'tilt HARVARD UNIVERSITY Library of the Museum of Comparative Zoology 0^ I MUS. CO MP. ZOOL. «g / 1^% LIBRARY OCCASIONAL PAPERS AUG 1 2 1Q74 of the HARVARD MUSEUM OF NATURAL HISTORY The University of Kansas Lawrence, Kansas NUMBER 29, PAGES 1-60 JULY 19, 1974 SYSTEMATIC RELATIONSHIPS OF NEOTROPICAL HORNED FROGS, GENUS HEMIPHRACTUS (ANURA: HYLIDAE) By Linda Trueb^ In the minds of all but the most resolute and devoted systematic biologists, taxonomic study frequently is viewed as a necessary tedium, economically dispatched so that one can move on to more interesting aspects of the biology of an organism. However, occa- sionally one chances upon an animal that because of its literary antiquity, bizarre appearance and habits, and questionable rela- tionships is of intrinsic interest. Such is the case of the extraordi- nary horned frogs of the genus HemipJiractus. These Neotropical frogs are denizens of undisturbed cloud forests, montane and low- land tropical rainforests from Panama south to Bolivia. Mating calls are unknown in the group, as are breeding congregations. Solitary individuals are found perched on low vegetation in the forest at night, and, only rarely, a fortunate collector has found a female carrying eggs or young upon her back. Although these frogs have been known for 150 years, onl\' about 200 preser\ed specimens exist, and more than 50 percent of these ha\'e been obtained in the course of extensive field work in Panama, Ecuador, and Peru within the last eight years. The familial association of this genus has been subject to constant debate since the first species was named in 1824. The morphological pecuHarities have been described, re- described, and largely misinterpreted for an equal length of time. And, \irtually nothing is known about the breeding biology, habits, or population structure of these frogs. In view of the uniqueness 1 Adjunct Curator, Di\ision of Herpetolog>-, Museum of Natural History, and Department of Systematics and Ecology, The Unixersity of Kansas. 2 OCCASIONAL PAPERS MUSEUM OF NATURAL HISTORY and relative rarity of Hemiphracttis and the apparent restriction of its species to the rapidly diminishing, undisturbed tropical forests, I feel that it is appropriate to review the systematic status of these unusual frogs. Hopefully, this constitutes only a preface to dis- covery and subsequent study of other aspects of their biology. The specific objectives of this study are threefold: (1) to define the genus Hemiphractus and the various species that properly are allocated to it; ( 2 ) to assess the intrageneric relationships and prob- able phylogenetic history of the group; and (3) to discuss the possible intergeneric relationships of this anuran genus. A historical resume provides a synopsis of the taxonomic history of these frogs. Morphological characters judged to be of systematic importance are discussed in an analysis of characters. Finally, wdthin each species account, I have included available information concerning the habitat, habits, food, color in life, and reproduction. MATERIALS AND METHODS Specimens examined include 182 preserved frogs inclusive of the type specimens of all nominal taxa except Hemiplir actus scutatus, for which the type no longer exists. Osteological observations were based on 13 dry skeletal preparations and one cleared and stained specimen; these were supplemented by radiographs of three addi- tional individuals. I have observed living individuals of all species except Hemiphractus johnsoni and H. scutatus. Colored photo- graphs were available for all species. Specimens are identified throughout the text by their catalogue numbers preceded by the appropriate museum abbreviation, as follows: AMNH American Museum of Natural History ANSP Academy of Natural Sciences of Philadelphia AUM Auburn University Museum BMNH British Museum (Natural History) BYU Brigham Young University CAS California Academy of Sciences CAS-SU Stanford University Collection ( In California Academy of Sciences) FMNH Field Museum of Natural History KU The University of Kansas Museum of Natural History LACM Los Angeles County Museum of Natural History MCZ Museum of Comparative Zoology, Harvard University MNCN Museo Nacional de Ciencias Naturales, Madrid NHRM Naturhistoriska Riksmuseet, Stockholm UIMNH University of Illinois Museum of Natural History UMMZ University of Michigan Museum of Zoology USC-CRE University of Southern California — Costa Rica Expedition USNM United States National Museum (=National Museum of Natural History) ZSM Zoologisches Sammlung Miinchen Because Hemiphractus lack any ob\'ious, external secondary sexual characteristics, the sexes of all adult and subadult specimens RELATIONSHIPS OF NEOTROPICAL HORNED FROGS Fig. I. Schematic dorsal and lateral views of skull of Hemiphractus scutatus showing various cranial dimensions measured. Skull approximately natural size; subtemporal fenestra indicated by hatching. Abbrexiations: HL, head length; HW, head width; PP, posterior projection of paraoccipital process. See text for explanation of subscript notations. were determined by examining the gonads. As a result, frogs less than 30 mm snoiit-vent length were classified as juveniles, whereas those with poorly developed gonads and snout-\ent lengths less than the observed ranges for sexually mature individuals (see Table 2) were classified as subadults. Adult females were divided into three classes depending upon the size of ovarian eggs which varied from minute (less than 1 mm diameter) to moderate-sized (between 1 and 3 mm diameter) to large (greater than 3 mm diameter). In several individuals the oviducts were distended but empty; usually such a female either had eggs or young associated with her or showed evidence (depressions on the back) of having recently carried young. When the information was available, date (month) of specimen collection was plotted with respect to repro- ductive state to determine what, if any, reproducti\'e cycles seem to prevail. These data are summarized in the Remarks section of each species account. Although most measurements taken on these frogs are standard and self-explanatory, a number of additional cranial dimensions have been recorded that require brief explanation. These are shown schematically in figure 1. Standard head length (HLi) was measured from the angle of the jaw to the medial articulation of the premaxillaries; unless otherwise noted, all references to head length denote this measurement. Medial head length ( HLo ) repre- sents the distance from the posteromedial articulation of the fronto- parietals to the medial articulation of the premaxillaries anteriorly. Lateral head length (HL3) is a measurement from the distal tip 4 OCCASIONAL PAPERS MUSEUM OF NATURAL HISTORY of the paraoccipital process to the premaxillaries. The length of the paraoccipital process projection (PP) beyond the angle of the jaw was computed as follows: (HLo-HLi)/HLi. Standard head width (HWi) represents the greatest head width measured across the maxillary arch; unless otherwise noted, all references to head width denote this measurement. The distance between the distal tips of the paraoccipital processes is noted as HWo. A number of new and/or special names or phrases are employed in this paper to describe certain cranial features. The most impor- tant of these and their definitions are listed below. Most, together with the general structures of the hemiphractine skull, are illustiated and labeled in figure 2. 1. Postorbital indentation. — (Fig. 2a); the emargination of the postero- dorsal rim of the orliit. 2. Paraoccipital process. — (Fig. 2a, h, c); the posterior, projecting horns composed of the squamosal and frontoparietal bones. 3. Otic margin. — (Fig. 2a); the lateral edge of the paraoccipital process and the squamosal. 4. Subtemporal fenestra. — (Fig. 2a, c); the lateral fenestra bounded anteriorly by the maxillary, dorsally by the squamosal and posteriorly by the quadratojugal. 5. Supraorbital ridge. — (Fig. 2a); the bony frontoparietal ridge located at or near the dorsal margin of the orbit and the lateral edge of the frontoparietals; variable in presence, shape and length. 6. Preoccipital ridge. — (Fig. 8); transverse frontoparietal ridge lying parallel to the occipital flange and uniting the posterior termini of the supraorbital ridges. 7. Postorbital ridge. — (Fig. 2a); the bony frontoparietal ridge located along the posteromedial margin of the postorbital indentation. 8. Zygomatic ridge. — (Fig. 2c); a bony ridge variable in degree of de- velopment and extent; ridge is constituted by squamosal posteriorly and maxillary anteriorly, and usually extends diagonally ventrad from the posterior border of the ventral arm of the squamosal to the postero- ventral margin of the orbit. 9. Canthus in section. — a phrase used to describe the configuration of the snout-loreal region in transverse section; thus, the canthus may be rounded {i.e., obtusely angular) or angular (i.e., acutely angular). ACKNOWLEDGMENTS I am grateful to the following persons for the loan of specimens and/or provision of working space in their respective institutions: F. Bernis, James E. Bohlke, James R. Dixon, Alice G. C. Grandison, Walter Hellmich, W. Ronald Heyer, Alan E. Leviton, Hymen Marx, Charles W. Myers, the late James A. Peters, Jay M. Savage, Doro- thy M. Smith, Greta Vestergren, Charles F. Walker, Ernest E. Williams, John W. Wright, and Richard G. Zweifel. Photographs of living frogs and valuable field notes were generously provided by Martha L. Crump, William E. Duellman, Charles W. Myers, Catherine A. Toft, John E. Simmons and David B. Wake. Michael E. Whitenton contributed many hours of xolunteer labor devoted RELATIONSHIPS OF NEOTROPICAL HORNED FROGS Maxillary Canthal / Premaxillary Alary process Prevomer Nasal Sphenethmoid Supraorbital ridge Frontoparietal Subtemporal fenestra Postorbital ndentction Paraoccipital process Sphenethmoid Cu I triform process of parasphenoid Pterygoid Columella Quadratojugal Squamosal Sphenethmoid Maxillary Occipital flange- Frontoparietal- Nasal- Squamosal Zygomatic ridge Occipital condyle Subtemporal fenesta Pterygoid Quadratojugal Angular Fig. 2. Skull of Hemiphractus prohoscideus (KU 124139, 9; ca. X 1.5). Top: dorsal \iew. Middle: \entral view. Bottom: Lateral view of skull and mandible. Dermal ornamentation not shown. 6 OCCASIONAL PAPERS MUSEUM OF NATURAL HISTORY to measuring preserved specimens. The success of my own and my colleagues' field work in Ecuador was significantly enhanced through the provision of transportation by the Texaco Petroleum Company and the hospitahty of Ing. Ildefonso Munoz B. of Santa Cecilia, Ecuador. William E. Duellman, Alan E. Leviton and John D. Lynch critically reviewed the manuscript. Juan Leon graciously provided a Spanish translation of the Summary. Although this manuscript is a result of my long-standing interest in the systematics and biology of casque-headed hylids, I wish to extend special thanks to George S. Myers and Alan E. Leviton who share my interest in these peculiar frogs and who generously con- tributed the unpublished results of their earlier investigations of Hemiphractus. I gratefully acknowledge their generosity and the impetus provided by both gentlemen towards the completion of this contribution to an understanding of these enigmatic frogs. HISTORICAL RESUME Frogs of the genus Hemiphractus are among the most bizarre of anurans. Taxonomically, they have been known since the first species was described and illustrated by Spix in 1824. Because of their extraordinary aspect and adventitious appearance in early collections (namely those of Spix, Jimenez de la Espada, and Moritz Wagner in the early to mid-19th Century), commentary on this peculiar group of anurans occurred quite frequently in the literature of the late 1800's and early 1900's. There was an early proliferation of specific names associated with the genus, and a spate of misinformation has been perpetuated throughout the litera- ture of the past 150 years. In an attempt to clarify the history of this genus, a review is presented below. Spix (1824:28, Pi. IV, Fig. 2) described Rana scutata; perhaps the most significant statement in his brief description is as follows, ''maxillae ensiformes, superior denticulafo, inferior non dentata, apice intus suhaculeata." Gravenhorst (1825:920) proposed a new leptodactylid genus, Stombus, and included three species in it — Rana cornuta Linne, R. megastoma Spix, and R. scutata Spix. Sub- sequently, Wagler (1828:735 and 743, PI. X, Figs. 1-5) provided a short description and set of illustrations of the type specimen of scutatus. As opposed to Spix's statement, Wagler contended that the frog had mandibular teeth; on that basis he erected a new genus, Hemiphractus. Furthermore, he proposed the substitute name Hemiphractus spixii. Subsequently, Wagler (1830) gave a more complete description of the type specimen, placing it in the order "Ranae" within the family "R. phaneroglossaeT Only passing mention was made of this single specimen of Hemiphractus in the succeeding 32 years, and in several herpetological works of this period, it