivados probablamente existio en Ame- 
60 BULLETIN OF LOS ANGELES COUNTY MUSEUM OF NATURAL HISTORY No. 12 
rica del Sur antes del comienzo del cenozoico. 
La cepa ancestral Melanophryniscus-Dendrophryn- 
iscus-Atelopus probablamente ocurrio en las saba- 
nas o bosques deciduos del Brasil sureste. Melano- 
phryniscus ha retenido muchas de las caracteristi- 
cas ancestrales generalizadas y actualmente se 
encuentra en el mismo tipo general de ambito 
natural. Dendrophryniscus fue derivado de la cepa 
de Melanophryniscus y fue adapto a la selva 
humeda tropical del este del Brasil y la cuenca 
Amazonica. Atelopus se ha adaptado al ambito 
natural a orillas del arroyo y se ha movido a las 
areas montanosas que se hicieron disponables con 
el levantamiento de los Andes en los finis del cre- 
taceous y el comienzo del terciario. Este ambito 
natural neuvo ha sido explotado con buen exito 
par Atelopus y ha sido un factor mayor contri- 
buyente a su radiacion especifica. Oreophrynella 
es una rana muy especializada que fue derivada 
de una cepa antigua de Bufonidae y que fue sub- 
secuentemente restringida al Cerro Roraima un 
parte antigua del Escudo Guianico. 
1971 MORPHOLOGY AND EVOLUTION OF NEOTROPICAL FROGS 61 
LITERATURE CITED 
AHL, E. 1938. tiber die Kaulguappe des Dendro- 
phryniscus moreirae Miranda-Ribeiro. Zool. Anz. 
124(5/6): 158-159. 
ANDERSSON, L. B. 1945. Batrachians from east Ecua- 
dor collected 1937, 1938 by Wm. Clarke-Mac- 
intyre and Rolf Blomberg. Arkiv Zool. 37A(2): 
1-88. 
BADENHORST, C. E. 1945. Die Skedelmorfologie van 
die Neotropiese Anure Atelopus moreirae de 
Miranda-Ribeiro. Ann. Univ. Stellenbosch 23(2): 
1-19. 
BALDAUF, R. J. 1959. Morphological criteria and their 
use in showing bufonid phylogeny. J. Morphol. 
104(3): 527-560. 
BARBOUR, T. 1923. Notes on reptiles and amphibians 
from Panama. Mus. Zool. Univ. Michigan, Occ. 
Papers 129: 1-16. 
BARBOUR, T., AND G. K. NOBLE. 1920. Some amphib- 
ians from northwestern Peru, with a revision of 
the genera Phyllobates and Telmatobius. Mus. 
Comp. Zool., Harvard Univ., Bull. 63(8): 395- 
427. 
BARRIO, A. 1964. Pe.culiaridades del canto nupcial 
de Melanophryniscus stelzneri (Weyenburgh) 
(Anura, Brachycephalidae). Physis 24(68): 435- 
437. 
BHADURI, J. L. 1953. A study of the urogenital system 
of Salientia. Zool. Soc. Bengal, Proc. 6(1): 1-111. 
BLAIR, W. F. 1962. Non-morphological data in anuran 
classification. Syst. Zool. 6(2): 72-84. 
 1963.   Evolutionary   relationships   of  North 
American toads of the genus Bufo: a progress 
report. Evolution 17(1): 1-16. 
BOETTGER, O. 1892. Katalog der Batrachier-Sammlung 
in Museum der Senckenbergischen Naturforschen- 
den Gesellschaft. Gebriider Knauer, Frankfurt. 
73 p. 
BOGART, J. P. 1968. Chromosome number difference 
in the amphibian genus Bufo: the Bufo regularis 
species group. Evolution 22(1): 42-45. 
BOGERT, C. M. 1960. The influence of sound on the 
behavior of amphibians and reptiles. Amer. Inst. 
Biol. Sci., Publ. 7: 137-320. 
BOKERMANN, W. C. A. 1962. Una nueva especie de 
Atelopus del nordeste de Brasil (Amphibia, Sa- 
lientia, Brachycephalidae). Neotropica 8(26): 
42-44. 
 1966. Lista anotada das localidades tipo de 
anfibios Brasileiros. Impreso Pelo Servico Docu- 
mentacao, Sao Paulo, 181 p. 
1967.   Observacoes   sobre   Melanophryniscus 
moreirae (Mir. Rib.)   (Amphibia-Brachycephali- 
dae). An. Acad. Brasileira Cien. 39(2): 301-306. 
BOULENGER, G. A. 1882. Catalogue of the Batrachia 
Salientia S. Ecaudata in the collection of the 
British Museum. Second Edition. British Mu- 
seum, London. 503 p., plates 1-30. 
 1894. On the genus Phryniscus of Wiegmann. 
Ann. Mag. Nat. Hist., Ser. 6, 14(83): 374-375. 
