Anurans from the Early Tertiary of Patagonia

ANURANS FROM THE EARLY TERTIARY OF PATAGONIA

BOBB SCHAEEFER

BULLETIN OF THE AMERICAN MUSEUM OF NATURAL HISTORY VOLUME 93: ARTICLE 2 NEW YORK: 1949 ANURANS FROM THE EARLY TERTIARY OF PATAGONIA ANURANS FROM THE EARLY TERTIARY OF PATAGONIA

BOBB SCHAEFFER Assistant Curator of Fossil Vertebrates Department of Geology and Paleontology

PUBLICATIONS OF THE SCARRITT-PATAGONIAN EXPEDITION, NUMBER 37

BULLETIN OF THE AMERICAN MUSEUM OF NATURAL HISTORY VOLUME 93 : ARTICLE 2 NEW YORK: 1949 BULLETIN OF THE AMERICAN MUSEUM OF NATURAL HISTORY Volume 93, article 2, pages 41-68, text figures 1-7, plates 16-19, table 1 Issued March 28, 1949 Price: $0.50 a copy CONTENTS

INTRODUCTION ...... 47

SYSTEMATIC DESCRIPTIONS ...... 49 Family Leptodactylidae . 49

Eophractus, new genus ...... 49 Eophractus casamayorensis, new species .49

Calyptocephalella Strand ...... 50 Calyptocephalella canqueli, new species .50

Eupsophus Fitzinger ...... 55

Eupsophus sp...... 55

Neoprocoela, new genus ...... 57 Neoprocoela edentatus, new species.57 DISCUSSION ...... 59

Anuran Differentiation ...... 59

Anuran Dispersal ...... 64

Trends in Anuran Evolution ...... 65

REFERENCES ...... 67

45 Page___ is blank. INTRODUCTION TEE: SOUH AMEixcAN CONTINENT has thus From this incomplete but representative list far yielded a small number of fragmentary it is evident that relatively few characters fossil anurans. Teracophrys (nomen nudum) are actually available in working out the was listed byAmeghino (1901) from theUpper affinities of a fossil anuran. The nature of Oligocene of Patagonia. The specimens, ap- the vertebrate column can often be deter- parently referable to the Leptodactylidae, mined, as well as a few skull characters, un- cannot be located in the Ameghino Collection fortunately rarely including the structure of of the Museo Argentino de Ciencias Natu- the prevomer. By using the criteria men- rales (Miss Noemi Cattoi, personal communi- tioned by Noble (1930), it may be possible cation). Ceratophrys prisca has been de- to postulate an arciferal or firmisternal con- scribed by Ameghino (1899) and Rovereto dition for the shoulder girdle. The terminal (1914) from the Upper Pliocene of Monte phalanges may or may not be present. Thus Hermoso, Argentina, and C. ensenadensis by the opportunity for exact identification is Rusconi (1932) from the Pleistocene (Ense- dependent upon the nature and completeness nadan) near Buenos Aires. of preservation and on the possibility of mak- Both Noble (1928) and Dunn (1931) have ing such identification without recourse to emphasized the importance of paleontologi- structures usually not fossilized. cal discovery in determining the approximate Parker (1940) has pointed out the difficul- times and routes of anuran dispersal. They ties associated with the interpretation of fos- have also pointed out the danger of assuming sil anuran material, and it is obvious that a that the failure to discover fossil anurans conservative approach in the matter of tax- belonging to particular groups is proof that onomic relationship is desirable unless a posi- such groups never existed in a given area. tive identification can be made, as in the case Dunn wisely states (ibid., p. 118), ". . . great of the Calyptocephalelka described in this weight should be given to positive evidence, paper. It might be pointed out that detailed and very little to negative evidence." It is of information on the osteology of the modern course true that in the absence of fossil data anurans is widely scattered and far from certain reasonable conclusions can be drawn complete. Herpetologists could render pale- on the basis of the distributional pattern of ontology (and possibly themselves) a great the present day faunas. There is, for instance, service by making comprehensive studies good reason for believing that the Discoglos- available. A single fossil discovery may be of sidae never entered North America and that the greatest importance in working out the the Pelobatidae did not invadeSouthAmerica. ramifications of anuran evolution and in de- The pelobatids were present in NorthAmerica termining routes of dispersal, providing a by the Pliocene (Taylor, 1937) but apparently reasonably accurate identification can be have migrated no farther south than Mexico. made. This is too often not possible and, as a The main taxonomic subdivisions of the survey of the literature indicates, has led to Anura now generally recognized are based many differences of opinion among competent on the five different types of vertebral col- students of the subject. umn occurring within the order (Nicholls, The present study is based on a collection 1916; Noble, 1922, 1931). The smaller cate- of lower Eocene and Middle and Upper gories are usually defined on suites of char- Oligocene anurans obtained by Dr. George acters involving the skull (particularly the Gaylord Simpson in 1930 and 1934 while on prevomer, the presence or absence of teeth, the First and Second Scarritt Patagonian and the middle ear), the shoulder girdle, the Expeditions. The specimens consist of nature of the terminal phalanges, the tongue, crushed and mostly dissociated skeletons the thigh musculature, the parotid gland, the embedded in very fine lacustrine or fluviatile presence or absence of Bidder's organ, and sediments. certain external features and proportions. The writer is greatly obligated to Dr. E. 47 48 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 93

R. Dunn for a number of helpful suggestions supplied important information from his and for reading the manuscript. He is like- field books and from preliminary notes made wise Indebted to Mr. K. P. Schmidt, Mr. on this collection. The drawings were pre- Arthur Loveridge, Mr. C. M. Bogert, and pared by Dorothea Kay Barlow and the Dr. J. A. Oliver for the loan of Recent speci- photographs were taken by Elwood Logan. mens for comparative purposes. Dr. Simpson SYSTE,MAT,I'IC DESCRIPTIONS FAMILY ?LEPTODACTYLIDAE lescrilbes the "Bird Clay" in his field notes BOPBRACTUS, NEW i as a "hardl, mlassive, bright green clay with intercalatedl )andss of whiite tuff." This la- GECNOTYPE;: EIophrahtus cas5amayorensis, new custrine deposit is exposel at the south end species. of Caffad16n 1londo near Paso Niemann, (;JNERIC DIA(GNOSIS: Ani anuirtan of leIpto- Trerritory of Chuhut, Arge¢ntina (see locating dactylidl or possibly hlylidi affinities. Ouiter diagram, Schaeffer, 1947). The anuran frag- surface of known (lermial hones of sktll(cov- ments were found in the white tuff, ere(l with (lee) rotinledi (lepressions which SPECIFIC DIAGNOSIS: Same as for genus. are surround( 'd by naIrrow, sharp, connecting DEscRIPToN: These fragments, consisting ridges. Portion of nasal bicone between orbit of the associated elements representing the an(I nares widle; nasals in 'ontItact alonig nidl- type andl in adldition a second partial maxil- dorsal line. Anterior atn(l msedlian border of lary andi a single vertebra (A.M.N.H. No.