was suggested that Hemiphractus was allied with lepto- RELATIONSHIPS OF NEOTROPICAL HORNED FROGS 7 dactylids and, in particular, the genus Ceratophrys (e.g., see Du- meril and Bibron, 1841:430, and Giinther, 1858:26). The first substantial contribution to our knowledge of Hemi- phractus is that of Peters ( 1862 ) . In this paper he proposed that the genus was allied to "climbing" or "treefrogs." Peters (p. 146) reinstated the specific name scut at us, which has priority over Wagler's substitute name, spixii, and on the basis of his examination of Spix's specimen, provided a detailed morphological account and illustrations of the frog (pp. 146-149, PI. 1, Figs. 1-6). To my knowledge, this is the last detailed published description of scutatus that is based on examination of the type specimen, although Noble (1926:18) briefly mentioned it. Presumably, the t\'pe, which was housed in the zoological collections at Munich, was destroyed or lost during World War II. Peters described and illustrated (1862: 149-152, PI. 2, Figs. 1-4) a second species, Hemiphractus fasciatus. He provided (p. 145) the first adequate definition of Hemiphractus and placed the genus within the family Hemiphractidae, which he distinguished from other anuran families on the basis of the follow- ing combination of characters: 1) absence of parotoid glands; 2) small transxerse processes of the sacral vertebrae; 3) presence of teeth on the palatine bones; 4) presence of rudimentary inter- digital \\ ebbing; 5) presence of a round, attached tongue; and 6) presence of flattened digital tips. Significanth', Peters did not in- clude the "presence of 'growing' teeth on the mandible" in his definition of the family, but this character is included in the ge- neric definition (p. 145) preceding that of the family. In his paper on the structure and distribution of arciferous Anura, Cope (1866:69) included the Hemiphractidae and Cystig- nathidae (=Leptodact\'lidae) in the third of three "groups" of arciferal frogs. The "group" was defined as ha\ing cylindrical sacral diapophyses and procoelous vertebrae. The Hemiphractidae was distinguished from the Cystignathidae by the presence of mandibular teeth in the former and their absence in the latter. In Cope's definition of the Hemiphractidae, he concluded (p. 88) that, "It [the Hemiphractidae] is probably intermediate between the Hylidae and Cystignathidae." Cope (1868:114) described a third species of Hemiphractus, H. divaricatus, collected by Pro- fessor James Orton from "Peru, Napo and Maranon rivers." Jimenez de la Espada ( 1871 ) reported on his extensive collec- tions from South America, including an account of Hemiphractus scutatus and the description of a new genus, Ceratlujla. He dis- tinguished (pp. 62 and 63) CeratJu/la from Hemiphractus in a general way, noting that Hemiphractus was "toad-like," whereas Cerathyla was "hylid-like." The latter conclusion probably was based on the larger size of H. scutatus and the reduction of its digital discs as contrasted with the smaller specimens with ex- 8 OCCASIONAL PAPERS MUSEUM OF NATURAL HISTORY panded digital discs allocated to Cerotlujla. Both genera were placed in the family Hemiphractidae, and the family within a "sec- tion" entitled Hemiphractina ( Opisthoglossa Platydactyla: Ba- trachia Salientia: Amphibia). In this paper (p. 64), four new species were described and placed in the genus CerathijJa. These are: 1) C. proboscidea from Sumaco, Ecuador, 2) C. bubahis from Archidona, Ecuador, 3) C. pabnarum from San Jose de Moti, Ecua- dor, and 4) C. braconnieri from Archidona, Ecuador. Brocchi (1877) published a detailed osteological study of Hemi- phractus scutatus. His osteological illustrations (Pi. 16) are plenti- ful and accurate. Although Brocchi correctly stated that H. scutatus possesses odontoids on the mandible instead of true teeth, he er- roneously described the vertebrae as being opisthocoelous. Bou- lenger (1882:451-455) maintained scutatus and divaricatus in the genus Hemiphractus, and in an obvious error, cited Cerathi/la as CeratohyJa. He distinguished the two genera on the basis of dilated (Ceratohyla) versus non-dilated (Hemiphractus) digital discs. Boulenger included braconnieri, pabnarum, bubahis, proboscidea, and fasciata in the genus Ceratohyla. The family Hemiphractidae (including three genera — Hemiphractus, Ceratohyla, and Amphodus [=PhyUodytes]) was included in Boulenger's Series Arcifera (suborder Phaneroglossa: order Ecaudata: class Batrachia). Unfortunately the mistakes of Brocchi and Boulenger were perpetuated and compounded by Gadow ( 1901 ) who recognized hemiphractines as a subfamily, Hemiphractinae, and allied it with the leptodactylids (p. xi; family Cystignathidae: suborder Phanero- glossa: order Anura: subclass Lissamphibia: class Amphibia). Gadow's Hemiphractinae contained three genera (Hemiphractus, Ceratohyla, and Amphodus) and, among other things, was char- acterized by 1 ) teeth on both jaws, the vomers, and palatine bones, or on the palatines and parasphenoids in Amphodus, and 2) opis- thocoelous vertebrae. He further stated (p. 210) that the skin of the head is involved in cranial ossification in Hemiphractus. Two additional species of Cerathyla were described in 1917. The first, C. panamensis, was described by Stejneger (1917:31) from Signal Loma on the north coast of Panama. The second, C. johnsoni, was named by Noble (1917:789) from Santa Rita Creek, Departamento Antioquia, Colombia. Miranda-Ribeiro (1926: 118) described Hemiphractus bouJenjieri from the Buckley col- lection from "Ecuador." Finally, Noble (1926:18-19) synonymized Cerathyla with Hemiphractus, noting that both genera are char- acterized by procoelous vertebrae, sliglitly dilated sacral diapoph- yses and broad, overlapping coracoid cartilages in the pectoral girdle. Noble ( 1931 ) later recognized the subfamily Hemiphrac- tinae (family Hylidae: suborder Procoela: order Salientia) within which he included all hylids in which tlie females carry eggs on RELATIONSHIPS OF NEOTROPICAL HORNED FROGS 9 their backs. This assemblage comprises Hemiphractus, Crypto- hatraclnis, Gcisfrotlwca, and Amphignatliodon; parenthetically, it should be noted that Amphignathodon does possess tiaie mandibular teeth. The most recent species to ha\'e been described in this group is Ceratohijla cristata from the Rio Pastaza watershed in Ecuador by Andersson (1945:18), who placed the genus in the family Leptodactylidae. Contemporaneously, Myers and Carvalho (1945: 18-19) described a recently acquired individual of Hemiphractus scutatus. In comparing the specimen ( Museu Nacional No. A842 from the state of Amazonas, Brasil) with Boulenger's (1882:452) description of scutatus and that of Miranda-Ribeiro (1926:118) for Hemiphractus houJengeri, they implicitly concluded that houlengeri was not distinct from scutatus. FinalK', Myers and Carvalho (pp. 20-21) described a large female Hemiphractus, also from Amazonas, Brasil. On the basis of comparison of this specimen with Bou- lenger's (1882) account of ])uha}us and Noble's (1926) description of joluisoni, they erroneously placed jolinsoni in the synonymy of huhaJus and designated Cerathyla huhaJus Jimenez de la Espada as the type species of the genus Cerathyla. The description of this specimen ( Museu Nacional No. A841 ) and accompanying photo- graphs (Figs. 17 and 18) clearly indicate that the indixidual is identified correctly as Hemiphractus phnsoni and not Hemiphractus huhalus. Hemiphractus W'agler, 1828 Hc'i}ii])Jiiactus Wagler, 1828, column 713 [T\pe species. — HemipJiracttis spixii Wagler, 1828 i=Rana scutata Spix, 1824) by monotypy]. Nohle, 1926:19 synonymy of Ceratliyla with Hemiphractus. Ccrathijla Jimenez de la Espada, 1871:63 [Type species. — CcratlujJa huhalus Jimenez de la Espada, by subsecjuent designation ( Myers and Carvalho, 1945:21)]. Ceratolujla Boulenger, 1882:453 [lapsus for Cerathyla]. Diagnostic definition. — 1) Skull extensively casqued, character- ized by prominent, posterolaterally projecting paraoccipital proc- esses; 2) dermal roofing bones exostosed in a reticulate and/or tuberculate pattern of dermal sculpturing; .3) dermal roofing bones not co-ossified; 4) maxillary, premaxillary, and prevomerine teeth fang-like and curved inward when present; 5) palatine edentate, bearing serrations; 6) mandible bearing odontoid serrations and a pair of enlarged, fang-like odontoids in the area of the mandibular symphysis; 7) cultriform process of parasphenoid o\erlapped an- teriorly by prevomers; 8) arciferal pectoral girdle with minute omosternum, small clavicles, broadly overlapping epicoracoid carti- lages, and massive coracoid bones; 9) presacral vertebrae procoe- lous, with enlarged neural spines which protrude dorsally in living and preserved specimens; 10) ilium bearing distinct dorsal crest; 10 OCCASIONAL PAPERS MUSEUM OF NATURAL HISTORY 11) vocal sacs and vocal slits absent; 12) submentalis muscle ex- tremely small and aiaphic; 13) intermandibularis muscle uniffer- entiated, incompletely developed anteriorly, and bearing an exten- sive median aponeurosis; 14) small, fleshy proboscis on terminus of snout; 15) one or more prominent, fleshy tubercles on the eyelid; 16) tympanum large and vertically ovoid with its surface plane ori- ented at a distinct angle to the longitudinal axis of the body; 17) hands large with we]l-de\eloped, non-bifid subarticular tubercles, ovoid, elevated thenar palmar tubercle and diffuse, cordiform outer palmar tubercle; 18) fingers basally webbed or lacking webbing, length from shortest to longest 2-1-4-3; 19) nuptial excrescence absent; 20) toes basally to one-half webbed; 21) pupil horizontal; 22) palpebrum clear; 23) eggs carried on back of female where young undergo direct development and are attached to dorsal skin of female by means of gill structures. Content. — As defined here, the genus Hetnipliractus contains five known species (Table 1): H. fasciatus Peters, H. huhalus (Jimenez de la Espada), H. prohoscideus (Jimenez de la Espada), H. johnsoni ( Noble ) , and H. scutatus ( Spix ) . Distribution. — Upper Amazon Basin of Colombia, western Brasil, Ecuador, Peru, and Bolivia ascending the slopes of the Andes to elevations of 1830 m; northern and eastern slopes of the Cordillera Occidental and Cordillera Central of the Andes in northern Co- lombia; Pacific slopes and lowlands of northwestern Ecuador, Colombia, and eastern Panama; Caribbean slopes of western Pan- ama ( to elevations of 1600 m ) . ANALYSIS OF CHARACTERS CraniuDi. — The cranial architecture of HemipJnactus is unique and, in its own fashion, as strange as the better known pipids and ceratophryine leptodactylids. The dermal roofing bones are ex- Table L — Alphabetical Synonymy of the Species of Hemiphractus. Trivial Name, Original Generic Name, Author, Date Current Name houlengeri, Hemiphractus, Miranda-Rilieiro, 1926 H. sctttotus hiaconnieri, Cciathijla, Jimenez de la Espada, 1871 H. huhalus huhalus, Cerathyla, Jimenez de la Espada, 1871 H. huhalus cristata, Ceratohijla, Andersson, 1945 H. prohoscideus divaricatus, Hemiphractus, Cope, 1868 H. scutatus fasciatus, Hemiphractus, Peters, 1862 H. fasciatus johnsoui, Cerathyla, Noble, 1917 H. johusoni })almarum, Cerathyla, Jimenez de la Espada, 1871 H. Jjuhalus panamensis, Cerathyla, Stejneger, 1917 H. fasciatus prohoscidea, Cerathyla, Jimenez de la Espada, 1871 H. prohoscideus scutatus, Raua, Spix, 1824 H. scutatus spixii, Hemiphractus, Wagler, 1828 H. scutatus RELATIONSHIPS OF NEOTROPICAL HORNED FROGS 11 tensively casqued, so that in dorsal aspect the skulls appear to be heavily armored triangular structures characterized by robust, pos- terolaterally projecting paraoccipital processes (Fig. 2). The latter processes are formed jointly by posterolateral growth of the fronto- parietals and otic rami of the squamosals. Although the skulls are casqued and ornately exostosed in a reticulate pattern (Figs. 8, 11, and 14), the dermal bones are not co-ossified with the covering skin. In each species the neurocranium is solidly roofed. The maxillaries are characterized by massive de\elopment of the pars facialis both preorbitally and postorbitally (Fig. 2b). An analogous lateral element has arisen from the quadratojugal in the temporal region. The quadratojugal is characterized further by the presence of one or t\vo bony tubercular projections posterolaterally. Denti- tionally, most Hemiphractus are armed with fang-like, inwardly curved maxillar\', premaxillary, and prevomerine teeth (Fig. 2b, c). The premaxillary series is interrupted medially by a slight diastema into which the mandibular odontoids fit when the mouth is closed. When present, the prevomerine dentition is limited to one to three teeth clustered on a low pedicle on the anterior part of the pre- vomer; //. joJin.soni lacks prexomerine teeth. The mandible has an exceedingl)- broad and massi\e articular surface. The angular and dentary bones ha\e a well-developed, thin bony ridge bearing a series of odontoid serrations (Fig. 2c). Anteromedially, there are paired odontoid structures which are similar in size and con- figuration to the maxillar\- teeth and distinctl\- larger