 1895a.   Description   of   a   new   Batrachian 
(Oreophryne quelchii) discovered by Messrs. J. J. 
Quelch  and F.  McConnell on the summit of 
Mount Roraima. Ann. Mag. Nat. Hist., Ser. 6, 
15(90): 521-522. 
 1895b. Correction to p. 521  ('Annals', June 
1895). Ann. Mag. Nat. Hist., Ser. 6, 16(91): 125. 
 1898. An account of the reptiles and batra- 
chians collected by Mr. W. F. H. Rosenberg in 
western Ecuador. Zool. Soc, London, Proc. 1897: 
107-126, plates 10-18. 
 1900. Batrachians. In E. Ray Lankester. Re- 
port on a collection made by Messrs. F. V. Mc- 
Connell and J. J. Quelch at Mount Roraima in 
British Guiana. Linn. Soc, London, Trans., Ser. 
2, Zool., 8(2): 55-56, plate 5. 
 1902.  Descriptions of new batrachians  and 
reptiles from the Andes of Peru and Bolivia. 
Ann. Mag. Nat. Hist., Ser. 7,  10(59): 394-402. 
 1903. Descriptions of new batrachians in the 
British Museum. Ann. Mag. Nat. Hist., Ser. 7, 
12(71): 552-557. 
 1905. Descriptions of new tailless batrachians 
in the collection of the British Museum. Ann. 
Mag. Nat. Hist., Ser. 7, 16(92): 180-184. 
 1914. On a second collection of batrachians 
and reptiles made by Dr. H. G. F. Spurrell, F. 
Z. S., in the Choco, Colombia. Zool. Soc, London, 
Proc. 1914(3): 813-817, plates 1-2. 
BRATTSTROM, B. H. 1957. The phylogeny of the 
Salientia based on skeletal morphology. Syst. 
Zool. 6(2): 70-74. 
BROCCHI, M. P. 1878. Observations sur les genres 
Atelopus et Phryniscus et sur le genre Ollotis 
(Cope). Soc. Philomatique Paris, Bull. Ser. 7, 
2:96-97. 
 1881-1883. fitude des Batraciens de l'Ameri- 
que Centrale. Mission Scientifique au Mexique 
et dans l'Amerique Centrale, Recherches Zoo- 
logique, Pt. 3, sec. 2: 1-122, plates 1-21. 
BUCHHOLZ, R., AND W. PETERS, 1875. Hr. W. Peters 
las tiber die von Herr Professor Dr. R. Buchholz 
in Westafrica gesammelten Amphibien. Akad. 
Wiss., Berlin, Monats., 1875: 196-212. 
BUDGETT, J. S. 1899. Notes on the batrachians of the 
Paraguayan Chaco, with observations upon their 
breeding habits and development, especially with 
regard to Phyllomedusa hypochondrialis, Cope. 
Also a description of a new genus. Quart. J. 
Microscop. Sci., new ser. 42(167): 305-333. 
CARVALHO, A. L. de. 1949. Notas sobre os habitos de 
"Dendrophryniscus brevipollicatus" Espada (Am- 
phibia, Anura). Rev. Brasil. Biol. 9(2): 223-227. 
CHILDS, O. E., AND B. W. BEEBE. 1963. Backbone of 
the Americas. Tectonic history from pole to pole. 
Amer. Assoc Petrol. Geol., Mem. 2: 1-320. 
COCHRAN, D. M. 1948. A new frog from Para, Brazil. 
Mus. Royal Hist. Nat., Belgique, Bull. 24(24): 
1-4. 
 1961. Living amphibians of the world. Hamish 
Hamilton Ltd., London. 199 p. 
COCHRAN, D. M., AND C. J. GOIN. 1970. Frogs of 
Colombia. U.S. Nat. Mus., Bull. 288: 1-655. 
COLE, C. J., C. H. LOWE, AND J. W. WRIGHT. 1968. 
Karyotypes of eight species of toads (Genus 
Bufo) in North America. Copeia 1968(1): 96- 
100. 
62 BULLETIN OF LOS ANGELES COUNTY MUSEUM OF NATURAL HISTORY No. 12 
COLIFAX, A.  1956. New information on the Corro- 
boree frog, Pseudophryne corroborae (Moore). 
Linn.  Soc.  New South Wales,  Proc.  80(379): 
258-266. 
COPE, E. D. 1864. On the limits and relations of the 
Raniformes. Acad. Nat. Sci. Philadelphia, Proc. 
1864, 16: 181-183. 
 1865. Sketch of the primary groups of Batra- 
chia Salientia. Nat. Hist. Rev. 1865(17): 97-120. 
 1866. On the structures and distribution of the 
genera of arciferous Anura. J. Acad. Nat. Sci. 
Philadelphia, Ser. 2, 6(2): 67-112. 
 1867. On the families of the raniform Anura. 
J. Acad. Nat. Sci. Philadelphia, Ser. 2, 6(2): 
189-200. 