I'JO. 1. Eaphrttdus casumayorensis, new genus and species. A..NI.IN . No. .3165, type specitnens. A. )orsal view of isolated le(ft. nsilKal in artivulation with Issoxiatednmaxillary illustrated be- low. 1Probsahl}e shape., of fronrit part of skull indicated. B. I.ateral view of inconipilete left mtiaxillary. X3/2. orbit elevtated. M.axillary deep, with teeth; 3164), represent the oldest known anuran ascen(ling l)roe3s >of m. xillary robust, indli- remains from South America. Tlhe few really cating at lea-st partital roofing of the temnporal diagnostic characters observable indicate area, an(d separated from thiickene(d(ldeig- affinity with either the Leptodactylidae or erous portion by shallow gr(x)v. Single the Hiylilae. Of the six anuran families repre- known vertebrapr(xa(pIreous, with cylindrical sented in South America having an entirely centrum. or partly procoelous vertebral column, the Eophractus tenolency towards a secondlary deposition of castamayerensit, niew species dlermal bone over the temporal area occurs Typjr: A.M.N.HI. No. 3165, associated left only in the two just mentioned. The ascendling nasal and left maxillary. process on the Eophractus maxillary is broa(l HORIZON AND LoCALITY: "B3ird (lay, andl high andi may possibly have an articular Casamayoran StagRe, Lower Loerne. Simpson surface for the squanlosal, but this is not 49 so BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 93 certain. In any case, at least partial roofing Calyptocephdeltak STRAND, 1926, Arch. Natur- of the temporal area is clearly indicated. The gesch., div. A, vol. 92, no. 8, p. 55 (new name size and depth of the maxillary and the form for CalyptocephalusDumeril and Bibron, 1841, of the ascending process are very similar to preoccupied). the condition in Calyptocephalelka, the only GENOTYPE: Calyptocephalus gayi Dum6ril Recent South American leptodactylid with a and Bibron. roofed temporal region. The Australian Cyclorana has a similar broad articulation Calyptocephalella canqueli, new species between the maxillary and squamosal (Par- TYPE: A.M.N.H. No. 3429, partial skele- ker, 1930). Among the Neotropical Hemi- ton consisting of crushed skull, right half of phractinae there are various degrees of roof- pectoral girdle, right fore limb, and first three ing, ranging from Amphignathodon with none vertebrae, also isolated vertebrae and limb to Hemiphractus in which the temporal area fragments belonging to same individual. is completely covered. HoRizoN AND TYPE LOCALITY: Sarmientan The nature of the dermal bone sculpturing Group, Deseadan Stage, Lower Oligocene. is highly variable in both the Leptodacty- Scarritt Pocket, southwest portion of Rin- lidae and the Hylidae, although it does ap- conada de los Lopez, west side of Sierra pear to be constant within a genus. The Canquel, central Chubut, Argentina. pitted ornamentation of Eophractus is some- SPECIFIc DIAGNOSIS: Very close to Calyp- what similar to that occurring in very small tocephalella gayi in all observable characters, subadults of CalyptocephaleZla (Reinbach, but differing from the Recent species in the 1939) but is very different from the fully greater width of the nasal bone between the adult denticulated pattern. It agrees most orbit and the nares, the shape of the orbit closely with the sculpturing of Amphigna- in adult individuals, which is narrower in thodon sp. from the Oligocene and Miocene relation to length, and the shape of the pos- of Franoe (Piveteau, 1927) and A. guentheri terolateral projection of the squamosal that is from Ecuador. relatively larger and more rounded in C. The elevation of the orbital rim is charac- gayi. teristic of the hylids rather than the lepto- DESCRIPTION: The most striking similarity dactylids, as is the groove on the outer between the fossil and the recent species is surface of the maxillary. A broad contact the ornamentation of the cranial roof. between the nasals occurs in various genera Reinbach (1939) has pointed out that in of both families (e.g., Calyptocephalella, subadults of C. gayi the surface of the dermal Hemiphractus). bones is covered with shallow depressions The skull of Eophractus was heavily os- which are separated from one another by sified and in general appearance and pro- rather sharp and narrow ridges. With increase portions must have resembled that of in age, denticles develop on the nrdges which Calyptocephalella very closely. It is thus very gradually encroach upon the depressions and tempting to consider that the two genera are finally obliterate them. The skull roof of closely related, particularly since the Oligo- C. canqueli is completely denticulated as is cene Calyptocephalella canqueli, tobedescribed the skull of a specimen of C. gayi of approxi- next, has a wvide nasal bridge. There is no mately the same size. proof for such an assumption, however, and The only preserved skull of C. canqueli the true affinities of Eophractus must await lacks the premaxillaries, excepting for the the recovery of additional specimens. ascending processes, and also the left maxil- HYPODIGM: The type and A.M.N.H. No. lary. The ascending processes are relatively 3164, partial left maxillary and single pro- broad as in C. gayi. The dentigerous right coelous vertebra. maxillary has a more extended articulation FAMiLy LEPTODACTYLIDAE with the nasal than the Recent species, while the contact with the squamosal is of about CALYPTOCEPHALELLA STRAND the same length. As pointed out above, the Calyptocephalus DuMEREL AND BIBRON, 1841, nasal bridge between the orbits and the nares Erp6tologie g6n6ral, p. 450. is wider in the fossil form, a character as- 1949 SCHAEFFER: ANURANS FROM TERTIARY OF PATAGONIA 51 sociated with the longer naso-maxillary arti- The shape of the orbit in C. canqueli and culation. A logarithmic plotting of the length in C. gayi is related to the length of the con- of the nasal along the middorsal line to the tact between the frontoparietal and the

SC/

vo? B

/ rodul

_,hu

FIG. 2. Calyptocephklella canqueli, new species. A.M.N.H. No. 3429, type specimen. A. Dorsal view. B. Ventral view. Abbreviations: cl, clavicle; cor, coracoid; frpr, frontoparietal; hu, humerus; mand, mandible; na, nasal; pt, pterygoid; radul, radio-ulna; sc, scapula; spheth, sphenethmoid; vo?, vomer. X1/1. width of the nasal bridge for five skulls of squamosal. When this suture is relatively C. gayi of different sizes and for the single short, the posterior rim of the orbit is ex- specimen of C. canqueli (fig. 3A) further tended backwards. As the logarithmic plot- suggests that the nasal of the latterexhibits ting of orbit length to orbit width in C. gayi a slightly different relative growth pattern. demonstrates, the orbit increases in width 52 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 93 more rapidly than in length with increase in about the same relative size in the Oligocene skull size. The single record for C. canqueli and Recent species. The posterolateral corner suggests a slower width increase, although of the squamosal in canqueli is neither ex- this obviously cannot be proved on the basis tended posteriorly nor rounded as in C. gayi of a single point. In any case, the shape of the but is more angular and has a slight lateral orbit and the narrower frontoparietal- extension. In this respect, canqueli again squamosal contact in C. canqueli are very agrees with the young adult condition in C. similar to the condition found in small sub- gayi. The arm of the squamosal articulating adults of C. gayi. with the quadrate, with due allowance for The frontoparietals and squamosals are of crushing, is of the same length and robustness

8i 7. 30f L. 6 5. + 20+ 0 -C -4-4-f 4. 4- 3. I0.9. F+ .0 2 f 4- 8' S-: A L- 7. o 65 B 5- 4 0 a a, al . -JL I ,tr- 3 4 5 6i760 20 6 20 30 Length na. Orbit lenqth

4 ^_w430 70-|ts|@||-|--- 60' I20 450 2Q 30- 5010 to 4-~ E. -C20- 4-S. C cr D -c C: 4-

Ift & a a-Ia 8 10 20 30 4( I--erzUJ_ ' I --I5060'30'90._ A Lenqth fr.-p& Width~f0 cranial30.40 ta~ble FIG. 3. Logarithmic graphs illustrating growth changes for the indicated skull characters in Calypto- cephaklila gayi. The crosses represent determinations for the skull of C. canqueli. Regression lines determined by method of least squares. 1949 SCHAEFFER: ANURANS FROM TERTIARY OF PATAGONIA 53 -.16 .14 -.12 -JO -.08 -.06 -04 -.02 O .02 .04 .06 D8 .10 .12 .14 .16 r f * * ii * -- .1 | l 1 i I LENGTH NA.+FR.-PA. WIDTH CRANIAL TABLE LENGTH NASAL WIDTH NA.ORBIT TO NARES WIDTH MX.AT ORBIT LENGTH ORBIT WIDTH ORBIT * . LENGTH TEMP. FENESTRA

