HARVARD UNIVERSITY Library of the Museum of Comparative Zoology B R E V I 0 R A useuiii of Comparative Zoology HAR\ ARD UNIVERSITY Numbers 380^09 1971-1973 CAMBRIDGE, MASS., U.S.A. 1974 Edited by Penelope Naumann BREMORA Museum of Comparative Zoology CONTENTS Numbers 380-409 1971 No. 380. The Taxonomic Position of the Genus Petria Semenow (Coleoptera: Allecuhdae). By John F. Lawrence. 9 pp. June 15. No. 381. Tetraploidy in a Hybrid Lizard of the Genus Cnemi- dophorus (^Teiidae). B\' \\'illiani B. Xeaves. 25 pp. December 20. 1972 No. 382. Osteology and Description of Thrattidion noctivagus, a Minute, New Freshwater Clupeid Fish from Cam- eroon, with a Discussion of Pellonulin Relationships. By Tyson R. Roberts. 25 pp. February 25. No. 383. Two New Genera of Bembidiine Carabid Beetles from Australia and South America with Notes on their Phylogenetic and Zoogeographic Significance (Co- leoptera). B) Terr\ L. Erwin. 19 pp. February 25. No. 384. An Attempt to Determine the Systematic Position of EUopostoyna ynegaloyyiycter, an Enigmatic Fresh- water Fish from Borneo. By Tyson R. Roberts. 16 pp. February 25. No. 385. The Chafiares (Argentina) Triassic Reptile Fauna. XII. The Postcranial Skeleton of the Thecodont Chayiaresuchus. By .Alfred Sherwood Romer. 21 pp. Februarys 25. No. 386. A Review of the Australian Species of Elaphrosyron and Telostegus, with Notes on Other Genera (Hy- menoptera, PompiHdae). By Howard E. Evans. 18 pp. February 25. No. 387. Growth Changes in Amphilimna olivacea (Lyman) and the Systematic Status of Amphitarsus spinifer Schoener. By Lowell P. Thomas and Amy Schoener. 9 pp. March 15. No. 388. Intense Low-Frequency Sounds from an Antarctic Minke Whale, Balaenoptera acutorostrata. By Wil- liam E. Schevill and William A. W'atkins. 8 pp. April. No. 389. The Chahares (Argentina) Triassic Reptile Fauna. XI IL An Early Ornithosuchid Pseudosuchian, Gra- cilisuchus stipanicicorum, Gen. et Sp. Nov. By Al- fred Sherwood Romer. 24 pp. August 1 1 . No. 390. The Chanares (Argentina) Triassic Reptile Fauna. XIV. Lewisuchus admixtus, Gen. et Sp. Nov., a Further Thecodont from The Chaiiares Beds. By Alfred Sherwood Romer. 1 3 pp. August 1 1 . No. 391. The Relationship of Island Area and Isolation to Color Polymorphism in Liguus fasciatus (Pulmonata, Bul- imulidae). By Michael A. Rex. 15 pp. August 11. No. 392. A Fossil Pelomedusid Turtle from Puerto Rico. By Roger Conant Wood. 13 pp. August 11. No. 393. Minute Jamaican Prosobranch Gastropods: Stoastoma and its Congeners. By Kenneth J. Boss. 13 pp. x\ugust 1 1 . No. 394. The Chahares (Argentina) Triassic Reptile Fauna. XV. Further Remains of the Thecodonts Lagerpe- ton and Lagosuchus. By Alfred Sherwood Romer. 7 pp. August 11. No. 395. The Chanares (Argentina) Triassic Reptile Fauna. XVI. Thecodont Classification. By Alfred Sher- wood Romer. 24 pp. November 6. No. 396. The Chaiiares (Argentina) Triassic Reptile Fauna. XVn. The Chanares Comphodonts. By Alfred Sherwood Romer. 9 pp. November 6. No. 397. Xyloredo, a New Teredinid-like Abyssal Wood-borer (Mollusca, Pholadidae, Xylophagainae) . By Ruth D. Turner. 1 9 pp. November 6. 1973 No. 398. Pseudobeaconia, a Perleidiform Fish from the Triassic Santa Clara Formation, Argentina. By Peter Hutch- inson. 24 pp. March 6. No. 399. The Evolution of Mammalian from Reptilian Denti- tions. By J. W. Osborn and A. W. Crompton. 18 pp. March 6. No. 400 The Carolina Salt Marsh Snake : a Distinct Form of Matrix sipedon. By Roger Conant and James D. Lazell, Jr. 13 pp. March 6. No. 401. The Chanares (Argentina) Triassic Reptile Fauna. XVm. Probelesodon minor, a New Species of Car- nivorous Cynodont; Family Probainognathidae Nov. By Alfred Sherwood Romer. 4 pp. March 6. No. 402. Speciation in the Genus Ochthoeca (Aves: Tyran- nidae). By John W. Fitzpatrick. 13 pp. June. No. 403. A New Late Paleocene Phenacodont (Mammalia: Condylarthra) from Western Colorado. By Br) an Patterson and Robert M. West. 7 pp. June. No. 404. New Studies on a Montane Lizard of Jamaica, Anolis reconditus. By Robert Hicks. 23 pp. June. No. 405. A Fossil Trionychid Turtle from South America. By Roger Conant Wood and Bryan Patterson. 10 pp. September 20. No. 406. Two New Lygosomine Skinks from New Guinea with Comments on the Loss of the External Ear in Ly- gosomines and Observations on Previously Described Species. By Allen E. Greer, Jr. 25 pp. September 20. No. 407. The Chafiares (Argentina) Triassic Reptile Fauna. XIX. Postcranial Materials of the Cynodonts Pro- belesodon and Probainognathus. By Alfred Sher- wood Romer and Arnold D. Lewis. 26 pp. Septem- ber 20. No. 408. A Taxonomic Comparison of the American Upogebia (Decapoda, Thalassinidea), Including Two New Species from the Caribbean. By Da\'id Thistle. 23 pp. September 20. No. 409. The Classification of the Cotingidae (Aves). By D. W. Snow. 27 pp. September 20. BREVIORA Museum of Comparative Zoology INDEX OF AUTHORS Numbers 380-409 1971-1973 No. Boss, Kenneth J 393 CoNANT, Roger 400' Crompton, a. W 399 Erwin, Terry L. 383 Evans, Howard E. 386 Fitzpatrick, John W 402 Greer, Allen E 406 Hicks, Robert 404 Hutchinson, Peter 398 Lawrence, John F 380 Lazell, James D 400 Lewis, Arnold D 407 Neaves, William B 381 OsBORN, J. W 399 Patterson, Bryan 403, 405 Rex, Michael A. . 