than the other mandibular odontoids. These large mandibular odontoids seem to arise from the anteriormost end of the dentary bone. The den- taries lie closely adjacent to one another; there is no indication of mentomeckelian ossification. A variety of other interesting osteological features is visible in the ventral aspect of the skull (Figs. 2b, 9, 12, and 15). The pre- vomers of the genus are attenuate structures which extend from the level of the palatines to the anterior end of the maxillary. In addition to the teeth described abo\e, each prevomer bears a median, serrate ridge which extends from the level of the teeth, posteriorly, to the terminus of the bone. The lateral wings of the prevomers are reduced, forming only the anteromedial margin of the choanae. The palatines are slender structures which are nar- rowly separated medially and which bear series of irregular odon- toid serrations. The parasphenoid has a noticeably long cultri- form process which extends anteriorly between the palatines and terminally is overlapped by, or juxtaposed to, the prevomers. The cultriform process is smooth in all species except johnsoni (Fig. 15) in which it usually bears an indistinct median keel. The most dominant ventral element is the pterygoid bone. The anterior ramus of this structure articulates with a massixe bony process of 12 OCCASIONAL PAPERS MUSEUM OF NATURAL HISTORY the maxillary (herein referred to as the pterygoid process of the maxillary) and extends posteriorly to the otic region of the neuro- cranium. The medial ramus arises at this level. In ventral view, only the ventral edge of the medial ramus is visible as it articulates with the anterior edge of the parasphenoid ala. The medial ramus consists of a diffuse, vertically oriented sheet of bone which faces the anterior surface of the otic capsule and extends laterally from the otic capsule forming a thin wall of bone. The bone attaches to the expanded ventral ramus of the squamosal, thereby forming a partition between the orbital and otic regions of the skull. The posterior ramus of the pterygoid is notable for its massive articular surface. With the exception of scutatus, which bears a series of odontoids along the anterior, lingual margin of the anterior ramus of the pterygoid, the pterygoids of Hemiphractus are edentate and smooth. There are two distinct types of skulls in the genus Hemiphractus. One group, composed of scutatus (Fig. 1) and fasciatus (Figs. 11 and 12), is characterized by especially massive skulls which lack a postorbital fenestra and which are obviously much broader than long (Table 2). Preoccipital and postorbital cranial ridges are absent and the supraoccipital, canthal, zygomatic and maxillary ridges tend to be indistinct, especially in large females. In lateral view, it is evident that the subtemporal fenestra is relatively small; consequently, the lateral elements of the quadratojugal, squamosal, and maxillary are greatly expanded as contrasted to the same elements in }ohnsoni (Fig. 14), bubalus (Fig. 8), and prohoscideiis (Fig. 2c). The snout regions of scutatus and fasciatus are not depressed; the alary processes of the premaxillaries are inclined posteriorly at approximately a 45 degree angle. Ventrally, the pterygoid process of the maxillary can be seen to arise at approxi- mately the midlength of the maxillary arch. Hemiphractus fasciatus can be distinguished readily from H. scutatus by the presence of pterygoid odontoids in the latter and the obvious difference (Figs. 1 and 11) in the respective orientations of the paraoccipital proc- esses. Moreover, the sphenethmoid is completely obscured dorsally in scutatus and joJinsoni, whereas a small part of this element is visible in fasciatus. The second group, composed of johnsoni, huhalus, and pro- hoscideus, is characterized by somewhat less extensively casqued skulls which are only slightly longer than broad (Table 2; Figs. 2, 8, 9, 14, and 15). Each of these species has a postorbital indenta- tion. The indentation is smallest in johnsoni and rounded; whereas huhahis and prohoscideus are characterized by larger indentations that are noticeably broader than long. All three species have zygo- matic, maxillary, canthal, and supraorbital cranial ridges; the ridges are least evident in johnsoni and best developed in hul)ahis. Hemi- RELATIONSHIPS OF NEOTROPICAL HORNED FROGS 13 Sell u > Z .5- § 13 -J O M! V C a « c/:) •- ■-a tj « ^ c a;) o ^ c a > 2 S tin -oj ^^ C VI N ^ ^ ti -C -§8 O N •SPa ^ bC o a> C/2 c ±3 C c C p .52 •'-' fe g ctf 0) Si) c o U N CM c q; T3 C/D £JC 0; bC T 'I " <>^ E^ O — o 4L M •- 3 c ?: o 1^ ^ c ""^ c CO ■^^ Z z X 4J CD tn o a C/D IC 00 ro CD 10 <^" ^ cc C- oc :^ c-i 01 01 ,— I ,-H cc 06 00 ^_ CO CO ■— I o '-^ ro •'^ 1/^ ^^ 00 '-^ "10 "V -h' 'T CD ';;' in T t-" 1001 G^cocD(^^ ocmtpcvi 10 (>lCD^0CcOO5i-H00 '"""'"'^^""^^"^^ a> CO 'c 00 i> c CO C5 O 05 CD 00 X 00 00 Si^ CO m 00 m i-- 'V "\ 10 ^. m ^. in . in ^! t ^. t ^. cd ^. cd ^ CD If-, 05 -f. CD i/j T}- ^ CT in CTin CT :^ CO ei ^. CD '^ >1 ^ ^. CO ^. S * — - cc t^ ^^ c^^f^ g C5 a> T-j< ,-^ ^ -f CO fM 3! CO -^ *^ -.^ i, CO ' m A ^ i5 -« t^ °o Tt ^ in t^ Tfi O — - t^ ^ Cvl — - Tf TT Tf S!4^ Oi CD ^ irT ^ T^" <^1 ~r ^ 05 '^ OC ^ X m ^' rf ^ t- .;l m =c s^ 01 ^ X 5! "* 2 -^ I: iQ 06 t^. 00 0. k o d X ': t- ". X I- ^ X 'CO ==i^. t- o X (M X ^ c^ j^ ci g t^ "-. l-^ '^. t- t^ t- -^ X -S X '• ■■; CD co^— 1^^ Tt<^-^t^^^ t^^_oo^-^ Tf-^-'oq l^t^ t^t^ t^t^ COCO in CD '^1 t- d P ^ ^_ q c^ ^ CO i: X Jq CO [^ C5 § X ^. 2§ ^. § q S ^. ^ ' t-' ' X ^1 06 'CD '^. ^ ^ in ^. ^ t^ in •'^ -f "^ CO '-^. -^ X — 'CO---' a ' — -co^ co^d--^ cm'^ CO ^ CO CO TP CD ■<*' (M tM .>i . .. nvr^>nfTTTrmjr Fig. 9. Skull of Hemiphractus huhalus (KU I24I40, 5; ca. X 4). Top: Ventral \ie\v. Bottom : Posterior \ie\v. however, there is no direct evidence of this. Specimens have been collected in January, March, June, August, and October. Females (2) with large to moderate-sized o\^arian eggs were obtained in August as were most mature males (4 of 5). In contrast, October 28 OCCASIONAL PAPERS MUSEUM OF NATURAL HISTORY Fig. 10. Distribution of Hemiphractus bubalus in Ecuador. collections resulted in acquisition of 14 juveniles and one subadult female and no adults; juveniles also were collected in June (3) and August (2). Hemiphractus fasciatus Peters Heiniphracttis fasciatus Peters, 1862:149 [Holotype. — ZSM 36/0 from "Pastaza Valley, Ecuador" (in error); Moritz Wagner collector]. Ceratohyla fasciata — Boulenger, 1882:454. Cerathyla pana7nensis Stejneger, 1917:31 [Holotype.— USNM 55320 from Signal Loma, 5 km south of Santa Isabel, Provincia Colon, Panama; Mr. Charles Gordon Holland collector]. New synonym. Hemiphractus panainensis — Duellman, 1970:140. Justification of sijnomjmij. — Examination of the holotypcs of Cerathyla panamensis and Hemiphractus fasciatus has shown that these two frogs are extremely similar. Stejneger's specimen of panamensis is a large female, whereas Peters' specimen of fasciatus RELATIONSHIPS OF NEOTROPICAL HORNED FROGS 29 is a much smaller female; both individuals fall well within the range of observed variation of snout-vent length, and head length, head width, and tibia length relati\'e to snout- vent length ( compare Tables 2 and 5). Although ponamemis was described from Panama and fasciatus from Amazonian Ecuador ( a type locality which must be in error, because no other specimen of fasciatus has been col- lected from the Amazonian side of the Andes south of northwestern Colombia), the specimens are nearly identical in all diagnostic features — 1) the crania are the same; 2) the tongues are ovoid; 3) the skin is finely granular; 4) although Stejneger's specimen is faded, his color notes generally concur with those of Peters; and 5) both specimens have a small, dorsoventrally depressed proboscis. Because of these ob\ious similarities, I herein place the name panamensis as a junior synonym of fasciatus which has priority. Table 5. — Comparison of Morphological Measurements ( mm ) of Holot>'pes of Hctni- pJiractiis fasciatus and Cerathylo panamensis. (SVL=snout-\ent length; HW=head width; HL=head length) H. fasciatus, 9 (ZSM36/0) C. panamensis, 9 Original (USNM 55320) Original Measurements Measurements description; by description; by Character Peters, 1862 True!) Stejneger, 1917 Trueb Snout-vent length -.._ 46.0 46.5 60.0" 59.1" Head length, medial .... 18.0 23.1 Head length 21.0 (45.2) — 28.7 (^SVL) .-._ .... (48.6) Head length, tip of snout to para- occipital horn .... 23.0 ..__ 33.0 Head width .... 21.7 21.4 (46.0) 32.0 (53.3) 33.1 (%SYL) .... ._.. (47.2) (56.0) Interorbital distance 8.3 (38.8) 12.0 (37.5) 11.2 (%HW) .... (33.8) Internarial distance 2.7 3.5 Nostril to orbit distance 5.5 7.0 7.0 (%HL) (26.2) (24.4) Nostril to snout distance 4.9 6.0° 4.0" Tibia length 24.5 31.5 33.0 (fSVL) .... (52.7) (52.5) (55.8) Forelimb length - -.-. 34.0 40.0 — Hind limb length .... 80.0 — 99.0 — " Stejneger measured lengths involving the head from the tip of the fleshy proboscis, whereas I did not; hence, his values are greater than mine. 30 OCCASIONAL PAPERS MUSEUM OF NATURAL HISTORY Diagnosis. — 1) Size moderate, sexual dimorphism evident; maxi- mum observed snout- vent length in males 55.7 mm (x=49.6), in females 68.7 mm (x=57.3); 2) postorbital indentation absent; 3) posterior projection of paraoccipital process relatively greater in males (11.7% head length, N=ll) than in females (4.0% head length, N=10); width between distal tips of processes about the same in males ( 87.9% head width ) as in females ( 85.5% head width ) ; 4 ) otic margin of paraoccipital process not parallel to maxillary arch in dorsal view; instead, approximately parallel to longitudinal axis of body, slightly concave laterally, and characterized by low but dis- tinct ridge (Fig. 11); 5) canthal ridge indistinct, anterolaterally concave; canthus angular in section; 6) supraorbital ridges absent; 7) preoccipital ridge absent; 8) postorbital ridge absent; 9) zygo- matic ridge indistinct and poorly developed, extending diagonally downward from rear margin of ventral arm of squamosal to postero- ventral margin of orbit; 10) subtemporal fenestra small, relatively smaller in females than in males; 11) pterygoid edentate; 12) culhl- form process of parasphcnoid not keeled (Fig. 12); 13) snout not depressed in lateral profile; 14) clavicles not well developed; 15) neural spines of presacrals III-VIII and sacrum enlarged, with that of the fifth presacral being the largest; 16) width of sacral dia- pophyses less than width of transverse processes of presacral VIII and approximately equal to width of presacral III; 17) ventral shaft of urostyle laterally expanded; IS) prehallux moderately developed; 19) proboscis small, triangular and dorsoventrally depressed; 20) eyelid granular with a single enlarged tubercle amid sex^eral smaller, scattered tubercles; 21) tubercles on dorsal surface of hind limb indistinct and aligned in transverse rows; 23) small, poorly devel- oped calcar present on heel; 24) skin of flanks and venter with scat- tered, low subconical tubercles; 25) fingers not webbed, or only basally webbed; 26) toes basally webbed; 27) adhesive discs pres- ent; those of the toes slightly smaller than those of the fingers. HemipJiractus fasciatus can be distinguished from other mem- bers of the genus by the combination of its lack of a postorbital indentation (Fig. 11) and the presence of well-developed adhesive discs on the digits. It is distinguished further by having a diffuse pale mark bordering the dorsal margin of a dark inguinal mark; the pale mark extends dorsal to the cloaca to form an indistinct cloacal stripe. Distribution. — This species is widespread throughout Panama, northern Colombia, and the Pacific slopes of Colombia and north- western Ecuador between elevations of 300 and 1600 m (Fig. 13). Remarks. — This species is perhaps the best known of the genus. It has an extensive range and a wide altitudinal distribution; more- over, its habits, habitat preferences, coloration, and behavior are well documented in the copious field notes of Charles W. Myers, RELATIONSHIPS OF NEOTROPICAL HORNED FROGS 31 who collected many indi\iduals in the course of three years' field work m Panama. He reported (1966) some initial observations on the distribution and behavior of fasciatus, and included a photo- graph of its defense or "gaping" posture and a report of a release call (one of two such reports, the other being for prohoscideus) . It seems appropriate here to note that Myers reported (field notes of 18 January 1966) that the Panamanian nati\es refer to fasciatus as "rana brava," and to supplement this statement with an anecdote excerpted from Myers' field notes of 13 December 1964: "I made the mistake of sticking a finger in the gaping mouth to see if it [the frog] would bite. It did and held on like a vise. The two fangs in the lower \i\\\ (which I was unaware of, else would not have been so stupid) made two puncture marks (like a small snake bite) which bled real blood." Benefiting from Myers' experience, those of us who have had the opportunitv- to collect Hemiphractiis on subsequent occasions find