LENGTH CORACOID LENGTH HUMERUS x LENGTH RADIUS LENGTH METACARP 11

II 3d~~I OIX ~~~~~~~v

LENGTH TI B.- F1. LENGTH RT. ASTRAGALUS FIG. 4. Ratio diagram comparing various measurements of Calyptocephalella canqueli (dotted line) and Calyptocephaklla gayi (solid line). as in C. gayi. The temporal fenestra appears C. cangueli is made, it is found that the to be somewhat shorter than in the Recent width along the posterior border of the table species (in relation to the length of the fronto- is about the same as in C. gayi in relation to parietal) in spite of some crushing in this the length of the nasal plus the frontoparietal area (fig. 3C). (fig. 3D) as measured along the middorsal The cranial table is gently convex in C. line. gayi. The difference in the width of the table A portion of the right vomer is preserved, when arched and when flattened is about 2 but it is badly crushed and reveals no im- mm. After such an allowance for crushing in portant characters. The palatines appear 54 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 93 identical in shape and extent with those of subadult skull characters such as orbit shape C. gayi, although they are likewise obscured and a slightly shorter temporal fenestra, by crushing. The sphenethmoid is completely in addition to possessing features that appear ossified. The posterior portion of the para- to be unaffected by relative growth changes sphenoid, and the otic and occipital regions such as the wider nasal bridge. The fore are obscured by the remains of the pectoral limb elements are consistently longer in C. girdle and left fore limb. gayi than in canqueli. The mandible is edentulous. The coronoid Although the fossil remains just described process of the angulosplenial is larger and have been compared with other recent lep- more clearly defined than in C. gayi, but todactylids, the characteristic skull orna- otherwise the similarity with the Recent mentation and the secondary deposit of bone species is again very close. on the skull immediately suggest close affinity The vertebrae are clearly procoelous but with Calyptocepialella, and this appears to be are otherwise too poorly preserved for de- confirmed on the basis of the available fossil tailed description or comparison. The first specimens. The difficulty with which a num- two presacral vertebrae in the type have ber of the Tertiary anurans are separated been compressed between the skull and the from Recent species on an osteological basis third presacral. The transverse processes of suggests that many species may actually the third are robust and of about the same extend back well into the Tertiary. The very relative length as in C. gayi. close resemblance between C. gayi and C. The left coracoid and the right scapula canqueli would appear further to support are well preserved in the type specimen and this thesis. again are not distinct from these elements in C. gayi. The clavicles are badly broken MEASUREMENTS IN MILLIMETERS and comparison is not possible. The humerus Calypto- Calypto- and radio-ulnar are of similar relative length cephalella cephaleUa in both species. The metacarpals are some- gayi canqueli what shorter and more delicate in C. can- M.C.Z. A.M.N.H. queli. The distal phalanges agree in having No. 447 No. 3429 slightly bulbous tips. Length of nasal +fronto- The pelvic girdle parietal 34.2 35.0 and hind limbs are pre- Width of cranial table 44.9 51.0 served in a referred specimen (A.M.N.H. Length of nasal 12.3 13.0 No. 3427). Although the skull is missing, the Length of frontoparietal 22.0 23.0 specimen may be confidently assigned to C. Width of nasal, nares to canqueJi. The sacral diapophyses are moder- orbit 3.7 5.0 ately dilated as in C. gayi. The coccyx has Length of orbit 12.7 13.0 two condyles and a strongly developed dorsal Maximum width of orbit 11.0 9.5 ridge. The hind limbs offer no additional Length of temporal fenestra 19.1 16.0 differentiating characters. Length of coracoid 14.0 14.7 The logarithmic regressions tend to dem- Length of humerus 31.0 26.8 onstrate the close similarity of C. canqueli Length of radio-ulna 20.8 17.7 and C. gayi, while the ratio diagram (fig. 4) Length of tibiofibula 41.8 40.2 emphasizes such differences as do exist. Length of right astragalus 21.5 20.5 Although the differences may appear hardly REFERRED SPECIMENS worthy of specific separation, they are con- From same horizon and locality as the sistent within the limits of the available type adult material, both and specimen: fossil Recent. The A.M.N.H. No. 3400, dissociated lower jaw and ontogenic series of C. gaYi skulls shows a limb fragments number of interesting changes in proportion A.M.N.H. No. 3427, partial skeleton showing with increase in size, the most obvious being coccyx, pelvis, and hind limbs the relative increase in orbit width and the length of the temporal fenestra. Apparently, From upper portion of Sarmientan Group, adults of C. canqudi retained some of Colhuehuapian Stage, Upper Oligocene. these South end of Lago Coli-Huapi, 15 kilometers rLIN AMER. Mus. NAT. HIST. VOL 93, PLATE 16

Pp. --E--I.., t, _' ophractus casamayorensis, new genus and species. 1. A.M.N.H. No. 3165, type specimen. Doral aspect of rnplete left nasal bone on left, lateral aspect of incomplete left maxillary on right. 2. A.M.N.H. No. 3164. An- )r aspect of presacral vertebra on left, median aspect of incomplete right maxillAry on right. X 3/1 BuurN AMum. Mus. NAT. HIST. VoL. 93, PLATIM 17

c1yp au. c. , new speei A.M.N.H. No. 3429, type qei. 1. Dorsal view. 2. Ventral view. X i/i BuUXnN Aiu. Mus. NAT. HIsT. VOL. 93i PLA7Z 18

:4I.

ti'

:1%atx

Iirx X* I 9 ... '. .'k. t

.. 11, " BuuxrmAmU MusNAT. Hs. VOL. 93, FA 19

- - . - .d f

4 1. .;e:.; , N.' . ;* s- /

.. 4.. r vi; L '. .*." 4 f '4 1949 SCHAEFFER: ANURANS FROM TERTIARY OF PATAGONIA 55 southwest of Casa Ramos, central Chubut, Eupsophus. The greatly elongated hind limbs, Argentina. These fragments may be de- particularly the feet, appear very similar in finitely assigned to Calyptocephalella and, so their proportions to the limb segments of far as can be determined, show no specific Hylorhina. Actually, the differences in the differences from C. canqueli in spite of the relative lengths of the limb elements in time interval involved. Hylorhina silvatica, Eupsophus grayi, and A.M.N.H. No. 3236, fragments of maxillary the form here described are hardly great A.M.N.H. No. 3237, fragments of maxillary enough to be of much taxonomic significance. A.M.N.H. No. 3238, isolated sphenethmoid The total length of the hind limb compared A.M.N.H. No. 3239, isolated sphenethmoid and with the length of the axial skeleton in H. fragment of right ilium silvatica is about 1.9, while in both E. grayi A.M.N.H. No. 3241, maxillary fragments, por- and the form here described it is no more than tions of skull roof, and two presacral procoel- 1.8. Ratios of the hind limb segments on the ous vertebrae basis of a femoral length of two units indicate FAMILY LEPTODACTYLIDAE the following (figures represent: femur, EUPSOPBUS FITZINGER tibiofibula, astragalocalcaneal segment, fourth digit): H. silvatica, 2, 2.3, 1.3, 2.5; Cystignathus DUMPMUL AND BIBRON (in part), 2.2; E. sp., 2, 2.1, 1.0, 2.3. 1841, Erp6tologie g6n6ral, vol. 8, p. 392. E. grayi, 2, 2.2, 1.2, Borborocoetes BELL, 1843, The zoology of the The nasals are well ossified and are of the voyage of H.M.S. Beagle, pt. 5, reptiles, p. 34. same relative size as in E. grayi. In contrast Eupsophus FITZINGER, 1843, Systema reptilium, to the condition in this species, however, p. 31. (For discussion of generic synonomy, see these elements are in contact, which may or Parker, 1932, Ann. Mag. Nat. Hist., ser. 10, vol. may not be a consequence of crushing. The 10, p. 342.) frontoparietals are separated for most of their length, but the details of their extent GENOTYPE: Cystignatzus roseus Dum6ril and of the size of the fontanelle cannot be and Bibron. determined. One skull preserved in ventral Eupsophus sp. aspect (A.M.N.H. No. 3407) demonstrates HORIZON AND LOCALITY: Sarmientan that the portion of the sphenethmoid anterior Group, Deseadan Stage, Lower Oligocene. to the prevomers is, ossified as in E. grayi. Scarritt Pocket, southwest portion of Rin- The presence of prevomers is indicated, but conada de los Lopez, west side of Sierra no details are preserved. Canquel, central Chubut, Argentina. DESCRIPTION: A small leptodactylid not separable from Eupsophus on the basis of available specimens. The maxillaries are shallow, with teeth. The nasals are relatively large and possibly in contact. A frontoparietal fontanelle is present. The vertebrae are procoelous, the sacral articulation is double in some specimens, the sacral diapophyses are moderately expanded, and there is a double condyle on the coccyx. The hind limbs are elongated as in Recent species of this genus. The observable characters of the skull and postcranial skeleton indicate relationship with Eupsophus or Hylorhina. The osteological characters separating these genera, such FIG. 5. Eupsophus sp. A.M.N.H. No. 3422. Par- as the presence of a bony sternum in Hylo- tial skeleton preserved in dorsal aspect. Abbrevia- rhina, cannot be distinguished in any of the tions: frpr, frontoparietal; hu, humerus; mand, mandible; mx, maxillary; na, nasal; spheth, fossil specimens. A comparison of the skulls, X2/1. however, does suggest closer affinity with sphenethmoid; sq, squamosal. 56 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 93 MEASUREMENTS IN MILLIMETERS Eupsophus Eupsophus sp. grays A.M.N.H. A.M.N.H. A.M.N.H. A.M.N.H. A.M.N.H. No. 22104 No. 3407 No. 3424 No. 3426 No. 3430