 1868.  An  examination  of the  Reptilia and 
Batrachia obtained by the Orton Expedition to 
Equador and the Upper Amazon, with notes on 
other species. Acad. Nat. Sci. Philadelphia, Proc. 
1868, 20: 96-140. 
 1871. Ninth contribution to the herpetology 
of Tropical America. Acad. Nat. Sci. Philadel- 
phia, Proc. 1871, 23: 200-224. 
 1874. On some Batrachia and Nematognathi 
brought from the Upper Amazon by Prof. Orton. 
Acad.   Nat.  Sci.  Philadelphia,  Proc.   1874,  26: 
120-137. 
..  1887. Catalogue of Batrachians and reptiles 
of Central America and Mexico. U. S. Nat. Mus., 
Bull. 32: 1-98. 
CORNALIA, E. 1849. Vertebratorum synopsis in Museo 
Mediolanense extantium quae per novam orbem 
Cajetanus Osculati collegit annis 1846-47-48. 
16 p., 1 plate, (only seen in G. Osculati, 1850. 
Esplorazione delle regioni equatoriali lungo il 
napo ed il fiume delle Amazzoni frammento di 
un viaggio tatto nelle due Americhe negli anni 
1846-48. Milano, 320 p.). 
COTT, H.  B.   1957. Adaptive coloration in animals. 
Methuen and Co., Ltd., London. 500 p., plates 
1-48. 
DALY, J. W., AND C. W. MYERS. 1967. Toxicity of 
Panamanian Poison Frogs (Dendrobates):  some 
biological   and   chemical   aspects.   Science   156 
(3777): 970-973. 
DARLINGTON, P. J., JR. 1957. Zoogeography: the geo- 
graphical  distribution   of  animals.   John  Wiley 
and Sons, Inc. New York. 675 p. 
DAVIS, D. D. 1935. A new generic and family position 
for Bufo borbonica. Field Mus. Nat. Hist., Zool. 
Ser. 20(12): 87-92. 
 1936. The distribution of Bidder's organ in 
the Bufonidae. Field Mus. Nat. Hist., Zool. Ser. 
20(15): 115-125. 
DEMAN, J. G. 1874. Myologie comparee de l'extre- 
mite  posterieure  chez  les  Amphibies.  Niederl. 
Arch. Zool. II: 53-88. 
DONOSO BARROS, R. 1969. Una nueva especie de Ate- 
lopodidae de Bolivia  (Anura). Physis 28(77): 
327-330. 
DUBOIS, R. 1947. On the distribution of Bidder's organ 
in bufonids. Zool. Mededeelingen Rijksmuseum 
Nat. Hist., Leiden, 28:275-279. 
DUELLMAN, W. E., 1967. Social organization in the 
mating calls of some Neotropical anurans. Amer. 
Midland Nat. 77(1): 156-163. 
DUELLMAN, W. E., AND J. D. LYNCH. 1969. Descrip- 
tions of Atelopus tadpoles and their relevance to 
Atelopodid classification. Herpetologica 25(4): 
231-240. 
DUGES, A. 1835. Recherches sur l'osteologie et la 
myologie des Batraciens a leurs differens ages. 
Acad. Sci., Paris, Mem. 6: 1-216, plates 1-18. 
DUMERIL,  A.  M.  C, AND G.  BIBRON.   1841.  Erpe- 
tologie generate ou histoire naturelle complete 
des reptiles. Paris. Vol. 8, 792 p. 
DUNLAP, D. G.  1960. The comparative myology of 
the pelvic appendage in the Salienta. J. Morphol. 
106: 1-76. 
 1966. The development of the musculature of 
the hindlimb in the frog, Rana pipiens. J. Morphol. 
119(2): 241-258. 
DUNN, E. R. 1931. New frogs from Panama and Costa 
Rica. Boston Soc. Nat. Hist., Occ. Papers 5: 
385-401. 
 1933. Amphibians and reptiles from El Valle 
de Anton, Panama. Boston Soc. Nat. Hist., Occ. 
Papers 8: 65-79. 
Du TOIT, G. P., AND C. G. S. DE VILLIERS. 1932. Die 
skedelmorphologie van Hyperolius horstockii as 
voorbeeld van die Polypedatidae. South African 
J. Sci. 29: 449-465. 
EATON,  T.  H,  JR.   1959.  The  ancestry  of  modern 
Amphibia: a review of the evidence. Univ. Kans. 
Publ. Mus. Nat. Hist. 12(2): 155-180. 
EISELT, J. 1941. Der Musculus opercularis und die 
mittlere Ohrsphare der anuren Amphibien. Arch. 
Natur., new ser., 10(2): 179-230. 
ELIAS, H., AND J. SHAPIRO. 1957. Histology of the skin 
of some toads and frogs. Amer. Mus. Novitates 
1819: 1-27. 