391 Roberts, Tyson R 382, 384 RoMER, Alfred Sherwood 385, 389, 390, 394, 395, 396, 401, 407 Schevill, William E 388 ScHOENER, Amy 387 Snow, D. W 409 Thistle, David 408 Thomas, Lowell P 387 Turner, Ruth D, 397 Watkins, Willl\m E 388 West, Robert M 403 Wood, Roger Conant 392, 405 B R E V I O'^'teA Museiiiii of Comparative Zoology Cambridge. Mass. 15 June, 1971 Number 380 THE TAXONOMIC POSITION OF THE GENUS PETRIA SEMENOW (COLEOPTERA: ALLECULIDAE) John F. Lawrence^ Abstract. The beetle genus Petria Semenow includes a few species in- habiting the Kara-Kum Desert. U.S.S.R. They are known only from males, which are soft-bodied, with short elytra, long antennae, and large eyes. The genus is usually placed in a separate family, the Petriidae, within the section Heteromera, but some Russian workers have included it within the family Alleculidae. A detailed study of Petria antcnnata Semenow provides abundant evidence supporting the inclusion of Petria within the subfamily Omophlinae of the Alleculidae. Brief descriptions and figures are given for the prothorax, metendosternite, hindwing, tarsal claw, abdomen, and male genitalia, and comparisons are made with members of the Alleculidae and other heteromeran families. Speculations are made concerning the habits and habitat of the larva and female based on knowledge of related omoph- lines and analogous types of desert-inhabiting Coleoptera. The genus Petria Semenow includes a few species of peculiar looking beetles that inhabit the Kara-Kum, a desert east of the Caspian Sea (Turkmen S.S.R. ). The group is known only from males, which are attracted to lights at night. They are relatively small (3.0-4.25 mm), soft-bodied, and lightly pigmented insects, with shortened elytra, long antennae, and large, prominent eyes {see illustrations in Jacobson, 1913, 1915; Ogloblin and Znoiko, 1950; and Znoiko, 1936). The genus has been placed in a sep- arate family, the Petriidae. which is considered in most general texts to be of uncertain phylogenetic position within the section Heteromera (Crowson. 1955). Although Semenow ( 1 893. 1 896) made Petria the type of a new family, he considered the genus to be related to members of the ' Museum of Comparative Zoology, Cambridge, Mass. 02138 2 BREVIORA No. 380 Allcculidae and particularly the omophline genus Steneryx Reitter, also from Central Asia. Most later authors recognized the family Petriidae, but Znoiko (1936) presented evidence for the inclusion of Petria within the alleculid subfamily Omophlinae and pointed out a transition in general form, eye size, wing venation, tarsal claws, and antennal structure, among the species of Petria, Stene- ryx, and a third genus, Cnecosochara Reitter, also known from males only. In spite of Znoiko's conclusions, the Petriidae is still treated as a heteromeran family of doubtful affinities, and Crowson (1955) made no attempt to place it, since specimens for dissection were not available to him. The major problem in classifying Petria on the basis of super- ficial characters is that, while it closely resembles several Allcculi- dae, it has been thought to lack the main diagnostic features of that family, namely, the closed procoxal cavities, connate basal ab- dominal sternites (3-5), and pectinate tarsal claws. In order to obtain material for dissection, I wrote to Leningrad, and, through the kindness and cooperation of Dr. G. S. Medvedev of the Zoologi- cal Institute, Academy of Sciences, U.S.S.R., received three speci- mens from Semenow's series of P. antennata, collected at Utsh- adzhi in May of 1889. Sincere thanks are due to Dr. Medvedev and the Zoological Institute. I am also grateful to R. A. Crowson for his continual inspiration, comments, and criticisms, and to P. J. Darlington, Jr., and T. F. Hlavac for their useful suggestions and critical review of the manuscript. Although specimens of Steneryx and Cnecosochara were not available for study, dissec- tions were made of Cteniopus flavus (Scopoli) and examinations were made of several alleculids and other Heteromera. A study of the above material revealed the following features of Petria antennata that shed some light on its phylogenetic relation- ships: 1 ) The procoxae are subconical, projecting, and almost con- tiguous, the intercoxal process of the prosternum being laminate (Fig. 1 ), but each coxa bears a relatively large articular region (a), which is concealed by the coxal cowling (c), so that the articula- tion with the pleuron is internalized (Fig. 2). 2) The pleuro-coxal mechanism of the prothorax is of the tene- brionoid type, with the endopleuron fused to the notal wall and the trochantin apparently absent. 1971 GENUS PETRIA 3 3) The procoxal cavities are closed internally and open exter- nally or posteriorly (Fig. 1). 4) The mesocoxal cavities are not closed outwardly by the meso- and metasterna and are thus contiguous laterally with the mes- epimera. 5) The metendosternite is of the tenebrionoid type, with a narrow stalk, no laminae, and the anterior tendons out near the apices of the lateral arms (Fig. 3). 6) The tibial spurs are neither serrate nor pubescent. 7 ) The tarsal claws each have two or three toothlike projections (Fig. 5). 8 ) The hindwing is similar to that of most tenebrionoids, with four well-developed anal veins (in the main group), an anal or wedge cell (w) present, and a short stalk of Rs extending basad of the radial cell (not shown in Znoiko, 1936) (Fig. 4). 9) Abdominal sternites 3, 4, and 5 are connate. This is not easily observed without sectioning, since all the segments are broadly overlapping (Figs. 6 and 7). 10) Abdominal sternite 7 has a pair of posterolateral gland openings (Fig. 9) similar to those found in Cteniopus {see Ken- dall, 1968). ^ 1 1 ) Abdominal sternite 8 is developed into a pair of claspers (Figs. 6 and 8) that are similar to those of alleculids {see Camp- bell, 1966; Champion, 1888; Kaszab, 1969; McDonald, 1960; and Oglobhn and Znoiko, 1950). 