Length of skull . 12.0 9.60 Width of skull ...... 14.3 - - - Length of vertebral column (including uro- style) .23.5 17.5 Length of humerus .10.5 8.0 - 9.5 - Length of radio-ulna .7.5 - - 6.5 Length of femur .17.5 13.3 - 12.2 13.0 Length of tibiofibula .19.0 14.0 14.0 14.0 14.3 Length of astragalocalcaneal segment . . 10.5 6.0 6.5 6.0 7.0 Length of fourth digit .19.0 15.4 15.2 16.5 Total length of pes .28.7 - 21.4 - 23.5

There are nine procoelous vertebrae which A.M.N.H. No. 3415, pelvis and hind appendages might be distinguished from those of E. of small individual grayi only by a somewhat stronger develop- A.M.N.H. No. 3420, skull fragments and hind ment of the neural spines, actually merely appendages longitudinal ridges, on the fifth through A.M.N.H. No. 3422, crushed skull, left forearm, seventh vertebrae. It is reasonable to sup- and vertebral column pose, however, that the size of these ridges is subject to at least this much variation. Two specimens (A.M.N.H. Nos. 3422 and 3426) appear to have a double sacral articulation. The diapophyses of the ninth vertebra are moderately expanded, while those of the eighth, which are in contact with the ends of the ilia, are of uniform width. This condition may be associated with a relatively shorter coccyx than in E. grayi. A double articulation between the sacral vertebra and the coccyx is evident in one specimen (A.M.N.H. No. 3407). The elements of the pectoral girdle cannot be observed in any of the specimens. The fore limbs and, as pointed out above, the hind limbs are similar in their propo-rtions to those of the Recent species with the excep- tions noted. Assignment to Eupsophus is indicated by all the observable characters. The poor preservation precludes detailed description and the determination of specific differences, if any are actually present, between this form and E. grayi. REFERRED SPECIMENS FIG. 6. Tentative restoration of Eupsophus sp. based on all available specimens to show propor- A.M.N.H. No. 3407, complete skeleton, skull tions of axial skeleton and limb segments. Approx- badly crushed imately X3/2. 1949 SCHAEFFER: ANURANS FROM TERTIARY OF PATAGONIA 57 A.M.N.H. No. 3424, hind appendages Group, Deseadan Stage, Lower Oligocene. A.M.N.HI. No. 3425, anterior portion of skull Scarritt Pocket, southwest portion of Rin- A.M.N.HI. No. 3426, postcranial skeleton conada de los Lopez, west side of Sierra A.M.N.H. No. 3430, two partly superimposed Canquel, central Chubut, Argentina. and crushed skeletons SPCIuFIC DIAGNOSIS: Same as for genus. FIAMILY LEPTODACTYLIDABE DESCRIPTION: In a number of respects this is the most interesting form in the collection, NEOPROCOELA, NrW GENiuS and it is unfortunate that it is represented GF.NOTYPE: Neoprocoela edentatuts, new only by the type specimen. The large fon- species. tanelle, edexntulous upper jaw, wide maxillary, GENERIC DIAGNOSIS: A presumed lepto- ancl dilated sacral diapophyses suggest rela- dactylid with several prinmitive, criniid-like tionship with the Australian leptodactylidls. pmx