ESTES, R., AND R. WASSERSUG. 1963. A Miocene toad 
from   Colombia,  South  America.   Mus.   Comp. 
Zool., Harvard Univ., Breviora 193: 1-13. 
FERNANDEZ, K. 1927. Sobre la biologia y reproduccion 
de Batracios Argentinos. (Segunda Parte). Bol. 
Acad. Nac. Cienc. Cordoba, 29: 271-328. 
FITZINGER, L. I. 1843. Systema reptilium. Vienna. 
106 p. 
FUHRMAN, F. A., G. J. FUHRMAN, AND H. S. MOSHER. 
1969. Toxin from skin of frogs of the genus 
Atelopus: differentiation from dendrobatid toxin. 
Science 165: 1376-1377. 
GADOW, H. 1901. Amphibia and reptiles. Macmillan 
and Co., Ltd., England. 668 p. 
GALLARDO, J. M. 1961a. La ubicacion sistematica y 
distribucion geografica de Brachycephalidae 
Argentinos. Reunion Trabajos Comunicaciones 
Cienc. Nat. Geogr. Argentino, Univ. Nac. Litoral. 
Santa Fe, 1961:205-212. 
 1961b.  Nuevo  genero de  Brachycephalidae. 
(Amphibia Anura). Neotropica 7(24): 71-72. 
_. 1966. Zoogeografia de los anfibios Chaquefios. 
Physis 26(71): 67-81. 
GAUPP, E. 1896. A. Ecker's und R. Wiedersheim's Ana- 
tomie des Frosches. Dritte Auflage, Erste Abtei- 
1971 MORPHOLOGY AND EVOLUTION OF NEOTROPICAL FROGS 63 
lung. Skelet und Muskel. Braunschweig. 264 p. 
GOIN, C. J. 1960. Amphibians, pioneers of terrestrial 
breeding habits. Smithsonian Rep. 1959: 427-445. 
GOIN, C. J., AND O. B. GOIN. 1962. Introduction to 
Herpetology. W. H. Freeman and Co., San Fran- 
cisco. 341 p. 
GRAY, J. E. 1825. A synopsis of the genera of Reptiles 
and Amphibia, with a description of some new 
species. Ann. Phil., Ser. 2, 10: 193-217. 
GREGORY, W. K. 1951. Evolution emerging. Macmillan 
Co., N. Y. vol. 1, 736 p. 
GRIFFITHS, I. 1954a. On the nature of the fronto- 
parietal in Amphibia, Salientia. Zool. Soc, Lon- 
don, Proc. 123: 781-792. 
 1954b.  On  the  otic element in Amphibia, 
Salientia. Zool. Soc, London, Proc. 124: 35-50. 
 1959. The phylogeny of Sminthillus limbatus 
and the status of the Brachycephalidae (Amphi- 
bia, Salientia). Zool. Soc, London, Proc. 132: 
457-487. 
 1963. The phylogeny of the Salientia. Biol. 
Rev. 38: 241-292. 
GUNTHER, A. 1858a. On the systematic arrangement 
of the tailless Batrachians and the structure of 
Rhinophrynus dorsalis. Zool. Soc. London, Proc. 
26: 339-352. 
 1858b. Catalogue of the Batrachia Salientia 
in the collection of the British Museum. British 
Mus., London, xvi, 160 p., 12 plates. 
1900.  Biologia Centrali-Americana:   Reptilia 
and Batrachia:   1885-1902:  i + XX:  326 p., 76 
plates. 
HAFFER, J. 1967. Zoogeographical notes on the "non- 
forest"   lowland   bird   faunas   of   northwestern 
South America. El Hornero 10(4): 315-333. 
1969. Speciation in Amazonian forest birds. 
Science  165 (3889): 131-137. 
HAM, C. K., AND L. J. HERRERA, JR. 1963. Role of sub- 
andean fault system in tectonics of eastern Peru 
and Ecuador. In O. E. Childs and B. W. Beebe 
(eds.), Backbone of the Americas. Tectonic his- 
tory from pole to pole. Amer. Assoc. Petrol. 
Geol., Mem. 2: 47-61. 
HARRINGTON, H. J. 1962. Paleogeographic develop- 
ment of South America. Amer. Assoc. Petrol. 
Geol., Bull. 46(10): 1773-1814. 
HECHT, M. K. 1962. A reevaluation of the early 
history of the frogs. Part I. Syst.Zool. 11(1): 
39-44. 
1963. A reevaluation of the early history of 
the frogs. Part II. Syst. Zool. 12(1): 20-35. 
INGER, R. F. 1954. Systematics and zoogeography of 
Philippine Amphibia. Field Mus. Nat. Hist., 
Fieldiana Zool. 38(4): 181-531. 
 1966. The systematics and zoogeography of 
the Amphibia of Borneo. Field Mus. Nat. Hist., 
Fieldiana Zool. 52: 1-402. 