12) The aedeagus is of the normal (not inverted) heteromeroid type, with a long basal piece (b), short tegmen (apical or cap piece) (t), and the median lobe (m) membranous except for a ventral strut (or two fused struts) at the apex (Fig. 10). This is a typical alleculid aedeagus as illustrated in Campbell (1966. 1968); McDonald (1960); Marshall (1970a, 1970b); and Sharp and Muir (1912). Except for the lack of an external coxal closure, the condition of the prothorax in Petria is strongly suggestive of a relationship to the tenebrionoid complex of the Heteromera (Tenebrionidae, Lagriidae, Alleculidae, and Nilionidae). The internalization and concealment of the pleural articulation, fusion of the endopleuron to the notum, and reduction of the trochantin are characteristic of 4 BREVIORA No. 380 the tenebrionoid families and a few related groups, such as the Colydiidae, Zopheridae, Monommidae, Prostomidae, and Daco- deridae; a similar condition also occurs in certain Clavicornia (Propalticidae, cerylonoid complex) (Crowson, 1955; Hlavac, per- sonal communication; Watt, 1967). The internal closure of the procoxal cavities occurs in most Heteromera, but is absent in the Mycetophagidac, Ciidae, Pterogeniidae, Tetratomidae, Perimylopi- dae, Zopheridae, Monommidae, Pythidae, and Pyrochroidae. The development of conical, projecting procoxae and a laminate inter- coxal process has taken place in various members of the tene- brionoid association (Lagrio, Mycetochara, Cteniopus), but externally open procoxal cavities are unknown in this large group, except in a few sub-Antarctic forms that are doubtfully included or have been removed (Crowson, 1955; Watt, 1967). The laterally open mesocoxal cavity is a fairly common feature in the Heteromera, but in the families Prostomidae, Zopheridae, Monommidae, Elacatidae, Mycteridae, Inopeplidae, and Salpingi- dae the cavity is closed by the meeting of the meso- and metasterna (Crowson, 1955; 1967). The metendosternite of Petria is also of the tenebrionoid type, with no laminae and with laterally placed tendons, and differs from that found in most groups of Heteromera. A similar struc- ture may be found, however, in certain Melandryidae and Scrap- tiidae, which differ from Petria by having serrate or pubescent tibial spurs and free abdominal sternites (Crowson, 1938, 1944, 1955, 1966). The wing venation also points to the tenebrionoid complex, be- ing almost identical with that of certain omophline Alleculidae and a number of Tenebrionidae; in most Melandryidae, the venation is simpler with the anal cell absent, but in Melcmdrya it is essentially the same as that of Petria (Bernet-Kempers, 1923; Crowson, 1955, 1966; Forbes, 1922; Znoiko, 1936). The fusion of the basal three abdominal sternites and the presence of glands on the seventh sternite are probably the strong- est pieces of evidence for the association of Petria with the Tenebrionidae and their relatives. The fusion of sternites is not uncommon in the Heteromera and has probably occurred several times, but the connation of the first three appears to be restricted to the tenebrionoids, certain Colydiidae, {Mrymechixemis, Pyc- nomerus, Anchomma) and Meryx (Merycidae). Two sternites 1971 GENUS PETRIA 5 are united in the Pteiogeniidae, Prostomidae, Cononotidae, Myc- teridae, Lo^rioida (Anthicidae), and a few other genera of doubtful affinities, while four are actually or apparently fused in the ma- jority of Colydiidae, Zopheridae, Monommidae, and Dacoderidae (Crowson, 1955, 1967; Watt, 1967). The situation is complicated by the fact that fusions are not always visible from the surface (as in Petria) or intersegmental membranes are concealed (as in Dacoderiis, see Watt, 1967). The apparent fusion of three basal abdominal sternites, however, is a consistent feature of the tene- brionoid families and is almost always correlated with several other adult and larval characters. The presence of defense glands on the seventh abdominal ster- nite is known, according to Kendall (1968), only in the families Lagriidae, Alleculidae, and Tenebrionidae. Although the actual glandular reservoirs were not seen in my material, openings and ducts, similar to those of Cteniopiis sulphuripes (Linnaeus) and C. ftuvus (Scopoli), were present at the posterior angles of the sternite. The presence of small teeth on the tarsal claws, the modification of the eighth sternite into abdominal claspers, and the form of the aedeagus all argue for the inclusion of Petria in the family Alle- culidae, especially when coupled with the several tenebrionoid features mentioned above. Male claspers may be found in other groups, such as the Oedemeridae and Cephaloidae, but the struc- ture of these organs is entirely different and the other tenebrionoid characters are lacking (Arnett, 1951; 1953). The arguments presented above, when added to the comparisons of Znoiko (1936), leave little doubt that Petria belongs in the family Alleculidae and is related to Steneryx, Cnecosochara, and their allies in the subfamily Omophhnae. It represents an adaptive extreme characterized by the loosely built and lightly sclerotized body, shortened elytra, large eyes, loss of the combs on the tarsal claws, loss of the external coxal closure, and sexual wing dimorph- ism. The last feature is not actually known in Petria but may be postulated by analogy with other groups known from males only (see below). Semenow (1893) speculated that the females of Petria, since they were never collected along with males, might be wingless parasites of Hymenoptera or Orthoptera, as is the case in certain Meloidae and Rhipiphoridae. In other omophlines, however, such 6 