that it is safer and wiser to grasp the species by the tips of the paraoccipital horns from the rear. This species, like other members of the genus, is a denizen of the forest; however, because of its relatively wide altitudinal dis- tribution, it inhabits lowland tropical rainforests, lower montane forests and cloud forests. By day it is found amidst leaf litter on the forest floor, whereas by night, it fa\ors perches above the ground. Most indi\iduals have been found sitting on sticks or low vegetation 0..3-0.6 m above the ground. Occasionally, larger indi\'iduals perch 1-2 m abo\c the ground on tree trunks, branches, and in the bases of arboreal bromeliads. There is e\idence that the species preys upon a variety of animals; one specimen, a large female (KU 93503) had an unidentified gastropod, a lizard (Ptt/cho- g,Iossus), and a dendrobatid frog (CoJostetlnis) in her stomach. Coloration in living fasciatus is highly \ariable. Generally, the dorsal ground color varies from dull olixe green or yellowish tan to dark brown; darker dorsal markings (usually brown) are nearly always present. The flanks, and anterior and posterior surfaces of the thighs are lighter than the dorsum, usually a pale tan or flesh color with brown mottling; in at least one indi\idual, the posterior surfaces of the thighs were dark brown. The venter varies from dark brown to gray; some mottling, \ arying between light browTi, yellow and white is present. In some individuals, the gular and pectoral regions are darkest and grade to a perceptibh' paler ab- dominal color. Nearly all indi\iduals have a distinct, pale gular spot. The iris is almost in\ariably pale, its tone \arying from pale yellow or pale orange-gold to light brown. The color may be uni- form, save for the horizontal brown stripe bisecting the iris through the pupil, or the iris may be two-toned. In the latter case, the lower part of the iris is darker than the upper portion and varies 32 OCCASIONAL PAPERS MUSEUM OF NATURAL HISTORY 0. / if Fig. 11. Skull of Hemiphractus fasciatus (KU 93510, $ ; ca. X 4.5). Top: Dorsal view. Bottom: Lateral view of skull and mandible. RELATIONSHIPS OF NEOTROPICAL HORNED FROGS 33 m :<• ^-J^sK, * ■^■4. -^J" , "^r;* Fig. 12. Skull of Hemiphractiis fasciatus (KU 93510, (^ ; ca. X 4.5). Top: Ventral view. Bottom: Posterior view. from gray-gold to reddish gray. The irises of some individuals are characterized by fine red or black reticulations. Hemiphractiis fasciatus has been collected in all months except March, April, August, and October. The one female with associated young (BYU 19142, see Duellman, 1970:143 and PI. 7 for a detailed discussion of this specimen) was collected in September. Females with large ( ca. 5-6 mm diameter ) eggs were obtained in November 34 OCCASIONAL PAPERS MUSEUM OF NATURAL HISTORY 10- 0 100 300 KM 70 Fig. 13. Distribution of Heiniphraciufi fasciatiis (circles) and the type locality of Hemiphractus johusoni (triangle) in lower Central America and north- western South America. See figure 17 for distribution map of H. johnsoni. (1), December (1), and May (1); females with moderate to small- sized ovarian eggs were obtained in January (4) and May (2). Mature males were collected in January (10) and May (1); sub- adult males and females were obtained at the same times, and juveniles in the months of February, May through July, and September. While examining .specimens at the Natural History Museum in Stockholm, I encountered a specimen of Hemipliractus with a manu- RELATIONSHIPS OF NEOTROPICAL HORNED FROGS 35 script name of Cerathyla minima applied to it by Andersson. The specimen (NHRM 1981 from El Tambo, La Costa, Departamento Cauca, Colombia, collected on 16 June 1936 by K. von Sneidern) is a juvenile male (25.8 mm snout-vent length). It is pale tan in color with no evident markings. The configuration of the skull confirms its identification as Hemiphractiis jasciatus. Hemiphractus johnsoni (Noble) new combination Cerathyla johnsoni Noble, 1917:798 [Holotype.— AMNH 1341 from Santa Rita Creek, 14 miles north of Mesopotamia, Departamento Antioquia, Colombia; R. D. O. Johnson collector]. Cerathyla htihahis — Myers and Carvalho, 1945:21 [in part]. Diagnosis. — 1) Size moderately large, sexual dimorphism evi- dent; maximum observed snout-vent length in males 52.9 mm (x=:48.2), in females 77.2 mm (x=68.2); 2) postorbital indenta- tion moderately small, rounded (Fig. 14); 3) posterior projection of paraoccipital process relatively less in males ( 19.5% head length, N=2) than in females (27.3% head length, N=2); width between distal tips of processes relatively less in males (82.1% head width) than in females (90.1% head width); 4) otic margin of paraoccipital process more nearly parallel to maxillary arch than to longitudinal axis of body in dorsal view; 5) canthal ridge distinct, anterolaterally concave; canthus angular in section; 6) supraorbital ridges moder- ately developed, slightly convergent posteriorly, terminating just anterior to occipital flange; 7) preoccipital ridge absent; 8) post- orbital ridge absent; 9) zygomatic ridge indistinct, extending from rear margin of \'entral arm of squamosal to posterolateral margin of orbit, parallel and ventral to zygomatic arm of squamosal; 10) subtemporal fenestia minute; 11) pterygoid edentate (Fig. 15); 12) cultriform process of parasphenoid usually with indistinct keel; 13) snout not depressed in lateral profile; 14) clavicles moderately well developed; 15) neural spines of presacrals IV- VIII enlarged, with none being distinctly larger than any other; 16) width of sacral diapophyses less than width of transverse processes of presacral VIII and greater than width of presacral III; 17) ventral shaft of urostyle laterally expanded; 18) prehallux poorly developed; 19) proboscis small, triangular and dorsoventrally depressed; 20) eye- lid coarsely granular with one slightly enlarged, conical tubercle; 21) tubercles on dorsal surface of forearm scattered; 22) tubercles on dorsal surface of hind limb absent or poorly developed; 23) small, poorly-de\'eloped calcar present on heel; 24) skin of flanks and venter smooth; 25) fingers not webbed; 26) toes basally webbed; 27) adhesive discs present; those of toes slightly smaller than those of fingers. Hemiphractus johnsoni differs from other members of the genus by the presence of a small, round postorbital depression. Further- 36 OCCASIONAL PAPERS MUSEUM OF NATURAL HISTORY ^^ tl Fig. 14. Skull of Hemiphractus johusxmi (KU 153436, $ Dorsal view. Bottom: Lateral view of skull and mandible. ca. X 4). Top: RELATIONSHIPS OF NEOTROPICAL HORNED FROGS 37 y^TTynrY-r i Fig. 15. Skull of Hemiphractus johnsoni (KU 153436, $ ; ca. X 4). Top: Ventral view. Bottom : Posterior \1e\v. more, it lacks the bold dorsal markings characteristic of bubalus, and has a small proboscis which distinguishes it from proboscideus. Distribution. — This species is known from 11 specimens repre- senting an extraordinaiy distribution that ranges from the northern Andean slopes of Colombia and the upper Amazon Basin and lower Amazonian slopes of the Andes from Ecuador to northern Bolivia 38 OCCASIONAL PAPERS MUSEUM OF NATURAL HISTORY (Figs. 13 and 17). It occurs between elevations of 300 and 1910 m. Remarks. — On the basis of available museum specimens, H. jolmsoni is the rarest member of the genus. There are 11 known specimens exclusive of young associated with an adult female. The only notes on habitat (Noble, 1917:801) and the localities from which specimens have been collected indicate that this species inhabits dense forests. One specimen (AMNH 58631) is a female from Peru with 18 young attached to her back; another female collected by David B. Wake and Richard Etheridge in Peru in July had 17 eggs on her back. Available field notes indicate that jolmsoni perches on low branches (0.3-0.7 m above ground) at night. In life, this frog is tan-brown dorsally with reddish mottling and darker brown spots. The venter is grayish white and the iris gold above and orange below with a horizontal brown streak. For notes on the location of the type locaHty, see Trueb (1971:28). It should be noted that Myers and Carvalho's (1945:19-21; Figs. 17 and 18) discussion of Ceratliyla huhaJus is referrable to Hemiphractiis johnsoni. Hemiphractus proboscideus (Jimenez de la Espada) New combination Ceratliyla proboscidea Jimenez de la Espada, 1871:64 [Holotype. — MNCN 173 from Smiiaco, Provincia Napo, Ecuador; Jimenez de la Espada collector], Ceratohyla proboscidea — Boulenger, 1882:453. Ceratohijla cristata Andersson, 1945:18 [Holotype. — NHRM 1911 from Rio Pastaza watershed, Ecuador; William Clarke-Maclntyre collector]. New synonym. Justification of synonymy. — Types of both proboscidea and cristata are from Amazonian Ecuador. The snout-vent lengths of each and their head widths, head lengths and tibial lengths relative to the snout-vent lengths fall within or near the range of observed variation for the individual's respective sex (compare Tables 2 and 6). The holotype of proboscidea is a small male which is poorly preserved and completely faded, making comparison with the larger female type of cristata difficult. Nonetheless, the cranial characteristics of both specimens are the same and unique among hemipractines. Furthermore, both have ovoid tongues, a small dermal protuberance on each eyelid, and a long, fleshy proboscis with an irregular dorsal edge. Because of these distinct resem- blances, I herein place the name cristata as a junior synonym of proboscideus. Diagnosis. — 1 ) Size moderate, sexual dimorphism evident; maxi- mum observed snout- vent length in males 49.7 mm (x=46.3), in females 67.1 mm (x=56.1); 2) postorbital indentation large, rounded, and about twice as wide as long (Fig. 2); 3) posterior projection of paraoccipital process relatively greater in males (24.5% RELATIONSHIPS OF NEOTROPICAL HORNED FROGS 39 Table 6. — Comparison of Morphological Measurements (mm) of Holotypes of Cerathyla proboscidea and Ceratohyla cristata. ( SVL=snout-vent length; HW=:head width) Cerathyla proboscidea, $ (MNCN 173) Character Measurements by Trueb Ceratohyla cristata, 5 (NHRM1911) Original description; Andersson, Measurements by Trueb 1945 62.0 62.0 28.3 30.0 (45.6) 29.6 (48.4) 10.0 (47.7) 9.0 (33.3) 2.5 (30.4) 2.6 6.0 5.5 6.0 6.9 36.5 36.2 (58.9) 28.9 (58.4) (46.6) Snout-vent length 42.8 Head length 19.0 (%SVL) (44.4) Head width .^ 20.5 (%SVL) (47.9) Interorbital distance 6.0 (%HW) (29.3) Internarial distance — - 1.9 Width of upper eyelid 3.2 Nostril to snout distance — - 4.3 Tibia length 26.3 (%SVL) (61.4) Foot length" 20.3 (%SVL) (47.4) ° Excluding tarsus head length, N=14) than in females (19.8% head length, N=24); width between distal tips of processes relatively greater in males (105.1% head width) than in females (97.7% head width); 4) otic margin of paraoccipital process approximately parallel to maxillary arch in dorsal view, characterized by a distinct, well-developed ridge; 5) canthal ridge distinct and straight; canthus angular in section; 6) supraorbital ridges distinct, slightly convergent postero- medially, terminating just anterior to occipital flange; 7) preoccipi- tal ridge absent; 8) postorbital ridge absent; 9) zygomatic ridge well developed, extending diagonally downward from rear margin of ventral arm of squamosal along posterolateral margin of orbit to merge inconspicuously with suborbital margin; 10) subtemporal fenestra large; 11) pterygoid edtentate; 12) cultiiform process of parasphenoid not keeled; 13) snout depressed in lateral profile; 14) clavicles moderately well developed; 15) neural spines of pre- sacrals IV- VIII and sacrum enlarged, with those of presacrals IV- VIII being distinctly larger than that of sacrum; 16) width of sacral diapophyses slightly less than width of transverse processes of presacral VIII and slightly greater than width of presacral III; 17) ventral shaft of urostyle laterally expanded; 18) prehallux poorly developed; 19) proboscis large, laterally compressed and char- acterized by an irregular dorsal edge; 20) eyelid with central, en- larged conical tubercle flanked by several slightly smaller tubercles; 40 OCCASIONAL PAPERS MUSEUM OF NATURAL HISTORY 21) tubercles on dorsal surface of forearm arranged in oblique rows; 22) tubercles on hind limb aligned in oblique rows; 23) small, distinct calcar present on heel; 24) skin of flanks and venter finely granular; 25) fingers not webbed; 26) toes one-fourth webbed; 27) adhesive discs present; those of toes slightly smaller than those of fingers. Hemiphracttis prohoscidetis can be distinguished from other members of the genus by its long, compressed proboscis, large postorbital indentation, and characteristic white or cream colored ( in preservative) discs on the inner two fingers. Distribution. — This species is known only from the upper Ama- zon Basin in Ecuador, where it occurs between elevations of 300 and 720 m along the Andean slopes (Fig. 16). Remarks. — Of the 72 preserved specimens of prohoscideus, 46 are from Santa Cecilia, Ecuador, and have been collected during a period of five years. Although intensive field work has been Fig. 