? 1967. The  development  of  a  phylogeny  of 
JACOBS, C, H. BURGL, AND D. L. CONLEY. 1963. Back- 
bone of Colombia. In O. E. Childs and B. W. 
Beebe (eds.), Backbone of the Americas. Tectonic 
history from pole to pole. Amer. Assoc Petrol. 
Geol., Mem. 2: 62-72. 
JAMESON, D. L. 1957. Life history and phylogeny in 
the salientians. Syst. Zool. 6(2): 75-78. 
JARVIK, E. 1968. Aspects of vertebrate phylogeny. In 
T. Orvig (ed.), Current problems of lower verte- 
brate phylogeny. Nobel Symposium 4. Almqvist 
& Wiksell, Stockholm, p. 497-527. 
JIMENEZ DE LA ESPADA, M. 1870. Zoologia. 1. Faunae 
neotropicalis species quaedam nondum cognitae. 
Amphibia. J. Sci. Math. Phys. Natur. Acad. Real 
Lisboa, second edition, 9: 57-65. 
 1875. Vertebrados del viaje al Pacifico: Batra- 
cios. Madrid. 208 p., 6 plates. 
KEFERSTEIN, W. 1867. Ueber einige neue Oder seltene 
Batrachier aus Australien und dem tropischen 
Amerika. Akad. Wiss. Gottingen, Nachrichten, 
1867: 341-361. 
 1868a. Beschreibung einiger neuen Batrachier 
aus Australien und Costarica. Akad. Wiss. Got- 
tingen, Nachrichten, 1868: 326-332. 
- 1868b. Ueber einige Batrachier aus Costarica. 
frogs. Evolution 21(2): 369-384. 
IZECKSOHN, E. 1968. Nova especie de "Dendrophryn- 
iscus" do estado do Rio de Janeiro (Amphibia, 
Salientia). Rev. Brasil Biol. 28(4): 357-362. 
Arch. Natur. 34(1): 291-300. 
KLAPPENBACH, M. A. 1968. Notas herpetologicas, IV. 
El genero Melanophryniscus (Amphibia, Salien- 
tia) en el Uruguay con descripcion de dos nuevas 
especies. Comun. Zool. Mus. Hist. Natur. Monte- 
video 9(118): 1-17. 
KLUGE, A. G. 1967. Higher taxonomic categories of 
gekkonid lizards and their evolution. Amer. Mus. 
Nat. Hist., Bull. 135: 1-59. 
KLUGE, A. G., AND J. S. FARRIS. 1969. Quantitative 
phyletics and the evolution of Anurans. Syst. Zool. 
18(1): 1-32. 
LANKESTER, E. R. 1900. Report on a collection made 
by Messrs. F. V. McConnell and J. J. Quelch 
at Mount Roraima in British Guiana. Linn. Soc. 
London, Trans., Ser. 2, Zool. 8(2): 51. 
LATSKY, L. 1930. A synopsis of researches relating 
to the validity of the South African Brevicepiti- 
dae as an autonomous family of the Anura. South 
African J. Sci. 27:451-468. 
LAURENT, R. 1942. Note sur les procoellens firmister- 
nes (Batrachia Anura). Mus. Royal Hist. Nat., 
Belgique, Bull. 18(43): 1-20. 
LlCHTENSTEIN, M. H. C, AND E. VON MARTENS. 1856. 
Nomenclator Reptilium et Amphibiorum Musei 
Zoologischi Berolinensis. Berlin. 48 p. 
LIEM, S. S. 1970. The morphology, systematics, and 
evolution of the Old World treefrogs (Rhaco- 
phoridae and Hyperoliidae). Fieldiana Zool., 
Chicago 57: 1-145. 
LIMESES, C. E. 1964. La musculatura del muslo en 
los ceratofrinidos y formas afines con un analisis 
critico sobre la significacion de los caracteres mio- 
logicos en la sistematica de los Anuros superiores. 
Univ. Buenos Aires, Fac. Cienc. Exactas Natur., 
Contrib. Cient., Ser. Zool. 1(4): 193-245. 
 1965. La musculatura mandibular en los cera- 
tofrinidos y formas afines (Anura, Ceratophryn- 
idae). Physis 25(69): 41-58. 
64 BULLETIN OF LOS ANGELES COUNTY MUSEUM OF NATURAL HISTORY No. 12 
LLOYD, J. J. 1963. Tectonic history of the South Cen- 
tral-American orogen. In 0. E. Childs and B. W. 
Beebe (eds.), Backbone of the Americas. Tectonic 
history from pole to pole. Amer. Assoc. Petrol. 
Geol., Mem. 2: 88-100. 
LUTZ, A. 1932. Sur la biologie des batraciens du Bra- 
sil. Soc. Biol., Paris, Compt. Rend. 109(9): 755- 
756. 
LUTZ, A., AND G. R. KLOSS. 1952. Short notes on some 
frogs from the upper Amazons and a few vicar- 
iant forms. Inst. Oswaldo Cruz, Mem. 50: 629- 
678. 