BREVIORA No. 380 as Omophlus, Podonta, and Cteniopus, the larvae are soil inhabi- tants, feeding on various roots and tubers, while the adults are usually found on flowers and fruits (Aguilar, 1962; Kaszab, 1969). A more reasonable hypothesis, then, would be that Petria larvae are root feeders and that the wingless females live either in the soil or at the bases of shrubs or grasses. This particular type of wing dimorphism and habitat selection is known in at least two other xerophilous beetles, the males of which resemble those of Petria, namely Vesperus (Cerambycidae) and Anorus (Dascilli- dae). Species of Vesperus inhabit the Mediterranean Region; the wingless female deposits her eggs above the ground, and the fast moving, triungulinlike larvae enter the soil and are transformed into radically dift'erent grubs, which feed on roots (Balachowsky, 1962). Anorus species occur in the deserts of southwestern North America and Chile; the larvae are unknown (but those of the re- lated Dascillus are root feeders), and the only known female is wingless and subterranean (Blaisdell, 1934). It is likely that the females and larvae of Petria, when they are discovered, will have similar habits. LITERATURE CITED Aguilar, J. d". 1962. Famille des Alleculidae, pp. 368-373. In A. S. Bala- chowsky (Ed.), Entomologie appliquee a I'Agriculture. Traite. Tome I. Coleopteres. Premier Volume. Paris, Masson, XXVII + 564 pp. Arnett, R. H., .jr. 1951. A revision of the Nearctic Oedemeridae (Coleop- tera). American Midland Nat., 45: 257-391. 1953. A review of the beetle family Cephaloidae. Proc. U. S. Nat. Mus., 103: 155-161. Balachowsky, A. S. 1962 Famille des Cerambycidae, pp. 394-434. In A. S. Balachowsky (Ed.), Entomologie appliquee a I'Agriculture. Traite. Tome I. Coleopteres. Premier Volume. Paris, Masson. Bernet-Kempers, K. 1923. Abbildungen von Fliigelgeader der Coleop- teren. Ent. Mitt., 12: 71-115. Blaisdell, F. E. 1934. Rare North American Coleoptera. Trans. American Ent. Soc, 60: 317-326. Campbell, J. M. 1966. A revision of the genus Lohopoda (Coleoptera: Alleculidae) in North America and the West Indies. Illinois Biol. Mon., 37: 1-203. 1968. A revision of the Mexican and Central American species of Isomira (Coleoptera: Alleculidae). Canadian Ent., 100: 449-469. 1971 GENUS PETRIA 7 Champion. G. C. 1888. Fam. Cistelidae, pp. 384-465, pis. 17-21. In F. D. Godman and O. Salvin (Eds.), Biologia Centrali-Americana, Insecta, Coleoptera. Vol. 4. Part 1. Heteromera (part). London, Porter. Crowson, R. a. 1938. The metendosternite in Coleoptera: a comparative study. Trans. Roy. Ent. Soc. London. 87: 397-415. 1944. Further studies on the metendosternite in Coleoptera. Trans. Roy. Ent. Soc. London, 94: 273-310. 1955. The Natural Classification of the Families of Coleop- tera. London. Lloyd. 187 pp. 1966. Observations on the constitution and subfamilies of the family Melandryidae. Eos, 41: 507-513. 1967. The natural classification of the families of Coleoptera. Addenda and corrigenda. Ent. Mon. Mag., 103: 209-214. Forbes, W. T. M. 1922. The wing-venation of the Coleoptera. Ann. Ent. Soc. America, 15: 328-352, pis. 29-35. Jacobson, G. G. 1913. Zhuki Rossii i zapadnoi Evropi. Vip. X. Leningrad, Devrient. Pp. 721-864, pis. 76-83. 1915. Zhuki Rossii i zapadnoi Evropi. Vip. XI. Leningrad, Devrient. Pp. 865-1024. Kaszab, Z. 1969. Famille: AUecuhdae, pp. 215-229. In H. Freude, K. Harde, and G. Lohse (Eds.). Die Kafer Mitteleuropas. Band 8. Kre- feld, Goecke and Evers. 388 pp. Kendall, D. A. 1968. The structure of the defense glands in Alleculidae and Lagriidae (Coleoptera). Trans Roy. Ent. Soc. London, 120: 139-156. McDonald, J. M. 1960. Morphology of the exoskeleton of Capnochroa jiiliginosa (Melsheimer) (Alleculidae). Coleopt. Bull., 14: 97-120. Marshall, J. D. 1970a. Isomira Mulsant in America north of Mexico (Coleoptera, Alleculidae): Redescriptions, new synonymies, and tax- onomic notes on eastern North American species. Ent. News, 81 : 41-49. 1970b. Isomira Mulsant in America north of Mexico (Co- leoptera: Alleculidae): Species of eastern North America concluded. Coleopt. Bull., 24: 88-95. Ogloblin, D. a., and D. V. Znoiko. 1950. Piltseyedi (Sem. Alleculidae). Ch. 2, Podsem. Omophlinae. Fauna S.S.S.R. Zhestkokriliye. Tom. 18, Vip. 8 (Nov. Ser., No. 44). Moscow and Leningrad. Akademia Nauk S.S.S.R. 135 pp. Semenow, a. 1893. De Coleopterorum familia nova. Melanges Biol., 13: 359-366; Bull. Acad. Imp. Sci. St.-Petersbourg (N. S.), 3(35) : 607-615. 1896. Revisio specierum generis Petria Sem. Annuaire Mus. Zool. Acad. Imp. Sci. St.-Petersbourg, 1: 25-30. 8 BREVIORA No. 380 Sharp, D., and F. Muir. 1912. The comparative anatomy of the male genital tube in Coleoptera. Trans. Ent. Soc. London, 1912: 477-642, pis. 42-78. Watt, J. C. 1967. The families Perimylopidae and Dacoderidae (Coleop- tera, Heteromera). Proc. Roy. Ent. Soc. London, Ser. B, 36: 109-118. Znoiko, D. V. 1936 Ueber die systematische Stellung der Familie Petriidae (Coleoptera). Trudi Zool. Inst. Akad. Nauk S.S.S.R., 3: 67-77. 1971 GENUS PETRIA Figures 1-10. Petria antennata Semenow. male (1 line = 0.125 mm unless otherwise indicated). Fig. 1. Prothorax, ventral view, with left coxa re- moved. Fig. 2. Procoxa, showing internal articular area (a). Fig. 3. Meten- dosternite, dorsal view. Fig. 4. Hindwing (0.50 mm). Fig. 5. Tarsal claw (0.038 mm). Fig. 6. Abdomen (0.25 mm). Fig. 7. Cross section of ab- dominal sternites 3-6 (0.076 mm). Fig. 8. Abdominal claspers and asso- ciated structures, ventral view. Fig. 9. Abdominal sternite 7, showing glandular ducts and openings. Fig. 10. Aedeagus, lateral view. M^S. COMP. 200L LIBRARY MAR 1 8 1985 B IR E V„,|^§ R A Museum of Comparative Zooloi^y Cambridge, Mass. 20 Dec-ember. 