A B

Fi';. 7. Neoprocoela edentatus, new genus and species. A.M.N.H. No. 3428, type specimen. A. Dorsal view. W1. Ventral view. Ab- breviations: cI, clavicle; cor, coracoid; frpr, frontoparietal; mand, mandible; mx, miiaxillary; na, nasal; pal, palatine; pmx, premaxil- lary; pro, prootic; ps, parasphenoid; pt, pterygoid; sc, scapula; spheth, sphenethmoid; sq, squamosal; vo?, vomer. X1/1. features. Premaxillaries T-shaped, edentu- All the Australian genera but two, however, lOus. Maxillaries broad, edentulous, with long have an incompletely fused vertebral con- pterygoi(l articulation. Nasals completely dyle (Parker, 1940), and this is the condition ossified andI in contact. Fontanelle present, in the Eocene Indobatrachus (Noble, 1930). with fronitoparietals widely separated. Sphen- In Neoprocoela there is complete fusion of the ethmoidl ossified and entire. Vertebrae pro- vertebral condyles. It may be concludedl on coelous, intervertebral condyle fused with the basis of available evidence that Neo- centruni; sacral diapophyses dilated, trian- procoeZa is a true leptodactylid retairning gular in shape; coccyx probably not fused several primitive leptodactylid features. The to sacrumn. Shoulder girdle as in Eupsophus. toothless condition was probably an indle- pendent development, as must be the case Neoprocoela edentatus, new species with many of the Australian genera and for TyPit: A.M.N.H. No. 3428. Partial skele- that matter with the Neotropical Batra- ton consisting of skuill, vertebral column, and chophrynus. proximal portion of left forearm. Although many of the characters of Neo- HORIZON AND LOCALITY: Sarmientan procoela suggest relationship with otler 58 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 93 families of procoelous anurans, these groups tered elements of the left manus were closely appear to be ruled out on the basis of other associated with the posterior portion of the more specific or diagnostic characters. There parasphenoid. While these bones and the is, for instance, a marked resemblance in surrounding matrix were being removed, in skull structure to that of Bufo, the principal order to expose fully the palatal area, a single difference being the presence of the fontanelle terminal phalanx was uncovered. This ele- in Neoprocoela. To consider this Oligocene ment shows no appreciable dilation at its form as a specialized Bufo, however, would distal end. require the presence of the Bufonidae (sensu The procoelous vertebrae, nine including sricto) in South America by no later than the the sacral, have cylindrical centra that are early Oligocene, an occurrence which is not longer than wide. The dilated, triangular supported by the known paleontological facts. sacral diapophyses present a difficult prob- The general nature of the skull thus sup- lem, as they are characteristic of Indobatra- ports affinity with the more generalized chus and the Australian leptodactylids as South American leptodactylids such as well as the Hylidae, Brachycephalidae, and Telmatobius or its edentulous derivative, Brevicipitidae (Microhylinae) but not of the Batrachophrynus. The nasals are in contact in Recent South American leptodactylids. In the latter, the maxillary is broad, the articu- spite of this fact, the weight of the observable lation between the maxillary and pterygoid is characters, as pointed out above, indicates extended, and a fontanelle is present, al- that Neoprocoela is a true but primitive though it is smaller than in Telmatobius. The leptodactylid with the edentulous upper jaw fontanelles of Neoprocoela and of Telmatobius as its principal specialized divergence. are of about equal size, separating the fronto- The absence of the centrum of the sacral parietals for most of their length. vertebra and the poor preservation of the The fragmentary prevomers show no indi- coccyx make it impossible to determine the cation of teeth, a condition commonly cor- nature of the sacrococcygeal articulation. related with an edentulous upper jaw. The The dilation of the end of the other palatal structures show no unusual proximal characters. coccyx suggests that a double condyle was The sphenethmoid is well ossified, present, but this is not certain. Furthermore, the bony portion ending at the level of the fusion between the sacral vertebrae and the prevomers. The squamosals are small with cannot be ruled out. delicate anterior processes of unknown length. coccyx definitely As will be pointed out in the discussion, there are a The left shoulder girdle is completely pre- number of served, consisting of a resemblances to the Atelopodidae narrow, rod-like which cannot at present be properly eval- clavicle, somewhat wider coracoid, and uated. typical leptodactylid scapula. The clavicle MEASUREMENTS: The fragmentary nature and coracoid are preserved in a divergent of the position, and the left clavicle, which is also single known specimen makes it im- present, shows evidence of being slightly possible to take accurate or meaningful arched. This suggests the arciferal pectoral measurements. girdle as pointed out by Noble (1930). Scat- HYPODIGM: Type specimen only. DISCUSSION ANURAN DIFFERENTIATION A GENERAL TREATMENT of the origin of the coelous, and the coccyx has a double condyle. Neotropical anuran fauna including consider- Romer (1945) has tentatively placed this ation of the more recently discovered fossil genus, together with Montsechobatrachus evidence has not been attempted since the (Vidal, 1902) from the Upper Jurassic of pioneer studies of Noble (1925 and 1931) and Spain and Stremmia of the same age from Dunn (1931). Not being an anuran specialist, East Africa, in the family Montsechoba- the writer embarks on such an effort with trachidae as a subdivision of the Amphi- even more hesitation than did Dunn, but in coela. As these genera are known only from the hope that the recently accumulated fragmentary remains and impressions, the paleontological data will throw some addi- family is very possibly not a natural one. tional light on the problem. They do demonstrate the presence of defini- There is still no general agreement among tive anurans at this time and testify to a herpetologists regarding the phylogeny and world-wide distribution in the Upper Juras- composition of many of the families of Salien- sic. tia. The list included in table 1 is based for The other amphicoelous family, the Lio- the most part on the most recent opinions of pelmidae, includes two Recent genera with a Dav'is, Dunn, Noble, and Parker for the well-known disjunctive distribution. Lio- modern families and of Watson, Piveteau, pelma occurs in New Zealand and Ascaphus and Romer for the extinct ones. In order to in the northwestern United States. As give the Neotropical fauna its proper setting Noble and others have pointed out, this and to present recently discovered fossil data distribution also suggests a former world- supporting a Holarctic origin for the stocks wide or at least Holarctic distribution of a from which the Neotropical fauna was de- generalized, primitive amphicoelous stock. rived, the probable origin and dispersal of all OPISTHOCOELA: This suborder, which is the families listed in table 1 will be briefly characterized by possessing opisthocoelous discussed. vertebrae and free ribs in the larval or adult The recent work of Watson (1941) has stage, has a thigh musculature which, ac- rather clearly demonstrated that the Salientia cording to Noble, closely resembles that of evolved from the labyrinthodonts. In the the Amphicoela. Of the two families included Pennsylvanian genera Amphibamus and Mio- in the Opisthocoela, the Discoglossidae and batrachus, which are grouped in the family the Pipidae, the former is presumably the Amphibamidae, there are already changes in more primitive with free ribs in the adult. the skull and vertebrae in an anuran direc- The discoglossids are first known from the tion. In the Triassic Protobatrachus, the only Upper Oligocene of France with the descrip- known member of the Protobatrachidae, this tion of Prodiscoglossus by Friant (1944). It trend is carried still further with a more appears to combine certain characters of anuran-like skull and an elongated ilium Bombina, Discoglossus, and Alytes. The mod- (Piveteau, 1939). With the Amphibamidae ern genera Discoglossus and Alytes occur also known only from one locality in North Amer- in the Lower Miocene of France and the ica and the Protobatrachidae only from Middle Miocene of Germany. Two extinct Madagascar, there is as yet no clue as to a genera, Pelophilus and Latonia, have been center of origin for the Salientia. This evi- found in the Upper Miocene of Germany. dence does indicate, however, that the group The discoglossids thus give every appearance must have reached its distinctive status by of being primarily an Eurasiatic group that no later than the early Jurassic. was differentiated in the early Tertiary AMPHICOELA: The Amphicoela were pos- about as completely as it is today. During sibly a distinct group by Upper Jurassic the Tertiary Bombina spread from some un- time. The single known vertebra of Eobatra- known center of origin until it covered most chus (Moodie, 1914) appears to be amphi- of Eurasia. Discoglossus reached Africa but 59 60 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 93

g Ci C) 0 I I I I i I I z I U5 1,i4P 0 t%4

CU ;2 A :E,

be Z~~~~~~~~~ c il 0c t~_X CKIH ____ -o

1-4 Ud

:4 0

1-4 U) I- 4U)w I I I I I I I I III I II

I -_; I L3 *-0 0 I L41 .o4 .- =*I 5w 0 C5~~~~~4 4))

C.)0 Ca ~~~~ce _) >M 0 '0 Cd ._ . 4) 4dVW() 0. cd .0 C) 4)4) I..4 .0 CD la-o U4Eco 0 (t J4)0 .4i 4-4 0:0 0 d-. 4.,I la - 4 4 ) -noc 4) C.) =ou .o co >o E 2EE X .0 0 4)0~ 4)4) co 4). ) U 0 :.U Cs4