MAYR, E., AND W. H. PHELPS, JR. 1967. The origin of 
the bird fauna of the south Venezuelan highlands. 
Amer. Mus. Nat. Hist, Bull. 136: 269-328. 
MCDIARMID, R. W. 1971. The systematic status of the 
frog Atelopus rubriventris Vellard. Physis (in 
press). 
MELIN, D. 1941. Contributions to the knowledge of 
the Amphibia of South America. Goteborgs 
Kungl. Vetenskaps, och Vitterh. Samh. Handl. 
Sjatte fol., Ser. B, 1(4): 1-71. 
MENCHER, E. 1963. Tectonic history of Venezuela. 
In O. E. Childs and B. W. Beebe (eds.), Back- 
bone of the Americas. Tectonic history from pole 
to  pole.  Amer.  Assoc.  Petrol.  Geol.  Mem.  2: 
73-87. 
MIRANDA-RIBEIRO, A. DE. 1920. Os Brachycephalideos 
do Museu Paulista. Mus. Paulista, Sao Paulo, 
Rev. 12(2): 305-315, 3 plates. 
MIVART, S. 1869. On the classification of the anurous 
batrachians. Zool. Soc, London, Proc. 1869: 
280-295. 
MOOKERJEE, H. K. 1931. On the development of the 
vertebral column of Anura. Royal Soc, London, 
Phil. Trans., Ser. B, 219: 165-196. 
 1936.   The   development   of   the   vertebral 
column and its bearing on the study of organic 
evolution.  Proc.   Indian  Sci.  Congr.  Zool.  23: 
307-343. 
MOOKERJEE, H. K., AND S. K. DAS. 1939. Further in- 
vestigation on the development of the vertebral 
column in Salientia (Anura). J. Morphol. 64: 167- 
209. 
MUNTZ, W. R. A. 1964. Vision in frogs. Sci. Amer. 
210(3): 110-119. 
NICHOLLS, G. E. 1916. The structure of the vertebral 
column in the Anura Phaneroglossa and its im- 
portance as a basis of classification. Linn. Soc, 
London, Proc. 1915-1916(128): 80-92. 
NOBLE, G. K. 1921. Five new species of Salientia from 
South America. Amer. Mus. Novitates 29: 1-7. 
 1922. The phylogeny of the Salientia. I. The 
osteology and thigh musculature; their bearing 
on classification and phylogeny. Amer. Mus. Nat. 
Hist., Bull. 46: 1-87. 
 1926. The pectoral  girdle of the  brachyce- 
phalid frogs. Amer. Mus. Novitates 230: 1-14. 
 1927. The value of life history data in the 
study of the evolution of the Amphibia. New 
York Acad. Sci., Ann. 30: 31-128. 
 1931. Biology of the Amphibia. McGraw-Hill, 
New York. 577 p. 
NUSSBAUM, M. 1898. Nerv and Muskel. II. Der Ober- 
schenkel einiger anuren Batrachien. Arch. Mikr. 
Anta. 52: 367-501. 
ORTON, G. L. 1957. The bearing of larval evolution 
on some problems in frog classification. Syst. 
Zool. 6(2): 79-86. 
PARKER, H. W. 1927. A revision of the frogs of the 
genera Pseudopaludicola, Physalaemus, and Pleu- 
rodema. Ann. Mag. Nat. Hist., Ser. 9, 20(118): 
450-478. 
 1932.  Parallel modifications in the skeleton 
of the Amphibia Salientia. Arch. Zool. Italiano 
16: 1239-1248. 
 1934. A monograph of the frogs of the family 
Microhylidae. British Museum, London. 208 p. 
1940.  The  Australian  frogs   of  the   family 
Leptodactylidae. Novitates Zool. 42(1): 1-106. 
PARKER, W. K. 1868. A monograph on the structure 
and development of the shoulder-girdle and ster- 
num in the Vertebrata. Robert Hardwicke, Lon- 
don, 237 p., 30 plates. 
 1876.  On the structure and development of 
the skull in the Batrachia. Part II. Royal Soc, 
London, Phil. Trans., part 2, 166(24): 601-669, 
9 plates. 
_. 1882. On the structure and development of 
the skull in the Batrachia. Part III. Royal Soc, 
London, Phil. Trans., part 1, 172(1): 1-266, 44 
plates. 
PERACCA, M. G. 1904. Viaggio del Dr. Enrico Festa 
nell'Ecuador e regioni vicine. Rettili ed Amfibii. 
Boll. Mus. Zool. Anat., Univ. Torino, 19(465): 
1-41. 
PERRIN, A. 1893. Contributions a l'etude de la myo- 
logie comparee: membre posterieur chez un cer- 
tain nombre de batraciens et de sauriens. Bull. 
Sci. France, Belgique, 24: 1-181. 