16. Distrilintion of Hemiphractus prohoscideus in Ecuador. RELATIONSHIPS OF NEOTROPICAL HORNED FROGS 41 carried out at this locality throughout the year, no prohoscideus have been obtained in February, September, or December. The majority of specimens (juveniles, adult males, and females with moderate or small-sized ovarian eggs; total of 34 frogs) was col- lected in June and July. In addition, juveniles and adult males were found from March through May and in August. Juveniles were collected also in October, and females in January, March, May through August and October and November. The temporal distribution of juveniles suggests that breeding must occur through- out the year, but it is disheartening to note that none of the KU females has large ovarian eggs, eggs on the back or any evidence of having recently had young on the back. One female (UMMZ 92110, 61.3 mm snout-vent length) has depressions on her back, thereby documenting the reproductive mode in this species. Based on examination of the Santa Cecilia collection, it is obvious that prohoscideus has a diverse appetite; among the stomach contents were a \'ariety of beetles and orthopterans in nearly all specimens. Other food items found were an unidentifiable frog (genus Hijla), a ju\'enile lizard (Anolis chn/solepis), two adult frogs (Eleuthcro- dacfylns conspicillattts and Ostcocephalus leprieurii), and a \aricty of unidentified eggs and an annelid. Hemiphractus prohoscideus seems to be restricted to lower mon- tane forest and lowland rainforest, where it occasionally is found amidst leaf litter by day, but most often is found perched on low vegetation (below 1.5 m abo\e the ground) at night. Typically, the species is found on palm fronds, vines, leaves of low herbs, bushes and trees. The irises of prohoscideus \ary between pale pink and pale gray with indistinct, diffuse gray triangles anterior and posterior to the pupil ( apices are located at the margin of the pupil ) . The tongue and buccal ca\'ity are bright \cllow. Otherwise coloration in this species is highly \ariable indi\idually, and within indi^•iduals from day to night. By night, the dorsal areas of most indi\iduals are yellowish tan with darker tan or brown markings; occasionally, indix'iduals are found which are gray \\'ith brown reticulations. By day, the typical dorsal coloration is transformed to a dark brown with darker brown markings. The \enter, which is pale colored or white at night, changes dramatically to a deep brown or reddish brown by day. This dark \'entral ground color is \ariously punc- tuated by orange or yellowish tan abdominal blotches and yellow to white spots in the gular area. The flanks tend to be pale flesh color and the thighs vary from yellowish tan with reddish brown bands to a dark gray posteriorly. According to field notes, the adhesive discs are orange in life. On at least two occasions this species has been observed to as- sume the "gaping" defense posture (discussed in the Remarks 42 OCCASIONAL PAPERS MUSEUM OF NATURAL HISTORY section of the account of fasciatus) when disturbed, exposing the bright yellow buccal cavity and tongue. Both individuals are males (48.2 and 43.1 mm snout-vent length), and neither one could be induced to bite. The larger of the two frogs emitted a "grunting" sound when collected. Hemiphractus scutatus (Spix) Rana scutata Spix, 1824:28 [Holotype. — formerly in Zoologisches Sammlung Miinchen; no number and now destroyed or lost]. [Stomhus] scutata — Gravenhorst, 1825:920. Hemiphractus spixii Wagler, 1828:743 [Substitute name for Rana scutata]. Hemiphractus scutatus — Peters, 1862:146. Hemiphractus divaricatus Cope, 1868:114 [Holotype. — ANSP 4132 from "Peru, Napo, and Maraiion rivers"; Prof. James Orton collector]. New synonym. Hemiphractus houlengeri Miranda-Ribeiro, 1926:118 [Holotype. — BMNH 1880.12.5.153 from "Ecuador"; Mr. Buckley collector]. Synonymy jide Myers and Carvalho ( 1945: 19 ) . Justification of synonymy. — Having examined the type of divari- catus and compared it and Cope's description with the descriptions of scutatus by Peters (1862) and Brocchi (1877), I find no grounds on which to recognize divaricatus as distinct from scutatus; accord- ingly, I place the name divaricatus as a junior synonym of scutatus. The differences described by Cope (1868) are minor variations in certain cranial features which he discerned by comparison of his specimen with Peters' description. Examination of table 7 shows that morphometrically, the types of divaricatus and houlengeri compare favorably with the measurements reported by Peters for scutatus and those reflecting variation within the sample I examined (Table 2). Diagnosis. — 1) Size large, sexual dimorphism evident; maxi- mum observed snout-vent length in males 57.4 mm (x=50.7), in females 80.5 mm (x=70.4); 2) postorbital indentation absent (Fig. 1); 3) posterior projection of paraoccipital process approximately the same in males (18.0% head length, N=7), as in females (17.9% head length, N=14); width between distal tips of processes about the same in males (93.8% head width) as in females (92.7% head width); 4) otic margin of paraoccipital process approximately parallel to maxillary arch in dorsal view ( Fig. 1 ) and characterized by a low but distinct ridge; 5) canthal ridge absent; canthus rounded in section; 6) supraorbital ridges absent; 7) preoccipital ridge absent; 8) postorbital ridge absent; 9) zygomatic ridge scarcely evident, extending diagonally upvv^ard from the postero- lateral margin of the orbit about half the distance to rear margin of ventral arm of squamosal; 10) subtemporal fenestra small; 11) anterior ramus of pterygoid bearing denticulate serrations on lin- gual margin; 12) cultriform process of parasphenoid not keeled; 13) snout slightly depressed in lateral profile; 14) clavicles poorly RELATIONSHIPS OF NEOTROPICAL HORNED FROGS 43 Table 7. — Comparison of Morphological Measurements (mm) of Holotypes of Hemiphractus scutatus, H. divaricatus, and H. hotilengeri. (SVL=: snout- vent length; HW=head width ) Character H. scutatus, sex? Description by Peters, 1862 H. divaricatus, sex? (ANSP'4132) Original description; Cope, 1868 H. boulengeri, $ BMNH 1880.12.5.153 Measurements by Trueb Snout-vent length 60.0 Head length, medial 23.0 Head length, tip of snout to paraoccipital horn 31.0 Head length (%SVL) .... Head width 32.0 (%SVL) (53.3) Forelimb length 37.0 Hind limb length 75.0 Foot length" (%SVL) __ Interorbital distance (%HW) __ Intemarial distance " Including tarsus 55.0 22.2 28.0 33.9 (61.6) 31.8 75.1 34.9 (63.5) 10.6 (31.3) 3.2 57.6 29.1 (50.5) 33.8 (58.7) 35.0 (60.8) 12.7 (37.6) 3.5 developed; 15) neural spines of presacrals I-VIII and sacrum en- larged, with that of the third presacral being the largest; 16) width of sacral diapophyses greater than \\'idth of transverse processes of presacral VIII and about equal to width of presacral III; 17) ventral shaft of urostyle laterally expanded; 18) prehallux well developed, anvil-shaped; 19) proboscis small, triangular and dorso- ventrally depressed; 20) eyelid coarsely granular with one slightly enlarged conical tubercle; 21) tubercles on dorsal surface of fore- arm arranged in oblique rows corresponding to dark bars; 22) tubercles on dorsal surfaces of hind limb aligned in oblique rows; 23) calcar absent on heel; 24) skin of flanks an venter coarsely granular; 2.5) fingers about one-fourth webbed; 26) toes basally webbed; 27) adhesive discs on fingers and toes vestigial, not visible in dorsal view. Hemiphractus scutatus differs from all other members of the genus by its lack (xestigial presence) of visible adhesive discs. It is distinguished further by its large size, and lack of 1 ) a postorbital indentation, 2) a canthal ridge, and 3) a pattern of labial stripes radiating from the suborbital margin to the edge of the lip. Distribution. — This species occurs in the upper Amazon Basin in Ecuador, Peru, and western Brasil between elevations of 250 and 1830m (Fig. 17). 44 OCCASIONAL PAPERS MUSEUM OF NATURAL HISTORY Fig. 17. Distribution oi Hemiphractiis sctitatus (circles) and Hemiphractus johnsoni (triangles) in Ecuador, Peru, and Brasil. See figure 13 for location of type locality of H. johnsoni in Colombia. Remurks. — Until the recent acquisition of two specimens of sctitatus from cloud forest at ele\'ations of 1400 and 1830 m, this species was thought to be an inhabitant of lower montane and low- land rainforest exclusively. It differs in one respect from its con- geners in that available field notes indicate that sciitahts is always found on the forest floor — a habit to be expected in view of its loss of functional toe pads. Color notes are rare (see Myers and Car- valho, 1945, for a resume); the most recent notes reported that an adult sctitatus was pale tan above marked with darker brown. The specimen had black marks on the upper lip and a black border around the eye; the iris was bronze. RELATIONSHIPS OF NEOTROPICAL HORNED FROGS 45 There are three preserved females with associated embryos (eggs). One of these (KU 129751, 72.3 mm snout- vent length) was collected in November and carried five eggs, each of which is about 9 mm in diameter. The oviducts of this specimen were empty and greatly distended. A second female (UMMZ 92106, 80.5 mm snout- vent length), for which there is no date of collection, has four depressions on her back; the associated embryos are 10 mm in diameter and the oviducts are empt\'. The third female (FMNH 121855, 67.3 mm snout-vent length) was collected in September; only two embryos are associated with this individual. The balance of the specimens for which data are available are females with moderate (ca. 6 mm) to small-sized ovarian eggs obtained in Janu- ary, February, May and November. An adult male was collected in December and a juvenile in February. There are no records of a call for this species, no documentation of the color of the buccal cavity, and no account of its assuming the "gaping" defense posture typical of other members of the genus. INTRAGENERTC RELATIONSHIPS In order to hypothesize on intragcneric relationships within Hemiphractus, it is necessary to establish the probable evolutionary patterns or directions of those character states that will be utilized in the subsequent analysis. Recently, there has been considerable commentary on the kinds of characters useful in anuran classifica- tion and the nature of their change through time ( see various papers in Vial, 1973). In working with Hemiphractus, I am confronted with two basic problems. The evolutionary status of most anuran characters has been discussed at broader taxonomic levels than that of the genus. Secondly, Hemipliracttis is exti"emely specialized and probably a highly derived group; thus, my conclusions cannot be expected to be necessarily comparable or parallel to those deter- mined for other Neotropical genera {e.g., see Trueb and Duellman, 1971; Trueb, 1971). Nonetheless, in the discussions that follow, the criteria for the various states of primitiveness, derivation, ad- vancement, specialization and generalization are the same as those discussed in an earlier paper (Trueb, 1973:65-66). Most of the characters to be considered are osteological; these, as a suite, clearly establish the advanced and derived status of Hemiphractus (see Trueb, 1973). Given this conclusion, I will attempt to delineate the evolutionary trends within the genus and their phylogenetic and, if possible, adapti\e significance. It has been shown that casque-headedness is a derived state ( Trueb, 1970; 1973), and that in casque-headed frogs, the degree of casquing is usually greater in the more derived species. Following this fine of reasoning, fasciatus and scutatus should be considered more advanced than their congeners that possess less extensively ossified 46 OCCASIONAL PAPERS MUSEUM OF NATURAL HISTORY skulls by reason of the presence of postorbital indentations, dor- sally exposed sphenethmoids, and moderate to large subtemporal fenestrae. However, it has been demonstrated also that there is an obligate ontogenetic sequence of ossification in which casquing always precedes exostosis — the appositional growth of dermal bone to form sculptured surfaces and, subsequently, ridges and flanges (Trueb, 1973). Because proboscideiis and bubahis possess com- plexly exostosed crania (Figs. 2 and 8), I suggest that the reduced amount of casquing in these species as contrasted to scutatus and fasciatus (Fig. 11) is the result of secondary reduction of ossifica- tion of a more extensively casqued skull which may have been similar to that of johnsoni (Fig. 14). Although not as extensively casqued as fasciatus and scutatus and less ornamented than pro- boscideus and bubalus, the skull of jolmsoni has developed dermal sculpturing to a degree that nearly all dermal cranial sutures are obscured; because of the sculpturing, it is impossible to ascertain externally whether or not the sphenethmoid is exposed dorsally. Two members of the genus exhibit species-specific specializa- tions; these are the parasphenoid keel of johnsoni (Fig. 15) and the pterygoid odontoids of scutatus. With respect