z.ce ceI to 0 _0 3 Ca 0 u)0 I-- 314 4) Od 4.4 E0 0 0d C.) '3.4. E 0. O I 0 P14 1949 SCHAEFFER: ANURANS FROM TERTIARY OF PATAGONIA 61 possibly not Asia. Alytes presumably dif- Eopelobates only lacks this increase in size. ferentiated in Europe and remained there. Scaphiopus and the related extinct genus A single genus, Barbourula, unknown in the Neoscaphiopus have been described by Tay- fossil state, reached or was differentiated in lor (1942) from the Upper Pliocene of Kansas. the Philippines. This primitive anuran family, The Pelodytidae are characterized, follow- on the basis of both fossil and Recent evi- ing Taylor, by an unornamented skull, no dence, had a relatively restricted and slow palatine bone, prehallux not forming a spade, dispersal. coccyx articulating with the sacral vertebra The Pipidae were already distinct from the by a double condyle, astragalus and calca- discoglossid stock by the early Tertiary. The neum more or less fused throughout their problematical Eoxenopoides (Haughton, 1931) length. The family may be represented in the from Lower Tertiary clays in Namaqualand, Middle Eocene of Germany by Propelodytes South Africa, appears to be a member of this (Weitzel, 1938), a form with procoelous verte- family. Its systematic position is not definite, brae and sacrococcygeal joint not fused. The however, mainly because of the surprising astragalus and calcaneum are fused, however, variation in the nature of the sacral articula- only at their proximal and distal ends, which tion. Xenopus stromeri (Ahl, 1926) has been may or may not represent an incipient stage described from the Miocene of southwest towards more complete fusion. A form named Africa. Although the differences between the Miopelodytes which is very similar to the South American and African pipids are not European Pelodytes has been described by great (E. R. Dunn, personal communication), Taylor (1941) from the Middle Miocene of those that do exist appear to be consistent. Nevada. It has the astragalus and calcaneum The African occurrences strongly indicate a fused throughout their length. late Mesozoic dispersal from some unknown Both families of the Anomocoela clearly Holarctic center, with a Protopipa-likeform had a Holarctic distribution with representa- reaching South America from the north prior tives reaching North America no later than to the break in the Panamanian bridge. One the middle of the Tertiary. There is no evi- other tempting hypothesis, for which there dence, Recent or fossil, that the Pelobatidae is no conv'incing morphological evidence, sug- or the Pelodytidae entered South America gests an independent origin of the African or Africa. and South American pipids from a primitive PROCOELA: Theearly Eocene Intertrappean Eurasian discoglossid stock. beds of India have yielded one of the oldest ANOMOCOELA: The Anomocoela, according representatives of the Procoela in the form to Noble (1924), evolved from the primitive of a leptodactylid bearing many characters Discoglossidae by a fusion of the interverte- in common with the Recent Australian bral cartilages resulting in a procoelous con- Leptodactylidae. Indobatrachus (Noble, 1930) dition, a reduction in the number of presacral may possibly belong to the subfamily Myo- vertebrae to eight, loss of ribs, appearance of batrachinae (Parker, 1940). This occurrence an acromion process, and expansion of the indicates a southern Holarctic origin for the sacral diapophyses. Although this suborder Procoela, a conclusion not opposed by the was formerly considered to include but one presence, in the Miocene of Europe, of two family, the Pelobatidae, Taylor (1941) has procoelous genera, Palaeobatrachus and ?Pro- recommended the recognition of an additional topelobates, which are placed in the separate family, the Pelodytidae (originally employed family Palaeobatrachidae. The relationships by Cope in a less restricted sense). of this family, which may very well be an The oldest known member of the Pelo- unnatural assemblage, are not clear (see batidae, Macropelobates, occurs in the late Wolterstorff, 1929). Oligocene of Mongolia (Noble, 1927). Eopelo- Most herpetologists have not followed bates, from the Miocene of Germany, is Noble's practice of lumping the Leptodacty- considered by Parker (1929) to represent a lidae with the Bufonidae. The two families structurally earlier stage than the Oligocene are practically indistinguishable from the form. Macropelobates is apparently a typical standpoint of osteology and myology, par- pelobatid with an enlarged prehallux, while ticularly since toothless leptodactylids have 62 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 93 been described. Davis (1936) has pointed however, that the East Indian genus Caco- out, however, that Bidder's organ is present phryne belongs to the Atelopodidae, indicat- in the Bufonidae but absent in the Lepto- ing that this family must have formerly been dactylidae, and he presents a logical thesis rather widely distributed in the Holarctic for believing that this structure first ap- area. The time of migration into South Amer- peared in the Bufonidae. He concludes on this ica is difficult to conjecture. It might be and other evidence that the Leptodactylidae considered as having occurred in the late and Bufonidae are natural groups with a Tertiary except for certain disturbing resemn- common origin, a concept that is certainly blances to Neoprocoela such as the dilated not in opposition to the paleontological sacral diapophyses and edentulous maxillary. record. Their separation may have taken The presence or absence of a firmisternal place some time in the Cretaceous. pectoral girdle in fossil Anura cannot be The Leptodactylidae must have had a definitely determined in most cases. The wide Holarctic distribution in the late Meso- single specimen of Neoprocoela lacks the zoic and probably extended into Australia sacrococcygeal articulation, and it is possible prior to its isolation in the Upper Cretaceous. that fusion did actually exist. The fontanelle Although the systematic position of several in this species may be a primitive leptodactyl- of the forms described in this paper is not id character which could have been lost in definite, the evidence clearly indicates that certain of its descendants. If Neoprocoeka is, representatives of this family entered South in fact, a true atelopodid, this family must America before that continent was isolated have then entered South America in the at the beginning of the Tertiary. Dispersal Paleocene. into Africa presumably occurred sometime On the basis of available evidence, it is during or before the Tertiary, although the reasonable to conclude that the Hylidae fossil evidence is still lacking for this assump- evolved from some primitive leptodactylid- tion. like stock in the late Mesozoic. The two The Bufonidae (Bufo) are first known from hylid subfamilies recognized by Noble, the the Oligocene of Europe (Filhol, 1877), and Hemiphractinae and the Hylinae, are pos- remains of Bufo have also been described not and from sibly valid, fossil remains cannot be the Upper Pliocene of Kansas (Taylor, assigned to either subfamily on the basis of 1937). The center of dispersal must have been osteology unless reference to a Recent genus somewhere in Eurasia, with Bufo only reach- is possible. One of the most unexpected dis- ing North America during the Tertiary and coveries in the paleontology of the Salientia entering the Neotropical region either by a is the presence of remains assigned to the sweepstakes route before the Middle Plio- genus Amphignathodon in the late Oligocene cene or by the Panamanian bridge after that and early Miocene of France (Piveteau, time. The failure of the bufonids to reach This member of Australia suggests a later radiation than oc- 1927). specialized the curred in the Hemiphractinae, the only definitive anuran case of the leptodactylids. The with mandibular teeth, is confined at the bufonids occurring in Africa and the Indo- time to Ecuador. The Malayan region present no particular dis- present other members persal of this subfamily, which are all Neotropical problem. and have no known The Brachycephalidae were considered by fossil representatives, Noble (1931) to be a composite family, are in some respects less specialized, in others and almost equally so. They all appear to be Davis (1935) divided the group into three derived families, the Rhinodermatidae, Dendrobati- from a Hyla-like ancestral stock dae, and Atelopodidae. The Rhinodermatidae which undoubtedly had a Holarctic dispersal. and Dendrobatidae are probably endemic to Furthermore, the possible occurrence in South America, being derived from lepto- Europe of Amphignathodon by the late dactylid types close to Syrrhophus and Cros- Oligocene and of Hyla by the Lower Miocene sodactylus, respectively, during the Tertiary (Noble, 1928) indicates that possibly both (Noble, 1931). Davis has subfamilies of the Hylidae were well differ- demonstrated, entiated in the Northern Hemisphere. The 1949 SCHAEFFER: ANURANS FROM TERTIARY OF PATAGONIA 63 Hemiphractinae presumably entered North pendently descended from the Ranidae (i.e., America in the Tertiary and South America Hemisinae and Cacosterninae). either by a sweepstakes route during the Parker considers the Microhylidae to be isolation period or after the bridge was re- an ancient offshoot from a primitive firmi- established. Tree frogs might be quite readily sternal stock and the Phrynomeridae to be transported on floating logs and other vege- derived from the same general line. He be- tation. The presence of Amphignathodon in lieves that this same stock at a later date Europe presents a problem regarding the gave rise to the Ranidae in the Holarctic place of origin of the other hemiphractid realm and the Rhacophoridae (= Polypedati- genera now confined to South America. This dae) in the Ethiopian region.This subdivision genus is clearly not ancestral to the other must have been well under way by the begin- hemiphractids, and it may be reasonably ning of the Tertiary. The Microhylidae are concluded that at least one other type, pos- unknown as fossils. They apparently had sibly resembling Gastrotheca, entered from the an early Tertiary radiation in Asia extending north which subsequently gave rise to Gastro- into Africa and North America. Only one theca itself, as well as Hemiphractus and subfamily, the Microhylinae, reached South Cryptobatrachus. Any of these genera may America, probably relatively late in the Ter- also have occurred in the Northern Hemi- tiary. The Phrynomeridae, containing the sphere. single African genus Phrynomerus, must have The great diversification of the Hylinae evolved from the microhylid stock early in in South America might suggest the Neo- the Tertiary if its degree of specialization be tropical area as their place of origin. The considered as an index of antiquity. occurrence of Hyla in the European Tertiary, The Ranidae, like the Microhylidae, al- the relic survival pattern of the entire group, most certainly became differentiated before and the parallelism and endemism of the the beginning of the Tertiary. The nearly African and Oriental species indicate a former universal distribution of Rana, including wide dispersal in the Holarctic. Migration Australia, and the great diversification of the into South America of some Hyla-like types, family in Africa and South Asia suggest a but not necessarily Hyla itself, must have possibly late Cretaceous dispersal of the occurred in the early Tertiary. Hyla, like Ranidae over much of Eurasia with migration Bufo, might have reached the Neotropical into Africa at least by the early Tertiary. area in the late Tertiary, although there is no Rana occurs in the Upper Miocene and later real evidence opposing an early Tertiary formations of Europe and in the Tertiary of arrival.- Much additional fossil evidence is Mongolia and China (Schlosser, 1924; Young, needed before the complicated picture of the 1936). Ranavus (Portis, 1885) and Aspherion radiation and dispersal of the Hylidae is (Von Meyer, 1847, p. 192), two genera rather well understood. similar to Rana, are found also in the Miocene DIPLASIOCOELA: The Diplasiocoela were of Europe. Rana reached North America by probably derived sometime in the late Meso- no later than the Upper Pliocene (Taylor, zoic from some primitive group of the Pro- 1942) and presumably entered South America coela. The taxonomic history of this suborder over the Panamanian bridge. The Racho- is complicated but has been greatly clarified phoridae (= Polypedatidae), derived from by Parker (1934). It appears that four famil- the ranid stock, presumably remained much ies must now be recognized: the Microhy- closer to their center of radiation, the limits lidae, the Phrynomeridae, the Ranidae, and of their range being Japan, southern Asia, the Phacophoridae. The Brevicipitidae, as the East Indies, and into Africa and Mada- defined by Noble, is thus considered to be gascar. The occurrence on Madagascar indian unnatural assemblage with many of its cates a long existence in Africa, with arrival subfamilies now included by Parker in the on this island either over a land bridge at the -Microhylidae, while others are considered to beginning of the Tertiary or by a sweep- be of uncertain position but probably inde- stakes route sometime later. 64 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 93 ANURAN DISPERSAL