1971 Number 381 TETRAPLOIDY IN A HYBRID LIZARD OF THE GENUS CNEMIDOPHORUS (TEIIDAE) William B. Neaves^ Abstract. An apparent hybrid between triploid. parthenogenetic Cneiuidophorits exsan(;iiis and diploid, sexual C. inonuitiis is shown to possess a tetraploid chromosom; complement. Evidence suggests that this tetraploid karyotype resulted from the fusion of a haploid sperm pronucleus of C. inornatus with an egg pronucleus carrying the unreduced triploid chromosome complement of C. cxsauiniis. Observations of cap- tive C. inornatus males show their propensity to engage in mating be- havior with C. exsangnis parthenogenones. These findings are discussed with regard to the origin and genetic mechanism of parthenogenesis in Cnciuidophorus. INTRODUCTION The hypotiiesis that parthenogenetic species of Cnefuiclophorus arose from the offspring of interspecific hybridizations (Lowe and Wright, 1966) has been supported by evidence from studies of lactate dehydrogenase (Neaves and Gerald, 1968) and adenosine deaminase (Neaves, 1969). A particular aspect of that hypothesis, namely, that triploid parthenogenones originated in crosses be- tween males of a sexual species and females of a diploid, par- thenogenetic species, has received support from reports that such hybridizations occur in nature. Taylor and Medica (1966) described apparent hybrids between Cnemidophorus inornatus, a sexual species, and C. neomexicanus, a diploid parthenogenone. ' Department of Anatomy, Harvard Medical School, Boston, Massa- chuseUs 02115 2 BREVIORA No. 381 Wright and Lowe ( 1967a) verified the occurrence of C. inornatus X C. neomexicanus hybrids, conveniently designated C. ''per- plexus," at several localities in New Mexico where the two parental species are sympatric. The question arises as to whether or not hybridization in Cnemidophorus ends with the attainment of triploidy. Although triploid parthenogenones are common (of seven parthenogenetic species in the sexlineatus group, five are triploid), no tetraploid species are known in the genus. Lowe and Wright (1966) men- tion a suspected tetraploid individual that they considered to represent a hybrid of C. inornatus and C. uniparens, the latter a triploid parthenogenone, but no evidence, karyotypic or otherwise, was presented to verify their suspicion. My report describes in detail a documented instance of tetraploidy in a Cnemidophorus hybrid and seeks to explain why there are as yet no tetraploid species in nature. Observations of interspecific mating behavior involving male C. inornatus and females of various parthenogenetic species are included in this report. Finally, these findings are discussed with regard to the origin and genetic mechanism of parthenogenesis in Cnemidophorus. MATERIALS AND METHODS Animals used in these studies were captured alive in their native habitats in Colorado, New Mexico, and Texas during the summers of 1967 and 1968. Soon after capture, these lizards were shipped to Boston and Cambridge. Upon arrival in the laboratories, they were housed either in a rectangular pen 12 feet by 14 feet or in oval pens 3.5 feet by 5 feet. The pens had sand floors with tiles and sheets of cardboard scattered about for shelter. Sun lamps installed over the pens were automatically controlled to provide a twelve-hour daily photoperiod. A constant supply of open water and available food, in the form of mealworms and crickets, was provided. Aspects of lizard behavior reported here were observed and photographed among the captive inhabitants of the large rec- tangular pen described above. Most incidents were recorded dur- ing the winter of 1968-1969 when the pen held approximately 150 Cnemidophorus distributed among five sexual and five par- thenogenetic species. 1971 TETRAPLOID CNEMIDOPHORUS 3 The tetraploid karyotype illustrated in this report was obtained from a phcnotypically aberrant lizard captured at Alamogordo, New Mexico, in August, 1967. This lizard was maintained alive until May, 1968, when it was used for karyotypic study. Mitotic figures were obtained from tissue cultures of 1 mm cubes of heart tissue according to the following procedure. Small Falcon flasks were loaded with 5.0 ml of a mixture containing 80 percent Eagle's Minimum Essential Medium (Microbiological Associates) and 20 percent fetal calf serum. With the flask in an upright position, tissue explants were placed against the dry wafl of the flask. The flask was slowly lowered to its side so that the explants were covered by the medium but not dislodged from the plastic surface. The flasks were incubated at 33" C. without agitation. The cul- tures were examined regularly with an inverted phase contrast microscope in order to follow cell proliferation. On the ninth day of incubation, when mitotic activity was intense, 0.25 microgram of Velban (Grand Island Biological Co.) was added to each flask. Twenty-four hours later, the medium was discarded and 5.0 ml of 0.1 percent trypsin in 0.8 percent sodium citrate was added to each flask. FoUowing a twenty-minute incubation at 37 C, the flasks were shaken vigorously to free the dissociating cells from one another and from the plastic. From this point, the cultures were processed according to a modification of the method outlined by Moorhead et al. (1960). The solution from each flask was decanted into a centrifuge tube and spun at 1000 g for five minutes. The cell pellet was fixed in two changes of ethanol: acetic acid (3:1) and was suspended in 45 percent acetic acid. Drops of this suspension