to the skull shape, fasciatus is the only species in which the otic margin of the para- occipital process is about parallel to the longitudinal axis of the body. This feature results from a slight difference in the configura- tion of the process (compare Figs. 1, 2, 8, 11, and 14) as contrasted with the remaining species in which the margin of the flange lies approximately parallel to the maxillary arch. Hemiphractus scu- tatus and fasciatus are similar in having a more rounded canthus as contrasted with their congeners; furthermore, these two species have robust, short snouts in lateral view (Figs. 1 and 11) as contrasted with the long, depressed snouts of proboscideus (Fig. 2c) and bubalus (Fig. 8). Hemiphractus johnsoni (Fig. 14) has a snout shape which is intermediate between these two groups. With the exception of some minor specializations {viz., the parasphenoid keel, absence of prevomerine teeth, small postorbital indentation, and highly developed dermal sculpturing), it appears that on the basis of cranial structure, johnsoni is most similar to the ancestral condition and intermediate between two distinct hemiphractine lines — one (scutatus and fasciatus) committed to increased armor- ment of the cranium, and a second (proboscideus and bubalus) in which armorment (and consequently weight) has been decreased and ornamentation markedly incremented. The results of critical analysis of cranial characters largely are corroborated by data derived from study of the postcranial skeleton. The general configuration of the vertebral column ( Table 3; Fig. 4a ) is one adapted to terrestrial situations (Trueb, 1973) and derived from an advanced group. The length and general uniformity of the RELATIONSHIPS OF NEOTROPICAL HORNED FROGS 47 lengths of the transverse processes, especially those of the posterior presacrals and the relatively narrow expansion of the sacral dia- pophyses represent terrestrial and perhaps semi-arboreal adapta- tions; the procoelous, non-imbricate nature of indi\ddual vertebrae suggests relatively recent derivation from an advanced group. As on the basis of cranial architecture, Hemiphractus is divisible into two groups by the nature of the cervical vertebrae. The scutatus- fasciatus line is characterized by a distinct notch between the cervi- cal cotyles (Fig. 4c-e), whereas the johnsoni-proboscideiis-biibalus group lacks such a notch (Fig. 4f-g). With respect to the relative widths of the transverse processes, scutatiis is the most divergent, having a relatively narrow sacrum with respect to the widths of the presacral transverse processes. Inspection of table 3 shows that fasciatus and joJinsoni are more similar to one another than either is to huhalus or proboscideus. The adaptive significance of the relatively great heights of the neural spines is an enigma; there are no muscle attachments to these spines. Each species has its own distinctix'c profile with regard to the relative heights of the neural spines; those of bubalus and proboscideus are more similar to one another than are any of the others to one another. Hemiphractus fasciatus and scutatus differ from the remaining three species in that the overlying skin is loosely applied to the protruding vertebrae. The proportions of the urostyle and ilia are inextricably related. Examination of table 3 and figure 5 clearly illustrates that scutatus has, proportionately, the shortest urostyle and ilia, and fasciatus the longest. Proportional shortness of these elements normally is associated with a trend towards a walking or hopping gait (Trueb, 1973), whereas increase in length usually reflects a trend towards more efficient jumping or leaping. The ilium bears only an indis- tinct crest in scutatus as contrasted to the remaining species, es- pecially fasciatus. The proportional lengths of the urostyles and ilia of jolinsoni, bubalus, and proboscideus lie between the extremes of scutatus and fasciatus, suggesting that these species are more generalized in their pelvic structure than either of the other two. Like fasciatus, each of these species bears a distinct ilial crest; however, in none is it as well developed as in fasciatus. Among the three, the crest is least developed in joluisoni and best de\'eloped in proboscideus; this may suggest a trend from generalized to more specific arboreality within these three frogs. Although the pectoral girdle structure varies considerably in Hemiphractus (Fig. 3), the interpretation of this variation is as questionable as it is among anurans in general. Hemiphractus scu- tatus is most distinctive by virtue of its greatly reduced clavicles and obviously short scapulae (Fig. 3a). The clavicles are some- what better developed in fasciatus (Fig. 3b), but still moderate in comparison with those of bubalus, proboscideus (Fig. 3c) and John- 48 OCCASIONAL PAPERS MUSEUM OF NATURAL HISTORY soni (Fig. 3d); the scapulae of the latter three species are inter- mediate in length between those of scutatus and fasciatus. Hemi- phractus johnsoni is distinguished by less massive coracoids than in the other species. As with consideration of the crania, and axial and vertebral morphology, a basic dichotomy is evinced in pectoral morphology. Hemiphractus scutatus and fasciatus seem to repre- sent one evolutionary sequence and prohoscideus and hubalus a second, and johnsoni emerges as the least specialized of the species. With respect to foot morphology there is only one character worth mentioning, and that is the presence of a well-developed pre- hallux in scutatus. The presence of this structure suggests that scutatus digs with its hind feet, a habit not unexpected in a forest floor dweller. In general, the distribution of soft morphological features closely parallels those evident osteologically. For example, the fingers and toes of hubalus, prohoscideus, johnsoni, and fasciatus tend to have reduced webbing, whereas those of scutatus are distinctly webbed. Hempliractus scutatus differs from its congeners in having vestigial toe pads, the remnants of which are visible only on careful examina- tion of the ventral digital tips. A number of dermal characteristics tend to support the alliances suggested above. All hemiphractines are characterized by the presence of a proboscis; however, that of prohoscideus is proportionately larger and distinctly different in shape (Fig. 6) than those of the other species. Hemiphractus johnsoni and scutatus have a similar dermal configuration of the eyelids. Likewise, the eyelids of prohoscideus and hidjahis are the same, but distinct from those of johnsoni and scutatus and those characteristic of fasciatus ( Fig. 7 ) . Apparently there has been an evolutionary trend towards development of calcars on the heels within the prohoscideus-huhaUis lineage. They are well developed in the two latter arboreal species, whereas calcars are poorly de- veloped in johnsoni and fasciatus, and absent in scutatus. This seems to indicate a counter trend involving the loss of the calcar in the more terrestrial, giound-dwelling species. In contrast to the smooth ventral skin of johnsoni, the venters of prohoscideus and hubalus are finely granular. The venter of fasciatus bears scattered, low tubercles and that of scutatus is coarsely granular. Available ecological data concerning favored microhabitat haunts of the various Hemiphractus confirm expectations based on observed morphological features. Hemiphractus prohoscideus and huhalus are characterized by relatively light weight skulls, generalized axial and pelvic osteology, robust pectoral girdles, extensive dermal ornamentation and disruptive color patterns. These two species most frequently perch between 1.0 and 2.5 meters above the ground on small branches and twigs amidst dense vegetation (see figure 18); clearly, they are arboreal species. On the other hand, johnsoni RELATIONSHIPS OF NEOTROPICAL HORNED FROGS 49 Fig. 18. Three species of Hemiphractus. Top: H. bubalus. Middle: H. johnsoni female carrying eggs (photograph by Da\id B. Wake). Bottom: H. fasciatus. All approximately 1.0-1.5 times natural size. 50 OCCASIONAL PAPERS MUSEUM OF NATURAL HISTORY and fasciatus are found less than 0.7 meters above the ground and sometimes on logs or the forest floor itself. Hemiphractus scutatus is exclusively a forest floor, leaf-litter dweller. The latter three species are characterized by a predominantly reddish or grayish brown color scheme with irregular, darker dorsal mottling which cryptically conceals them. Of the three, johmoni (Fig. 18) is the most generalized morphologically, save for a few minor features of the skull, and is at least semi-arboreal. Both fasciatus (Fig. 18) and scutatus seem to have adapted for a terrestrial life style in the lower strata of the forest by retaining and/or incrementing a heavily armored skull and apparently developing distinctly different modes of locomotion. The attenuate, crested ilia of fasciatus suggest that it is more adept at leaping than any other member of the genus, especially scutatus. Because it has a moderately well-developed pectoral girdle and large adhesive discs, fasciatus obviously is arboreal, although less specialized for this mode of existence than either proboscideus or huhahis. The short, low-crested ilia, massive skull, well-developed prehallux and effective loss of adhesive discs in scutatus seem to represent an adaptive adjustment to the hopping and walking gait and burrowing habits usually associated with in- habitants of the leaf litter of the forest floor. In summary, I suggest that johnsoni is the most generalized member of the genus and probably most closely resembles in appear- ance and habit the common ancestor of the five species. It is note- worthy that johnsoni is the widest ranging species of the group both latitudinally and altitudinally; it occurs over an elevational range of 1610 meters, from Colombia to Bolivia. From a South American johnsoni-\ike ancestor, I propose that two lineages arose — one giving rise to proboscideus and bubalus, which adopted an arboreal mode of existence, and a second giving rise to fasciatus and scutatus, which successfully exploited a terrestrial, semi-arboreal life style in one case and a completely terrestrial existence in the other. I am not prepared to comment on the relative degrees of specialization of bubalus and proboscideus on the basis of the present data. Both inhabit the upper Amazonian slopes of the Ecuadorian Andes. They are not known to occur sympatrically, and on the basis of the most reliable locality records, it seems that proboscideus is primarily an inhabitant of lower montane and low- land rainforests, whereas bubalus is found most frequently in cloud forests. Both scutatus and fasciatus have wide distributions, the ranges of which are separated by the Andes, with scutatus being Amazonian and fasciatus being Pacific. Given that they had a common ancestor not unlike johnsoni, it seems reasonable that a more generalized prototype may have specialized for their respec- tive modes of existence as a result of isolation and concomitant competition with the myriad of other anurans which ha\e e\'ohed on and inhabit the South American landmass. RELATIOxXSHIPS OF NEOTROPICAL HORNED FROGS 51 INTERGENERIC RELATIONSHIPS On the basis of the literature surveyed throughout this report, it is evident that the familial allocation of Hemiphractus has been uncertain. To those familiar with the Leptodactylidae (see Lynch, 1971, for a review of this family), the hemiphractines seem similar in superficial appearance, certain aspects of cranial and axial archi- tecture, and habit to some of the ceratophryine leptodactylids. However, Hemiphractus differs by the combination of the follow- ing three characters from all known leptodactylids: 1) claw-shaped terminal phalanges, 2) intercalarx' cartilages, and 3) a reproductive mode unknown among leptodactylids. The possession of these three characters allies Hemiphractus with a second, widespread Neotropical family, the Hylidae. In the most recent and comprehensi\ e treatment of Neotropical hylids, Duellman (1970:18-19) di\ided the New World Hylidae into four subfamilies — the Hylinae, Ph\llomedusinae, Amphignatho- dontinae, and Hemiphractinae. The hemiphractines and amphi- gnathodontines are distinguished from the bylines and ph>lIomedu- sines (which together include the majority of known genera) by their extensixely ossified skulls, "brooding" mode of reproduction, and unique larval gill structure. This mode of reproduction, in which the eggs and larxae are transported on the dorsum of the female, and the modified lar\al gill structure were described first by Boulenger (1898) in GastrotJieca cornuta. A more complete analysis and survey of these characters were presented by Noble (1927). He proposed that those hylids in which 1) larvae are transported on the back of the female, 2) the dorsal epidermis of the female is attenuated and \asculated, and 3) the lar\'ae ha\e specialized "bell-shaped" gills constitute a natural phylogenetic unit. Noble considered this unit to be distinct from other hylids which exhibit a \'ariety of reproductive modes exclusive of the "brooding" habit. Several questions now arise. Do the so-called "brooding" hylids actually represent a single phvletic unit? If so, what are the evolutionary patterns within this line and the inter- relationships of the nine genera currently included in the group? It is difficult for me to conceive of an evolutionarv' course in which the combination of the terrestrial brooding habit and the specialized gill structure ( both of which are unique among anurans ) could have arisen independently more than one time from a group of advanced frogs. Therefore, without evidence to the contrary, it is most parsimonious to conclude that these anurans {i.e., Amphi- gnathodontinae and Hemiphractinae ) do represent a single phyletic line which has diversified considerably. Duellman (1970) included eight genera in the subfamily Amphignathodontinae; these are: AnotJieca, Nyctimantis, Fritziana, Flectonotus, Gastrotheca, Am- phignathodon, Cnjptohatrachiis and Stefonia. The inclusion of Ano- 52 OCCASIONAL PAPERS MUSEUM OF NATURAL HISTORY theca and Nijctimantis is problematic. The first genus, Anotheca (from southern Mexico, Costa Rica, and Panama), does not brood its young. Eggs are known to be deposited in bromcHads and water- filled tree cavities where the larvae undergo development; further- more, the larv^ae lack specialized gills. The reproductive mode of the second genus, Nijctimantis (known only from Ecuador), is unknown. Females do not possess a brood pouch, and neither eggs nor larvae have been associated with Nyctinuintis: however, my colleagues and I have heard and collected males calling from water- filled cavities in bamboo. Perhaps the reproductive mode of this frog is the same as that of Anotheca. Morphologically, both of these montypic genera share a great many features with Gastrotheca; thus, despite the apparently more generalized reproductive mode and larval gill structure, Duellman (1970) elected to include them in the Amphignathodontinae. The placement of Anotheca and Nyctimantis- in the Amphignathodontinae obviously is debatable. Without further study of the entire subfamily, I prefer not to re- arrange Duellman's classification; furthermore, the subfamilial allo- cation of these two genera is not especially pertinent to the inter- generic relationships of Heyniplnactus. In the third genus, Fritziana (from the mountains of south- eastern Brasil), a single, shallow basin forms on the backs of females. In this basin a cluster of unprotected eggs develops. The larvae are characterized by bell-shaped gills. Flectonottis (from the mountains of southeastern Brasil and the coastal cordillera of northern Venezuela and the continental islands of Trinidad and Tobago) affords more protection for the developing young. The females of this genus are characterized by a brood pouch with a medial, slit-like opening. The larvae undergo direct develop- ment in this pouch. The way in which eggs are placed on the back of the female is unknown in both of these genera. In both Fritziana and Flectonotus the first finger is either equal in length to the second, or the second is shorter than the first. Gastrotheca is a genus represented by at least 34 species which, as a group, are widespread; it occurs in southeastern Brasil, mod- erate to high elevations from northern Argentina to Colombia and Venezuela, and low to moderate elevations in northwestern South America and Panama. All female Gastrotheca possess a dorsal brood pouch with a posterior opening, and all known larxae have specialized gill structures. Mertens (1957) discussed the passive introduction (by gravity) of eggs into the pouch of Gastrotheca ovifera, and Hoogmoed (1967) described the way in which a male and female cooperate during amplexus to maneuver fertilized eggs into the female's pouch in Gastrotheca riohamhae. The larvae of some Gastrotheca undergo partial development in the pouch and subsequently are expelled from the pouch to complete develop- RELATIONSHIPS OF NEOTROPICAL HORNED FROGS 53 ment in an aquatic medium. In other species, the eggs undergo direct development into small frogs without an intermediate aquatic lar\al stage. In some members of the genus, the second finger is shorter than the first. Like Gastrotheca, Amphignathodon (known only from Ecuador ) is characterized by larvae with specialized gills and the presence of a brood pouch in females. Osteologically, Amphignathodon is similar to certain of the casque-headed Gastro- theca; it is distinguished from the latter and all other known anurans by the presence of true teeth in the lower jaw. The remaining two genera, Cnjptohatrachus and Stefania, have relatively limited distributions and a number of peculiar characters in common. Neither has a brood pouch, but each has larvae with specialized gills which undergo direct development on the back of the female. In both Cnjptohatrachus (from the Andes and Sierra Santa Marta of Colombia) and Stefania (from the Guyana Highlands), the eggs reside within indi\ idual shallow depressions on the backs of females. The way in which eggs are positioned on the backs of the females is unknown, although in one species, Stefania scalae, Rivero (1970:463-464) provided some indication of the manner in which the\- adhere to the back once in place. He presented evidence suggesting that the female secretes a porous membrane, or otherwise modifies the epidermis of her back; thus, the skin surface may serve as a "template" for the secretion of a substance by indi\ idual embryos that ensures their secure adherence to the dorsum of the female. Cnjptohairachus and Stefania, like Flectonotus, Fritziana, and some Gastrotheca, are characterized by having the first and second fingers equal in length or the second distinctly shorter than the first. The significance of this modification is unknown, but Rivero (1970:465) pointed out that this character- istic may in some way relate to the egg-carrying capacity of these frogs. The Hemiphractinae, as defined by Duellman (1970) contains only one genus, Hemipliractus. Relatixe to the eight genera dis- cussed above, it shares the most reproductixe characters with CryptobatracJius and Stefania. The large-yolked eggs reside in indi- vidual shallow depressions on the back of the female, and there undergo direct development. Furthermore, the second finger is in\ariably shorter than the first in both sexes of each species. Be- cause of the suite of unique osteological features distinguishing Hemip]\ractus from the other eight genera, it has been placed in a separate subfamily. Noble (1927) assumed that Cnjptohatrachus and Hemiphractus were primitive among the hyfids that brood their young, because the eggs are carried fully exposed on the backs of females. By logi- cal association, we now would ha\e to include Stefania and Fritziana in this group also. Noble further stated that among those species 54 OCCASIONAL PAPERS MUSEUM OF NATURAL HISTORY having larvae with expanded gills, those in which the gills are composed of four separate sheets of tissue were more primitive than those in which either the posterior pair of gill sheets are lost (in at least one species of Cryptohatrachus) or fusion of the gill sheets had taken place. Hemiphractus is an example of the latter condi- tion. Each pair of gill sheets has fused, forming "bells" of vascular tissue laterally; together these bells enclose the ventral half of the embryo ( Noble, 1927 ) . Noble considered those frogs which brood their young within protective pouches to be the most advanced. This group includes Flectonotiis, Gastrotheca, and Amphignathodon. The gills of embryonic Gastrotheca are like those of Hemiphractus; the details of the larval gill structure of Flectonotiis and Amphi- gnathodon are unknown. It is evident that much remains to be learned about the breeding biology in general, and the specific aspects of reproduction and development of all of these frogs. Furthermore, detailed systematic studies of each of the genera should be executed before the status of their phylogenetic relationships can be assessed with any con- fidence. On the basis of the available information, it would seem that Anotheca and Nijctimantis might represent early derivatives of a hylid line committed towards evolution of behavioral and morpho- logical traits associated with parental care of the young. Thus, although Anotheca and Nijctimantis are not adapted morphologi- cally to keep their young with them, they may have elected an al- ternative solution, that is, to deposit eggs in places affording them some protection such as bromeliads and tree-cavities. A more ad- vanced derivative of this general hylid line might be represented by Fritziana in which larvae having specialized gills are carried in a shallow depression on the dorsum of the female. An anuran like Fritziana well may represent a reproductive and morphological prototype from which at least two, more specialized evolutionary lines could have been derived. The first and more primitiv^e of these probably is represented today by Cryptohatrachus, Stefania, and Hemiphractus. Members of this lineage failed to develop a maternal brood pouch, but did develop a means of affording indi- vidual embryos an intimate and discrete attachment with the female. Hypothetically, a second, derived lineage may be repre- sented by Flectonotiis, Gastrotheca and Ampliignathodon. The simplest form of a maternal brood pouch seems to be that of Flec- tonotus in which the dorsal skin of the female forms lateral flaps which occlude medially. The pouches of Gastrotheca and Am- phignathodon are complete, having only a posterior opening and affording the developing larvae the most protection of any of these hylids. RELATIONSHIPS OF NEOTROPICAL HORNED FROGS 55 SUMMARY Between 1824 and 1945 twelve nominal hemiphractine species were described. Of these twelve, five are recognized in this paper as comprising the genus Hemiphractus; these are: Hemipliractus huhalus (Jimenez de la Espada), H. fasciotus Peters, H. joJuisoni (Noble), H. proboscideus (Jimenez de la Espada), and H. scutatus ( Spix ) . The species are readily distinguished on the basis of cranial characters in addition to prevailing differences in sizes, proportions, tuberculation, and coloration. The genus is widespread, occurring in the upper Amazon Basin of Colombia, western Brasil, Ecuador, Peru and Bolivia, the Andes of northern Colombia, the Pacific slopes of northwestern Ecuador, Colombia and eastern Panama, and the Caribbean slopes of western Panama. Hemiphr actus occurs between elevations of 250 and 1S30 meters, and is an inhabitant of the lower strata of lowland rainforests and lower montane and cloud forests. Hemiphractus is allied with the Hylidac by possession of claw- shaped, terminal phalanges, intercalary cartilages, and a reproduc- tive mode which involves direct development of larvae with spe- cialized gills on the dorsum of the female. Osteologically, members of this genus differ from all other known hylids by the unique con- figuration of cranial castjuing, the presence of pter>'goid odontoids (in one species), the shape of the prexomers, the presence of an enlarged pair of mandibular odontoids and tooth-like mandibular serrations. They dift^er further by the presence of greatly enlarged neural spines. Because of this combination of characteristics, this genus is retained in a separate subfamily, the Hemiphractinae. Despite the unique suite of morphological characters ascribed to the Hemiphractinae, this hylid subfamily seems to be closely related to the Amphignathodontinae on the basis of their common repro- ductive mode. Both subfamilies differ from the two remaining New World hylid subfamilies ( the Hylinae and Phyllomedusinae ) in that the amphignathodontines and hemiphractines demonstrate ethological and morphological c\olutionary trends towards parental care of the young. In the simplest stage of this progression, eggs are deposited in protected environmental situations {e.g., water- filled cavities in trees). A significant advance is shown by those genera in which embr^'os are carried about exposed, on the backs of females. The embryos undergo direct de\'elopment and are equipped with specialized, "bell-shaped" gills which serve as an interface between the developing larxae and the maternal epi- dermis. In the most derived form of parental care, eggs are carried within a brood pouch on the dorsum of the female and, thus, are afi^orded maximal protection during their development; these larvae also are equipped with specialized gills. These data suggest that 56 OCCASIONAL PAPERS MUSEUM OF NATURAL HISTORY the Amphignathodontinae and Hemiphractinae represent a single phyletic line distinct from their Neotropical relatives, the Hylinae and Phyllomedusinae. RESUMEN Doce especies nominales de ranas hemiphractines se describieron entre 1824 y 1945. Cinco de ellas se reconocen en el presente tra- bajo en el genero Hemiphr actus; estas son: Hemiphractus hiibalus (Jimenez de la Espada), H. fasckitus Peters, H. johmoni (Noble), H. proboscideus ( Jimenez de la Espada ), y H. scutatus ( Spix ) . Las especies se pueden distinguir facilmente en base a las caracteristicas