On the basis of available paleontological action is even more striking in the case of evidence it is very tempting to consider single genera that migrated into North Amer- southern Asia as the primary center of dis- ica from Asia and finally into South America persal for the modern anuran families that towards the end of the Tertiary. Bufo and can be traced to the beginning of the Ter- Rana are the only members of their respective tiary.' The role that Africa has played, how- families to reach the New World, while the ever, is difficult to evaluate. Parker (1929) appearance of Hyla may have represented a believes that the Rhacophoridae (= Poly- second migration wave within this family. pedatidae) are of African origin, and there is Taylor (1942) has erected the genus Anchy- some support for such a conclusion on the lorana for some Upper Pliocene remains from basis of the distribution of modern genera. Kansas, which are apparently close to Rana Paleontological evidence is, however, com- but possibly distinguished from it by the pletely lacking. In any case, that continent fusion of the eighth and ninth vertebrae. If must have had a rather diversified anuran such fusion is not within the range of varia- fauna very early in the Tertiary or more tion of the vertebral column of Rana, al- probably in the late Mesozoic. though the evidence cited by Taylor indi- The factors affecting the extent and rate of cates it may be, the presence of a genus allied dispersal of the Salientiaarelargely unknown, to Rana in North America suggests some particularly in regard to migration across an Tertiary differentiation of the Rana stock entire continent, or from one continent to which subsequently became extinct. another across a land bridge. The discoglos- The paleogeographic factors affecting the sids apparently expanded their range very introduction of a terrestrial fauna into South slowly across Eurasia during the Tertiary America have been considered by Simp- but for some reason never reached North son (1943). Adopting his categories, which America. As Simpson (1940) has pointed out were devised for Neotropical land mam- in the case of certain mammalian groups, mals but which the should apply with equal age and area theory requires that the validity to the Salientia, the following may be Discoglossidae be a family of relatively recent recognized: origin. This is obviously not the case. The A. Anurans that entered South America Pipidae, which are probably not so ancient as prior to its isolation in the early Tertiary and the Discoglossidae and may have evolved evolved there this isolation: from the discoglossid stock, had a rapid during Pipidae, expansion Leptodactylidae, ?Atelopodidae, Hylidae which carried them into Africa (possibly but not necessarily excluding and South America at the beginning of the Hyla). Tertiary. The Leptodactylidae, and possibly B. Those that the Atelopodidae and Hylidae, arrived during the period also fall into of isolation by some sweepstakes route: this category, with an early Tertiary or even ?Bufonidae (Bufo only). a late Cretaceous dispersal. C. Those that entered Although both the Pelobatidae and the South America Pelodytidae entered after the connection with North America was North America during reestablished at the end the Tertiary, there is as yet no evidence that of the Pliocene: Hylidae (Hyla only, but possibly an early they reached South America. A filter-bridge Tertiary arrival), Microhylidae (Microhy- linae), Ranidae (Rana only). ' The important study by P. J. Darlington, Jr., on D. Those that were "The differentiated in geographical distribution of cold-blooded verte- South America, but were derived brates" (Quart. Rev. Biol., 1948, vol. 23, no. 1, pp. originally 1-26; no. 2, pp. 105-123) appeared too late for proper from forms that arrived prior to the isolation: consideration here. In regard to the existing anuran Rhinodermatidae, Dendrobatidae. families, the author concludes that they probably The accumulated evidence, fossil and Re- orginated in the Old World tropics and have dispersed cent, indicates that the distributional by northern routes, a thesis that is not at variance with pattern the conclusions reached in this paper. of the Salientia can be adequately explained on the basis of dispersal involving only the 1949 SCHAEFFER: ANURANS FROM TERTIARY OF PATAGONIA 65 Bering and Panamanian land bridges. As far migrated between Australia and South as the Neotropical anuran fauna is concerned, America via Antarctica or that the pipids and it appears manifestly unnecessary to agree (?) dendrobatids employed a South Atlantic with Joleaud (1939) that the "cystignathids" bridge.