PETERS, J. A. 1964. Dictionary of herpetology. Hag- 
ner Publishing Co., New York. 392 p. 
PETERS, W. 1873. trber die von Spix in Brasilien 
gesammelten Batrachier des Konigl. Naturalien- 
kabinets zu Miinchen. Akad. Wiss., Berlin, 
Monats.  1872: 196-227. 
PHILIPPI, R. A. 1894. Phryniscus Bibron ist nicht 
Phryniscus Wiegmann. Arch. Natur., Berlin, 60 
(1): 214-216. 
PONSE, K. 1924. L'organe de Bidder et le determin- 
isme des caracteres sexuels secondaires du Cra- 
paud  (Bufo vulgaris L.). Rev. Suisse Zool.  31 
(7): 177-336. 
RAMASWAMI, L. S. 1937. The morphology of the bufo- 
nid head. Zool. Soc, London, Proc. 106(4): 
1157-1169. 
_. 1939. Some aspects of the anatomy of Anura 
(Amphibia) ? a review. Indian Acad. Sci., Proc. 
10(1): 41-80. 
RAU, A. S., AND J. B. GATENBY. 1923. Notes on the 
distribution, morphology and cytology of the 
organ of Bidder. J. Royal Microscop. Soc, Trans. 
1923(2): 19-36. 
REIG, O. A. 1958. Proposiciones para una nueva 
macrosistematica de los anuros. Physis 21(60): 
109-118. 
1971 MORPHOLOGY AND EVOLUTION OF NEOTROPICAL FROGS 65 
REINBACH, W. 1939. Untersuchungen iiber die Ent- 
wicklung des Kopfskelettes von Calyptocephalus 
gayi. Jena Natur. 72(3): 211-362. 
RITLAND, R. M. 1955a. Studies on the post-cranial 
morphology of Ascaphus truei. I. Skeleton and 
spinal nerves. J. Morphol. 97(1): 119-178. 
1955b. Studies on the post-cranial morphology 
of Ascaphus truei. II. Myology. J. Morphol. 97 
(2): 215-282. 
RIVERO, J. A. 1961. Salientia of Venezuela. Mus. 
Comp. Zool., Harvard Univ., Bull. 126(1): 1- 
207, 1 plate. 
1963.   Five  new   species  of  Atelopus from 
Colombia with notes on other forms from Co- 
lombia and Ecuador. Carib. J. Sci. 3(2,3): 103- 
124. 
ROMER, A. S. 1962. The vertebrate body. Third edi- 
tion. W. B. Saunders Co., Philadelphia. 627 p. 
RUTHVEN, A. G. 1916. Description of a new species 
of Atelopus from the Santa Marta Mountains, 
Colombia. Mus. Zool., Univ. Michigan, Occ. 
Papers 28: 1-3, 1 plate. 
-  1922.   The   amphibians   and  reptiles  of  the 
Sierra Nevada de Santa Marta, Colombia. Mus. 
Zool., Univ. Michigan, Misc. Publ. 8: 1-69, 12 
plates. 
SANDERS, O. 1953. A new species of toad, with a dis- 
cussion of morphology of the bufonid skull. Her- 
petologica 9(1): 25-47. 
SAVAGE, J. M. 1967. An extraordinary new toad (Bufo) 
from Costa Rica. Rev. Biol. Trap. 14(2): 153- 
167. 
SAVAGE, J. M., AND A. G. KLUGE. 1961. Rediscovery 
of the strange Costa Rica toad, Crepidius epioticus 
Cope. Rev. Biol. Trop. 9(1): 39-51. 
SCHAEFFER, B. 1949. Anurans from the early Tertiary 
of Patagonia. Amer. Mus. Nat. Hist., Bull. 93: 
41-68. 
SCHMIDT, O. 1857. Diagnosen neuer Frosche des zoo- 
logischen Cabinets zu Krakau. Akad. Wiss., Wien, 
Sitzungsb., 24(1): 10-15. 
 1858.  Deliciae  Herpetologicae  Musei  Zoo- 
logici Cracoviensis. Akad. Wiss., Wien, Denk- 
schr. 14(2): 237-258. 3 plates. 
SEDRA, S. N. 1949. On the homology of certain ele- 
ments in the skull of Bufo regularis Reuss (Sa- 
lientia). Zool. Soc, London, Proc. 119: 633-641. 
SEDRA, S. N., AND M. I. MICHAEL. 1959. The onto- 
genesis of the sound conducting apparatus of the 
Egyptian toad, Bufo regularis Reuss, with a 
review of this apparatus in Salientia. J. Morphol. 
104(2): 359-376. 
SEXTON, O. 1958. Observations on the life history of 
a Venezuelan frog, Atelopus cruciger. Acta Biol. 
Venezuelica 2(21): 235-242. 
SHREVE, B. 1936. A new Atelopus from Panama and 
a new Hemidactylus from Colombia. Boston 
Soc. Nat. Hist., Occ. Papers 8: 269-271. 