were placed on cold, wet slides and flame dried. The chromosomes on these slides were stained with 1 percent toluidine blue in 1 percent borax and were photo- graphed with an Olympus photomicroscope. THE DISCOVERY OF THE TETRAPLOiD In the course of a collecting trip to New Mexico in August, 1967, I visited an exceptionally dense Cnemidophonis population within the city limits of Alamogordo in Otero County. The population was largely restricted to a weed bed approximately 20 meters wide and over 1000 meters long, bounded on the west by the roadbed of the Southern Pacific Railroad and on the east 4 BREVIORA No. 381 by the Alamogordo City Park. Two Cnemidophorus species were found in the area described. C. exsanguis, a triploid parthe- nogenone, was most abundant, occurring at a density of approxi- mately 50 animals per acre; C. inornatus, a sexual species, oc- curred at a density of about 10 animals per acre. No lizards were seen in the park itself, which had a cover of closely mown grass. West of the roadbed, the cover consisted of mesquite shrub and cactus on an eroded surface virtually devoid of weeds or grass. Here both C. inornatus and C. tigris, a second sexual species, were abundant. In two days of collecting west of the roadbed, only a single C. exsanguis was seen. No C. tigris were found east of the roadbed in the weeds where C. exsanguis was so abundant. Forty-three C. exsanguis and eleven C. inornatus were captured in the Alamogordo weed bed in 1967. In addition, two aberrant Cnemidophorus were taken. One of these, MCZ* 100080, was the size of C. inornatus, with partially fused paravertebral stripes, a white ventral surface, and a rusty tint on its dorsum, causing it to resemble C. exsanguis. The day after its capture, this animal died. Its abdomen was opened to expose the gonads, which ap- peared rudimentary. The nature of this specimen remains a mystery. A second aberrant specimen, MCZ 101991 (Plate 1), resembled a typical C. exsanguis except that its paravertebral stripes were fused and its ventral surface and tail were suffused with a brilliant blue characteristic of the same surfaces in C. inornatus males. This exsanguis-like animal was maintained alive for almost nine months before it was sacrificed for a study of its chromosomes. It wai immediately suspected that MCZ 101991 might represent a cross between the sympatric species of the weed bed, C. inornatus and C. exsanguis. This lizard clearly possessed attributes characteristic of both suspected parental species. A decision was reached to allow the animal to remain alive as long as possible so that it 5 behavior might be observed, but at the same time, an assessment of its ploidy was desired. Accordingly, blood smears were prepared from the exsanguis-Vike animal and from other Cnemidopliorus known to be either diploid or triploid on the '•' Museum of Comparative Zoology 1971 TETRAPLOID CNEMIDOPHORUS 5 basis of karyotypic evidence (Lowe and Wright. 1966; Wright and Lowe, 1967b). These were sent to N. B. Atkin for dcnsi- tometric estimation of DNA in the nucleated erythrocytes. Dr. Atkin reported (in litt.) that karyotypically triploid species (3n = 69±1) possessed about 50 percent more DNA per nucleus than karyotypically diploid species (2n = 46) and that the aberrant specimen (MCZ 101991 ) possessed about 20 percent more DNA per nucleus than triploids such as C. e.\sanguis. The details of erythrocyte DNA analysis in Cnemidophonis will be reported else- where. Although a true tetraploid should have demonstrated approximately 30 percent more DNA per nucleus than a triploid, the results of Atkin's survey encouraged the suspicion that the aberrant specimen represented a hybrid carrying some genetic elements characteristic of C. inonuitus in addition to the basic triploid complement characteristic of C. exsangiiis. In December 1967, Professor C. H. Lowe examined the aber- rant lizard and declared it unlike any Cnemidophonis he had ever seen. Lowe agreed that it might represent a tetraploid hybrid between C. inonuitus and C. exsangiiis. Meanwhile, this lizard thrived in captivity and proved itself a vigorous competitor for food; it commonly robbed typical C. exsangiiis of mealworms. In January 1968, the aberrant lizard laid two eggs, thus con- firming the suspicion that it was female and suggesting that its ovaries were functional. Attempts were made to provide suitable conditions for ovoposition, but she did not take advantage of them, and the eggs were dessicated by the time 1 discovered them in my daily check. She produced no more eggs during her captivity, which extended through May, 1968. When the aberrant lizard was eventually sacrificed for karyo- typic analysis, the results (Plates 2 and 3 and Table 1 ) confirmed its suspected status as a tetraploid hybrid. The modal chromo- some number, based on examination of eight apparently complete chromosome spreads, is 9L The karyotype in Plate 2 is from the single spread showing 92 chromosomes. Considering the large number of chromosomes in the karyotype, the small size of many, and the probability of overlap, it is not surprising that fewer than 92 chromosomes are evident in most spreads. The tetraploid karyotype can be divided into four apparently identical haploid complements, each closely resembling the haploid chromosome 6 BREVIORA No. 381 complement of C. inornatus (Lowe and Wright, 1966). Three inornatiis-WkQ haploid sets constitute the typical triploid karyotype of C. exscmguis, which is similar to that of another triploid parthe- nogenone, C. uniparens (Lowe and Wright, 1966). The fourth haploid chromosome complement