craneales, ademas de las diferencias en tamanos, proporciones, tu- berculos de la piel y coloracion. El genero esta ampliamente dis- tribuido, pues ocurre en la depresion alta amazonica de Colombia, occidente de Brasil, Ecuador, Peru, y Bolivia, los Andes al norte de Colombia, las faldas montaiiosas nor-occidentales de Ecuador, Colombia, y este de Panama. El genero Hemiphractus se distribuye a elevaciones que oscilan entre 250 y 1S30 metros e igualmente se encuentra en los bosques lluviosos, montanos y nublados. Hemiphractus esta emparentado con la familia Hylidae, ya que posee falanges terminales en forma de uira cartilagos intercalares, y un modo reproductivo caracterizado por desarrollo directo de la larva, con branquias especializadas, en el dorso de la hembra. Os- teologicamente, los miembros de este genero se diferencian de otros hilidos conocidos por tener una configuracion I'mica del casquete craneal, presencia de odontoides pterigoideos (en una especie), forma de los prevomeres, presencia de un par agrandado de odon- toides mandibulares, y endentaduras mandibulares en forma de dientes. Ademas, ellos se diferencian por poseer espinas neurales muy dilatadas. Tomando en cuenta esta combinacion de carac- teristicas, el genero se mantiene en una subfamilia separada, la Hemiphractinae. A pesar del grupo de caracteristicas que permiten distinguir a Hemiphractinae, esta subfamilia de hilidos parece estar muy em- parentada con Amphignathodontinae, en base a su modo reproduc- tivo comun que poseen. Ambas subfamilias difieren de las otras dos subfamilias de hilidos del Nuevo Mundo (Hylinae y Phyllo- medusinae) en que elllas tienen tendencias evolutivas etologicas y morfologicas hacia el cuidado parental de las crias. El primer paso hacia esta tendencia lo muestran aquellas especies que de- positan sus heuvos en habitaculos protegidos [i.e., cavidades con agua en los troncos de los arboles). Un avance significativo lo muestran aquellos generos cuyas hembras cargan los embriones en la espalda, pero carecen de bolsa o marsupio. Los embriones tienen desarrollo directo y estan en contacto con la epidermis materna mediante branquias campaniformes especiales. En la forma mas RELATIONSHIPS OF NEOTROPICAL HORNED FROGS 57 derivada de cuidado parental, hembras llevan los huevos dentro de bolsas o marsupios dorsales, de tal manera que hay una pro- teccion maxima de los huexos durante el desarrollo; estas larvas tabien estan equipadas con branquias especiales. Estos datos su- gieren que Amphignathodontinae y Hemiphractinae representan a una sola linea filetica distinta de sus relativas neotropicales, Hylinae y Phyllomedusinae. SPECIMENS EXAMINED Locality data for each specimen examined are listed in the fol- lowing paragraphs. The data are arranged as follow: alphabetically by country, state (department or province), and locality; alpha- betically by the first letter in the abbre\iations for the museums, and numerically after each museum abbreviation. Specimens lack- ing precise locality data are listed in the first most restricted political unit possible. Unless otherwise noted, all specimens are preserved in alcohol. Hemiphractus buhalus ECUADOR: Napo: Archidona, MNCN 176; Ri'o Azuela, 1740 ni, BMNH 1973.2850, KU 143107-17; San Jose de Moti, MNCN 175; SE slopes Cerro Siunaco, CAS 122207-08; S slope Cordillera del Due above Rio Coca, 1150 m, KU 123160-68, 124140 (skeleton); "Zarajioto, Orillas del Napo," MNCN 174. Pastaza: Canelos, 300 m, MCZ 17937; headwaters of Rio Arajuno, CAS 87364; headwaters of Rio Arajuno, about 10 km NE Puvo, CAS 122213; Mera, 1000 m, UMMZ 92109; Rio Alpayacu, 1 km E Mera, 1100 m, KU 120763; Rio Sandalia, tributar>- of Rio Puyo, USNM 193287. Hemiphractus fasciatus COLOMBIA: Antioquia: Alto de la Honda, near San Rafael, LACM 47147; 8 km S Valdixia, 1450 m, FMNH 63850. Cauca: La Costa, El Tambo, NHRM 1981. ECUADOR: Esmerddas: near the lower Rio Blanco [0° 02-03' N, 79° 10-15' W], CAS 122206; near the upper Cachabi [0° 56-58' N, 78° 35-38' W], CAS 122205. Pastaza: "Pastaza \'alley," ZSM 36/0; Mindo, UMMZ 55523. PANAMA: Bocas del Tow: 18.6 km W Almirante, upper Ri'o Changena camp, CAS 93836, CAS-SU 21772; north slope Cerro Pando, 1450 m, KU 107426; Rio Changena, 650 m, KU 128077 (cleared and stained). Co/on: Signal Loma (Loma Peak), 5 km S San Isabel, USNM 55320-21. Darien: north ridge Cerro Cituro, Serrania de Pirre, 900-1320 m, KU 107408-422, 107425; southeast slope Cerro Pirre, 1440-1450 m, KU 93504-508, 93509-10 (skeletons). Panama: Altos de Pacora, 800 m, KU 77107, 93503. San Bias: Camp Summit, 300-400 m, KU 116353-5. Hemiphractus johnsoni BOLIVIA: "Linimbane," 1000 m, BMNH 1901.8.2.51. COLOMBIA: Antioquia: Santa Rita Creek, 22.5 km N Mesopotamia, 1890-1910 m, AMNH 1341-42, 1343 (skeleton). ECUADOR: Pastaza: Canelos, 300 m, BMNH 1880.2.5.192. PERU: No specific localitij: "Eastern Perri," BMNH 1902.5.29.204. Cuzco: Marcapata Valley, KU 153436 (skeleton). Huduco: S slope Serrania de Sira, 690 m, KU 154671-72. Loreto: Rio Tapiche, Monte Alegre, AMNH 42862; 58 OCCASIONAL PAPERS MUSEUM OF NATURAL HISTORY Rio Ucayali, Cashiboya, AMNH 43452; Tangarana Hills, Alto Utoquinia, AMNH 58631 (adult female with series of 18 young). Pasco: Chontilla, Iscozazin Valley, USC-CRE 716, 717 (partial egg clutch). Hemiphractus proboscideus ECUADOR: No specific locality: "Eastern Ecuador," AMNH 52931; "Rio Pastaza watershed," NHRM 1911. Mowna-Santiago: Chiguaza, CAS 122211. Napo: Bemiejo No. 4, 720 m, 15 km ENE Umbaqui [77° 22' W, 00° 11' N], KU 123149-54, 124139 (skeleton); Lago Agrio, KU 126326; Loreto, CAS 122212; Puerto Libre, Rio Aguarico, 570 m, KU 123155-59; Rio Coca, 300 m, UMMZ 92108; Rio Napo, 300 m, UMMZ 92110; Santa Cecilia, 340 m, AUM 8046-51, BMNH 1973.2851, KU 104774-77, 106970, 109336-38, 111627-28, 123134-47, 124135-38 (skeletons), 126325, 146265-70, 150616-19; slopes of Cerro Sumaco, USNM 193288; Sumaco, MNCN 173. Pastaza: Copataza, a tributary of the Rio Pastaza, CAS 122209-10; Rio Pastaza, Andoas, AMNH 52929; Sarayacu, CAS-SU 11433. Hemiphractus scutatus BRASIL: Aitjazonas: Igarape Belem, near Rio Solinoes, ca. 70 km E Leticia, AMNH 85375-76. ECUADOR. Mowna-Santiago: Macuma, KU 129751, 129752 (skeleton), UIMNH 59616-17; Rio Piuntza, KU 147117-18; Taisha, Macas, USNM 193577. Napo: ? Archidona, USNM 193582; Amienia, 250 m, UMMZ 92105; Boca, Rio Coca, 300 m, UMMZ 92104; Loreto, CAS-SU 11434, USNM 193.578, 193584; San Jose de Moti, MNCN 192 (lot of two preserved specimens and one skeleton); SE slope Cerro Sumaco, CAS 87365, 87366 (skeleton), CAS-SU 11435, USNM 193583; top of Cerro Sumaco, San Jose Aliajo, AMNH 22210. Pastaza: Abitagua, UMMZ 92106; Alpayacu, 300 m, UMMZ 92103; bet^veen Abitagua and Rio Pastaza Valley, UMMZ 92107; Rio Arajuno, headwaters, USNM 193580; Rio Bufeo, tributary of Baja Rio Bolwnaza, USNM 193576; Rio Rutimo, tiiluitary of Rio Bobonaza, USNM 193580; Rio Talin, headwaters of Rio Bobonaza, USNM 193579. Tiinguiahtia: Mirador, AMNH 38527; Palmera, Pastaza River Valley, 1333 m, AMNH 39488. PERtJ: Ciizco: Quince Mil, Boca Rio Marcapata, FMNH 121855. Loreto: Caballococha, AMNH 43044; Cadena de Manseriche, AMNH 43403; Iquitos, AMNH 42768; middle Rio Utoquinia, CAS-SU 9332. NO DATA: UMMZ s-1015 (skeleton). LITERATURE CITED Andersson, L. G. 1945. Batrachians from east Ecuador. Arkiv Zool., 37A(2):l-88. BOULENGER, G. A. 1882. Catalogue of the Batrachia Salientia s. Ecaudata in the collection of the British Museum, ed. 2, London, xvi + 503 pp. 1898. An account of the reptiles and batrachians collected by Mr. W. F. H. Ro.senberg in Western Ecuador. Proc. Zool. Soc. London, pp. 107-126. Brocchi, M. 1877. Recherches sur I'ostcologie d'un batracien anoure provenant du Bresil {Hemiphractus). Ann. Sci. Nat., ser. 6, 5(7):1-18. Cope, E. D. 1866. On the structure and distribution of the arciferous Aniua. Jour. Acad. Nat. Sci. Philadelphia, ser. 2, 6:67-112. 1868. An examination of the Reptilia and Batrachia obtained liy the Orton expedition to Ecuador and the upper Amazon, with notes on other species. Proc. Acad. Nat. Sci. Pliiladelphia, 20:96-140. RELATIONSHIPS OF NEOTROPICAL HORNED FROGS 59 DUELLMAN, W. E. 1970. The hylid frogs of Middle America. Monogr. Mus. Nat. Hist. Univ. Kansas, l:xii + 753 pp. DuMERiL, A. M. C. and G. Bibron 1841. Erpetologie Generale ou Histoire naturelle complete des reptiles. Paris, vol. 8, 792 pp. Gadow, H. 1901. Amphibia and Reptiles. London, xiii + 608 pp. Gravenhorst, J. L. C. 1825. Stombus, eine neue Amphibien Gattung. Isis von Oken, 1825:920- 922. GUNTHER, A. C. L. G. "1858" [1859]. Catalogue of the Batrachia Salientia in the collection of the British Museum. London, xvi + 160 pp. HOOGMOED, M. S. 1967. Mating and early development of Gastrotheca marsupiata (Dumeril and Bibron) in captivity ( Hylidae, Anura, Amphibia). Brit. Jour. Herpetol., 4(l):l-7. Jimenez de la Espada, M. 1871. Faimae neotropicalis species quaedam nondum cognitae. Jour. Sci. Math. Phys. Nat., Acad. Real Sci. Lisboa, 3:57-65. Lynch, J. D. 1971. Evolutionary relationships, osteology and zoogeography of lepto- dactyloid frogs. Misc. Publ. Univ. Kansas Mus. Nat. Hist., 53:1-238. Mertens, R. 1957. Zur Naturgeschichte des venezolanischen Riesen-Beutelfrosches, Gastrotheca ovifera. Zool. Garten (N. F.), 23:110-133. MiRAXDA-RlBEIRO, A. DE 1926. Notas para servirem ao estudo dos gymnobatrachios (Anura) Brasilieros. Arch. Mus. Nac, Rio de Janeiro, 27:1-227. Myers, C. W. 1966. The distril:)ution and behavior of a tropical horned frog, CeratJiyJa panamensis Stejneger. Herpetologica, 22(1):68-71. Myers, G. S. and A. L. de Garvalho 1945. Notes on some new or little-known Brazilian amphibians, with an examination of the history of the Plata salamander, Ensatina platen- sis. Bol. Mus. Nac, Zool., 35:1-24. Noble, G. K. 1917. The systematic status of some batracliians from South America. Bull. Amer. Mus. Nat. Hist., 37(30) :793-814. 1926. An analysis of tlie remarkable cases of distribution among the Amphibia, with descriptions of new genera. Amer. Mus. Novitates, 212:1-24. 1927. The value of life history data in the study of the evolution of tlie Amphil^ia. Ann. New York Acad. Sci., 30:31-128. 1931. The liiology of the Amphiljia. McGraw-Hill Book Co., Inc., New York, xiii + 577 pp. Peters, W. 1862. Uber die Batrachier-Gattung Hemiphractus. Monatsber. Akad. Wiss. Berlin, pp. 144-152. RiVERO, J. A. 1970. On the origin, endemism and distribution of the genus Sfcfania Ri\ero (Amphil)ia, Salientia) witli a description of a new species from southeastern Venezuela. Bol. Soc. Venezolana Cien. Nat., 28( 117/118 ):456-481. 60 OCCASIONAL PAPERS MUSEUM OF NATURAL HISTORY SpIX, J. B. VON 1824. Animalia nova, sive species novae testudinum et ranarum, quas in itinera per Brasiliam, annis 1817-1820. Monaco, pp. 1-53. Stejneger, L. 1917. A new species of horned tree-frog from Panama. Proc. Biol. Soc. Washington, 30:31-33. Trueb, L. 1970. Evolutionary relationships of casque-headed tree frogs with co- ossified skulls (family Hylidae). Univ. Kansas Publ. Mus. Nat. Hist., 18(7):547-716. 1971. Phylogenetic relationships of certain Neotropical toads with the description of a new genus (Anura: Bufonidae). Los Angeles Co. Mus. Contrib. Sci., 216:1-40. 1973. Bones, frogs, and evolution. In Evolutionary biology of the anurans. Contemporary research on major problems. J. L. Vial, editor. Uni- versity of Missouri Press, Columbia, Missouri, pp. 65-132. Trueb, L. and W. E. Duellman 1971. A synopsis of Neotropical hylid frogs, genus Osteocephahis. Occas. Papers Mus. Nat. Hist. Univ. Kansas, 1:1-47. Vial, J. L. 1973. Evolutionary biology of the anurans. Contemporary research on major problems. James L. Vial, editor. University of Missouri Press, Columbia, Missouri, vii -f- 470 pp. Wagler, J. G. 1828. Auziige aus einem systema amphibiorem. Isis von Oken, columns 740-744. 1830. Natiirlisches System der Amphibien mit vorangehender Classification der Saiigethiere und Vogel. Munich, 354 pp. University of Kansas Publications MUSEUM OF NATURAL HISTORY The University of Kansas Publications, Museum of Natural History, beginning with volume 1 in 1946, was discontinued with volume 20 in 1971. Shorter research papers formerly pub- lished in the above series are now published as Occasional Papers, Museum of Natural History. The Miscellaneous Pub- lications, Museum of Natural History, began with number 1 in 1946. Longer research papers are published in that series. Monographs of the Museum of Natural History were initiated in 1970. All manuscripts are subjected to critical review by intra- and extramural specialists; final acceptance is at the discretion of the publications committee. Institutional libraries interested in exchanging publications may obtain the Occasional Papers and Miscellaneous Publica- tions by addressing the Exchange Librarian, University of Kan- sas Library, Lawrence, Kansas 66045. Individuals may pur- chase separate numbers of all series. Prices may be obtained upon request addressed to Publications Secretary, Museum of Natural History, University of Kansas, Lawrence, Kansas 66045. Editor: Richard F. Johnston" PRINTED BY UNIVERSITY OF KANSAS PRINTING SERVICE LAWRENCE, KANSAS % 3 2044 093 361 657