TRENDS IN ANURAN EVOLUTION With the possible exception of the Amphi- a single variant was confined to a particular coela, which may have been differentiated level, but rather that all or most of the as early as the Upper Jurassic, the modern stages existed simultaneously throughout the anuran suborders and, in fact, most of the time interval represented by the deposit. The modern families appear abruptly in the stages can be arranged in a sequence illustrat- early Tertiary. This circumstance, together ing a progressive forward shift in the sacral with the presence of definitive anurans in the articulation from one in which the diapophy- Upper Jurassic, implies that the major salien- ses of the first four coccygeal vertebrae tian radiation occurred during the Creta- (typically fused but with well-developed ceous. As this period had a duration of transverse processes) are involved to one in roughly 65 million years, it is not pos- which only the diapophyses of the last sible to determine the tempo and mode of presacral vertebra articulate with the ilia. this radiation, particularly since Cretaceous There is still another type in which a dia- anurans are completely unknown. It is evi- pophysis of the first coccygeal vertebra and dent, however, that the rate of evolution the opposite one of the last presacral make during the Tertiary, representing an almost up the sacral articulation. All these variations equivalent time interval, was relatively very must have permitted essentially normal loco- much slower. motion, as adequate support in the sacral The amphibamids and Protobatrachus in- region is indicated in every type. Ateolepus dicate, as Romer (1945) has pointed out, varius exhibits a similar but less extreme that the skull evolved more rapidly in an variation (Noble, 1931). The six species of anuran direction than the postcranial skele- Rana (Taylor, 1942) in the Upper Pliocene ton. By the end of the Jurassic, however, the Rexroad fauna of Kansas, described on the entire skeleton must have approached the basis of slight differences in the morphology biomechanical limit of salientian specializa- of the sacral vertebra, may actually represent tion. The subsequent modifications in the variants of a much smaller number of species. skull, vertebral column, pectoral girdle, and At the supraspecific level, the variation relative length of the limb segments were in characters usually considered to be of relatively minor refinements superimposed ordinal, familial, or generic significance may on an already highly specialized skeletal de- be very disconcerting. Although the Micro- sign. hylidae are included in the Displasiocoela, One of the most striking features of the Parker (1934) has pointed out that three of anuran skeleton is its capacity for variation the eight recognized subfamilies have uni- at both the intraspecific and supraspecific formly procoelous vertebrae, while a fourth levels. Among the many possible examples subfamily includes some genera with a proof intraspecific variation, an interesting one coelous column and others with a diplasiocoe- is that occurring in the sacral articulation of lous. Other examples of this type are further the early Tertiary ?pipid Eoxenopoides re- discussed by Parker (1932)? such as the mingi described by Haughton (1931). He presence or absence of teeth in the upper jaw divides the sacral variants into five "mor- as in the Leptodactylidae, the variation in phological stages." More than 50 specimens, the vomero-palatine region, in the degree of representing an exceptionally large sample dilation of the sacral diapophyses, and in the for fossil anurans, were found scattered ornamentation of the skull roof. through a very localized clay deposit over 100 While intraspecific variation may indicate feet in thickness. There is no evidence that weak selection towards structural uniformity, 66 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 93 supraspecific variation implies selection to- With the small samples of fossil anurans wards some sort of constant morphological that are usually available, it is very difficult and biomechanical pattern at some particular to consider variation properly, or to distin- taxonomic level. Why some genera of the guish between the two types just discussed. microhylid subfamily Microhylinae are pro- In connection with the frogs described in this coelous and others diplasiocoelous is un- paper, a number of such problems are evi- known. There appears to be no direct en- dent, such as the significance of the skull vironmental correlation although such, how- ornamentation in Eophractus, the increased ever subtle, probably exists. In spite of inter- width of the nasal bridge in Calypiocephalella esting experiments by Fick and others, there canqueli, and the edentulous condition and is still no satisfactory basic functional ex- dilated sacral diapophyses of Neoprocoela. planation for the different types of vertebral articulation. REFERENCES AmL, ERNST their bearing on classification and phy- 1926. Anura; Aglossa, Xenopodidae. In Kai- logeny. Bull. Amer. Mus. Nat. Hist., ser, Erich, Die Diamantenwulste Sud- vol. 46, art. 1, pp. 1-87, pls. 1-23. west-Africas. Berlin, Verlag von Die- 1924. A new spadefoot toad from the Oligo- trich Reimer, vol. 2, chap. 21, sect. 12, cene of Mongolia with a summary of the pp. 141-142, fig. 22, pl. 42. evolution of the Pelobatidae. Amer. AMEGHINO, FLORENTINO Mus. Novitates, no. 132, pp. 1-15, figs. 1899. Sinopsis Geologico-Paleontologica. Su- 1-7. plemento (Adiciones y Correcciones). La 1925. The evolution and dispersal of the frogs. Plata, Imprenta y Encuadernaci6n "La Amer. Nat., vol. 59, pp. 265-271. Libertad," pp. 1-13. 1928. Two new fossil Amphibia of zo6geo- 1901. L'age des formations sedimentaires de graphic importance from the Miocene of Patagonie. An. Soc. Cient. Argentina, Europe. Amer. Mus. Novitates, no. 303, vol. 51, pp. 65-91. pp. 1-13, figs. 1-6. DUNN, E. R. 1930. The fossil frogs of the Intertrappean 1923. The geographical distribution of am- beds of Bombay, India. Ibid., no. 401, phibians. Amer. Nat., vol. 57, pp. 129- pp. 1-13, figs. 1-3. 136. 1931. The biology of the Amphibia. New 1931. The herpetological fauna of the Ameri- York, McGraw-Hill Book Co., xiii +577 cas. Copeia, no. 3, pp. 106-119. pp., 174 figs. FILHOL, H. PARKER, H. W. 1877. Recherches sur les phosphorites du 1929. Two fossil frogs from the Lower Miocene Quercy. Ann. Sci. G6ol., Paris, vol. 8, of Europe. Ann. Mag. Nat. Hist., vol. 4, pp. 1-340, pls. 1-26. ser. 10, pp. 270-281, figs. 1-4. FRIANT, MADELEINE 1932. Parallel modifications in the skeleton of 1944. Caracteres anatomiques d'un batracien the Amphibia Salientia. Arch. Zool. Oligoc8ne de la Limagne, le Prodisco- Italiano, vol. 16, pp. 1239-1240, figs. 1- glossus vertaizoni. Compt. Rendus 11. Acad. Sci. Paris, vol. 219, pp. 561-562, 1934. A monograph of the frogs of the family 1 fig. Microhylidae. London, British Mu- HAUGHTON, S. H. seum (Natural History), 208 pp., 67 1931. On a collection of fossil frogs from the figs. clays at Banke. Trans. Roy. Soc. South 1940. The Australian frogs of the family Lep- Africa, vol. 19, pp. 233-249, figs. 1-10t todactylidae. Novitates Zool., vol. 42, pls. 23-24. pt. 1, pp. 1-106, figs. 1-120, pl. 1. JOLEAUD, LEONCE PIVETEAU, JEAN 1939. Atlas de pal6obiogeographie. Paris, Paul 1927. ltudes sur quelques amphibiens et rep- Lechevalier, 39 pp., 99 pls., plus descrip- tiles fossiles. Ann. Pal6ont., vol. 16, pp. tive text. 57-97, pls. 8-11. MEYER, H. VON 1937. Un amphibien du Trias inferieur; essai 1847. [Da ich die als zweite Abtheilung "zur sur l'origine et l'6volution des amphibien Fauna der Vorwelt" . . . I Neues Jahrb. anoures. Ibid., vol. 26, pp. 133-177, figs. Min. Geol. Palaiont., pp. 181-196. 1-30, pls. 19-20. MOODIE, R. L. PORTiS, ALESSANDRO 1914. The fossil frogs of North America. 1885. Resti di batraci fossili Italiani. Atti Amer. Jour. Sci., ser. 4, vol. 38, pp. 531- Accad. Sci. Torino, vol. 20, pp. 1173- 536, figs. 1-2. 1201, pl. 13. NICHOLLS, GEORGE E. REINBACH, W. 1916. The structure of the vertebral column in 1939. Untersuchunger iiber die Entwicklung the Anura Phaneroglossa and its impor- des Kopfskeletts von Calyptocephalus tance as a basis of classification. Proc. Gayi. Jenaische Zeitschr. Naturwiss., Linnean Soc. London, 128th Session, pp. vol. 72, pp. 211-362, figs. 1-33, pls. 13- 80-91, 1 fig. 18. NOBLE, G. K. ROVERETO, CAYETANO 1922. The phylogeny of the Salientia. I.-The 1914. Los estratos araucanos y sus fosiles. osteology and the thigh musculature; [Part] 4, Los f6siles de Monte Hermoso, 67 68 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 93 localidad tlpica del Hermosenes. An. of Nevada. Univ. Kansas Sci. Bull., vol. Mus. Nac. Hist. Nat. Buenos Aires, vol. 27, pp. 61-69, fig. 1, pl. 1. 25, pp. 116-179, figs. 53-70, pls. 1942. Extinct toads and frogs from the Upper RuscoNi, CARLOS Pliocene deposits of Meade County, 1932. La presencia de anfibios ("Ecaudata") Kansas. Thid., vol. 28, pp. 199-235, pls. y de aves f6siles en el piso ensenadense 14-20. de Buenos Aires. An. Soc. Cient. Argen- VIDAL, L. M. ina, vol. 113, pp. 145-149, 1 fig. 1902. Nota sobre la presencia del tramo Kime- SCHEAErR, BOBB ridgense en el Montesch (L6rida) y hal- 1947. An Eocene serranid from Patagonia. lazgo de un batracio en sus hiladas. Amer. Mus. Novitates, no. 1331, pp. 1- Mem. R. Acad. Cien. Barcelona, vol. 4, 9, figs. 1-4. pl. 263-26, pl. 4. SamssI, MAX WATSON, D. M. S. 1924. Tertiary vertebrates from Mongolia. 1941. The origin of the frogs. Trans. Roy. Soc. Palaeont. Sinica, ser. C, vol. 1, fasc. 1, Edinburgh, vol. 60, no. 7, pp. 195-213, pp. 1-132, pls. 1-6. fig. 1-23. SIMsoN, G. G. WEITZEL, KARL 1940. Mammals and land bridges. Jour. Wash- 1938. Propelodytes wagneri n. g. n. sp., ein ington Acad. Sci., vol. 30, no. 4, pp. 137- Frosch aus dem Mitteleozan von Mes- 163, figs. 1-6. sel. Notizbl. Hessischen Geol. Lande- 1943. Turtles and the origin of the fauna of sanst. Darmstadt, ser. 5, vol. 19, pp. 42- Latin America. Amer. Jour. Sci., vol. 46, pl. 6. 241, pp. 413 429. WOLTERSTORFF, W. STRAND, EMBRIK 1929. tYber fossile Fr5sche aus der Papierkohle 1926. Miscellanea nomenclatorica zoologica et von Burgbrohl (Laacher See). Jahrb. palaeontolOgica, I-II. Arch. Natur- Preussischen Geol. Landesanst, for gesch., div. A, vol. 92, no. 8, pp. 1-75. 1928, vol. 49, pt. 2, pp. 918-932, figs. 1- TAYLoI, E. H. 4, pls. 51-52. 1937. Una nueva fauna de batracios anuros del YOUNG, C. C. Plioceno medio de Kansas. An. Inst 1936. A Miocene fossil frog from Shantung. Biol. Univ. Nac. Autonoma de Mexico, Bull. vol. 7, pp. 513-529, pls. 1-2. Geol. Soc. China, vol. 15, pp. 189- 1941. A new anuran from the middle Miocene 193, figs. 1-2, pl. 1.