seen in the aberrant lizard's karyotype may have been contributed by C. inornatus itself, through an interspecific mating with C. exscmguis. OBSERVATIONS OF CAPTIVE SPECIMENS: INTERSPECIFIC MATING Observations of captive Cnemidophorus during autumn and winter of 1968-1969 suggest that C. inornatus males may be unusually disposed to interspecific mating with parthenogenetic females. Dozens of courtship encounters, including attempted copulation, were recorded between C. inornatus males and parthe- nogenetic females from the species C. exsanguis and C. tesselatus (Plate 4, figs. 5-9). The sequence of sexual behavior involved pursuit of a subject, often a parthenogenetic female but sometimes an individual of a sexual species, including other C. inornatus, for five to twenty seconds at a speed depending on the movement of the subject, the male apparently trying to stay near the tip of the subject's tail. Suddenly, the C. inornatus male would close with the subject, mount its back, and grasp in his jaws a fold of skin on the back of the subject's neck. If the subject acquiesced, the male would immediately execute sinuous movements while pressing its abdo- men against the subject's dorsum. With the subject still quiescent, the male would loosen its grip on the subject's skin and caress the back of the subject's head and neck with his jaw while continuing the sinuous body movements. The male seemed particularly in- tent on rubbing his ventral pelvis and tail base against the dorsal pelvis of the subject. Meeting no resistance, the male would maneuver his vent into the area between the subject's rear limb and tail base. When the subject was a large parthenogenetic female, this maneuver required the male to creep back from his original position in which his jaw touched the subject's head and neck (Plate 4, figs. 5-7). While struggling to approximate his vent to the subject's, the male would constantly stroke the sub- ject's dorsal pelvis and tail base with his rear hmb. At times, the 1971 TETRAPLOID CNEMIDOPHORUS 7 male"s hemipenis was visibly erect, but in spite of observed close apposition of tlie participants' vents, unequivocal intromission was not seen. It did not appear as though C. inornatus males met with any real cooperation, beyond mere quiescence, in their courtship efforts, but C. exsangiiis and C. tesselatiis at least permitted the sequence of events to proceed as far as described above. Other C. inoniatiis, particularly other males, made real elTorts to escape the attentions of an ardent but misguided suitor. Perhaps as a result, C. inornatus males exhibited a definite preference for the larger, parthenogenetic females over their conspecific females. Attempted intromission was observed only between C. inornatus males and parthenogenetic females. DISCUSSION The existence of a tetraploid lizard in a mixed population con- taining triploid parthenogenones and diploid, sexual lizards argues that tetraploid hybrids are synthesized in nature, although the genus Cnemidopliorus apparently contains no tetraploid species.^ That the tetraploid lizard described here is a hybrid and not an autopolyploid is suggested by its possession of phenotypic traits characteristic of both suspected parental species, by the propensity of the suspected male parent, C. inornatus, to engage in inter- specific mating behavior with parthenogenetic females such as C. exsanguis, and by the occurrence in nature of another interspecific hybrid, C. ''perplexus," which is considered to be the result of crosses between sexual and parthenogenetic lizards (Wright and Lowe, 1967a). Two questions are raised by the existence of the tetraploid hybrid. First is the question of the cytological events in gameto- genesis and fertihzation that permit a tetraploid to arise from a mating between a sexual male and a triploid, parthenogenetic female. This question, which is essentially concerned with genetic mechanisms operative in parthenogenesis, will be considered at some length in order to provide a background for the second question. The second question concerns the absence of tetraploid species in spite of the synthesis of hybrid tetraploids in nature. The first question has a trivial aspect, namely cytological events in the sexual male. Nothing more is required of the male lizard 1 See also the discussion of two tetraploid lizards in a mixed population of C. sunorae and C. tigris (Lowe et a!., 1970). 8 BREVIORA No. 381 than the production of typical sperm that carry a haploid chromo- some complement and that are capable of delivering this genome to the female pronucleus in a mature egg. A more complicated aspect of the question involves the cytological events that lead to the production of a mature egg whose pronucleus contains the unreduced somatic chromosome complement characteristic of the mother. This, in essence, is the question of the mechanism of parthenogenesis in Cnemidophorus, and it deserves detailed con- sideration. In Cnemidophorus, with the possible exception of C. lemniscatus (Vanzolini, 1970; Hall, 1970), interspecific hybridization is im- plicated in the origin of parthenogenesis. Oogenesis in parthe- nogcnones arising from interspecific hybridization must solve two major problems: 1 ) the maneuvering of potentially, and sometimes obviously, nonhomologous chromosome sets at meiosis, and 2 ) the production of a mature egg possessing the somatic chromosome number characteristic of the mother. Several solutions have been proposed: 1 ) mitotic egg production replaces the normal meiotic mechanism (Lowe and Wright, 1966), 2) failure of either first or second polar body