SIMPSON, G. G. 1949. The meaning of evolution. Yale 
Univ. Press, New Haven. 364 p. 
 1953. The major features of evolution. Co- 
lumbia Univ. Press, New York. 434 p. 
SMITH, A. C, AND I. M. JOHNSON. 1945. A phytogeo- 
graphic sketch of Latin America. In F. Verdoorn 
(ed.), Plants and plant science in Latin America. 
Ronald Press Co., New York, p. 11-18. 
Sprx, J. B. VON. 1824. Animalia nova sive species 
novae testudinum et ranarum, quas in itinere per 
Brasiliam, annis 1817-20 . . . collegit et descripsit 
... Monachi. 53 p., 22 plates. 
STARRETT, P. 1960. A redefinition of the genus Smi- 
lisca. Copei 1960(4): 300-304. 
 1967. Observations on the life history of frogs 
of the family Atelopodidae.  Herpetologica 23 
(3): 195-204. 
1968. The phylogenetic significance of the jaw 
musculature in anuran amphibians. Ph.D. Disser- 
tation, Univ. Michigan, 179 p. 
STOHLER, R. 1932. Das vorkommen des potentiellen 
ovars bei den Bufonidae. Natur. Ges., Basel, 
Verhandl. 42(1930-31): 196-210. 
TAYLOR, E. H. 1952. A review of the frogs and toads 
of Costa Rica. Univ. Kansas Sci. Bull. 35, pt. 
1(5): 577-942. 
THROCKMORTON, L. H. 1968. Concordance and dis- 
cordance of taxonomic characters in Drosophila 
classification. Syst. Zool. 17(4): 355-387. 
TIHEN, J. 1960. Two new genera of African bufonids, 
with remarks on the phylogeny of related genera. 
Copeia 1960(3): 225-233. 
 1962a.    Osteological   observations   on   new 
world Bufo. Amer.  Midi. Nat. 67(1): 157-183. 
 1962b. A review of new world fossil bufonids. 
Amer. Midi. Nat. 68(1): 1-50. 
1965.   Evolutionary  trends  in  frogs.   Amer. 
Zool. 5:309-318. 
TREWAVAS, E.  1933. The hyoid and larynx of the 
Anura. Royal Soc, London, Phil. Trans., Ser. B, 
222: 401-527. 
TSCHUDI, J. J. 1838. Classification der Batrachier, mit 
berucksichtigung  der  fossilen  thiere.   Soc.   Sci. 
Natur. Neuchatel 2: 1-99, 6 plates. 
TUMARKIN, A. 1955. On the evolution of the auditory 
conducting apparatus: a new theory based on 
functional  considerations.  Evolution  9(3): 221- 
243. 
VANZOLINI, P. E. AND E. E. WILLIAMS. 1970. South 
American anoles: the geographic differentiation 
and evolution of the Anolia chrysolepis species 
group (Sauria, Iguanidae). Arq. Zool., Sao Paulo 
19(1-2): 1-124, 5 pis. 
VELLARD, J. 1947. Un neuvo batracio del norte Argen- 
tino. Acta Zool. Lilloana 4(1947): 115-119. 
WAGLER, J. G. 1828. Auziige aus seinem Systema am- 
phibiorem. Isis von Oken, cols. p. 740-744. 
 1830. Natiirliches System der Amphibien, mit 
vorangehender Classification der Saugthiere und 
Vogel. J. G. Cotta'schen Buchhandlung, Mun- 
chen, Stuttgart und Tubingen. 354 p. 
WAKE, D. B. 1966. Comparative osteology and evolu- 
tion of the lungless salamanders, family Pletho- 
dontidae. Southern Calif. Acad. Sci., Mem. 4: 
1-111. 
WAKE, M. H. 1968. The comparative morphology 
and evolutionary relationships of the urogenital 
66 BULLETIN OF LOS ANGELES COUNTY MUSEUM OF NATURAL HISTORY No. 12 
system of caecilians.  Ph.D. dissertation, Univ. 
Southern California. 279 p. 
WERNER, F. 1899. Ueber Reptilien und Batrachier aus 
Columbien und Trinidad. Zool Botan. Ges., Wien, 
Verhandl. 1899(49): 470-484. 
WEYENBERGH, H. 1875. Apuntes pequefios. Soc. Zool., 
Argentina, Periodico Zool. 1(4): 331-333. 
WIEGMANN, A. F. A.  1834. Beitrage zur zoologie, 
gesammelt auf einer reise urn die erde von Dr. 
F. J. F. Meyen. Part 7. Ampbibien. Deutsche 
Akad. Naturforsch., Nova Acta Leopoldina 9: 
185-268. 
WITSCHI, E. 1933. Studies in sex differentiation and 
sex determination in amphibians. VI. The nature 
of Bidder's organ in the toad. Amer. J. Anat. 
52(3): 461-515. 
Accepted for publication December 15, 1970