formation, or fusion of one of these bodies with the egg nucleus after its formation (Beatty, 1957), 3) sup- pression of the first cleavage division (Beatty, 1957), or 4) endo- mitosis in the oogonium prior to the onset of meiosis (MacGregor and Uzzell, 1964). Solutions 2 and 3 cannot compensate for difticulties in bivalent formation that arise when nonhomologous chromosome sets are present at the onset of meiosis, nor can they compensate for situations in which more than two homologous chromosome sets enter meiosis. Since some diploid parthenogenones, such as the four Lacerta parthenogenones studied by Darevsky (1966), may be derived from closely related sexual species possessing fairly homologous karyotypes, solutions 2 and 3 cannot be ignored. In fact, Darevsky's ( 1966) study of parthenogenetic Lacerta suggests that the second meiotic division i> incomplete, with a diploid restitution nucleus resulting from the fusion of the daughter pro- nuclei during anaphase. Solutions 2 and 3 can, however, be dis- counted in all self-perpetuating triploids, such as C. exsanguis, and in diploids containing grossly nonhomologous haploid chromo- some complements, such as C. neomexicanus. If only solution 2 or 3 were operative in these cases, bivalent formation should fail, 1971 TETRAPLOID CNEMIDOPHORUS 9 owing to the absence of homologues, or, in those triploids that appear to have homologous haploid chromosome sets, owing to a tendency toward trivalent formation. Solution 1, mitotic egg production, overcomes difficulties in bivalent formation by avoiding the issue completely. This solution suffers from a lack of supporting evidence; no instance is known in which the germ line of a higher organism can facultatively abandon the meiotic theme. Furthermore, solution 1 is rendered dubious in Cnouidoplionis, the group for which it was originally proposed (Lowe and Wright, 1966), by evidence (personal ob- servations) that the nuclei of oocytes approximately 1 mm in diameter from ovaries of parthenogenetic species such as C. e.\Hini>uis, C. neonie.xicanus, and C tesselatus contain bivalent lampbrush chromosomes characteristic of meiotic diplotene. Hence, at least some components of typical meiosis occur in the ovaries of parthenogenetic Cnemidophonis. Solution 4, pre-meiotic endomitosis in the oogonia, has been shown to work in one group of parthenogenetic vertebrates, the ambystomatid salamanders (MacGregor and Uzzell, 1964), and appears likely in triploid poecilid fish (Schultz, 1967). Since pre- meiotic endomitosis can solve the problems of increased ploidy and nonhomologous chromosome complements, it is the most probable mechanism operative in parthenogenetic Cnemidophonis as well.^ This probability justifies discussion of its known features and its genetic implications. While studying preparations of diplotene lampbrush chromo- somes, MacGregor and Uzzell ( 1964) found 42 bivalents in each germinal vehicle from triploid parthenogenones (3n = 42), but in germinal vesicles from closely related, diploid sexual species (2n == 28), they found the expected 14 bivalents. This finding was explained by the postulation that endomitosis precedes meiosis in parthenogenetic oocytes, so that such oocytes enter meiosis with a hexaploid (6n = 84) chromosomal constitution. Meiotic DNA replication would then give an oocyte that is dodecaploid with respect to DNA, and meiosis would operate on ' A recent paper by Orlando Cuellar verifies the occurrence of this mechanism in the triploid parthenogenone Cnemidophonis uniparens. 10 BREVIORA No. 381 42 tetrads (bivalents) to yield a pronucleus with 42 chromosomes, which is the somatic complement for triploid ambystomatids. MacGregor and Uzzell (1964) suggest that only sister chromo- somes resulting from the endomitotic duplication associate to form bivalents. The genetic status of all parthenogenones that may rely on pre-meiotic endomitosis hinges on the question of bivalent formation, making a critical assessment of the MacGregor- Uzzell proposition necessary. The method of bivalent formation will determine whether the parthenogenetic lineage will retain the heterozygosity inherent in its hybrid origin or will experience an ultimate tendency toward homozygosity. While this question is not directly relevant to the issue of tctraploid hybrids, it is im- portant to an understanding of parthenogenesis in Cnemidophorits. Should sister chromosomes pair exclusively, a most conserva- tive pattern of inheritance would result; each offspring would, neglecting mutation, emerge with an exact copy of its mother's genome. Should homologous chromosomes also pair, independent assortment at first meiotic metcphase would cause the offspring to deviate from the mother's genotype, owing to the loss of alterna- tive alleles in the first polar body. In triploid parthenogenones relying on pre-meiotic endomitosis, the probability of losing both representatives (sisters) of a single chromosome from a homol- ogous set of three in a single generation can be calculated (p = 0.066), assuming random pairing of sister and homologous chro- mosomes, no crossing-over, and independent assortment. Similarly, after only two generations, p = 0.0066 that two chromosomes from a homologous set of three will be lost, leaving complete homozygosity at all loci on that chromosome. A diploid parthenogenone in which sister and homologous chromosomes pair randomly experiences an even stronger tendency toward homozygosity. The probability is 0.33 that one of a set of two homologous chromosomes will be lost in a single generation if crossing-over does not occur. In either diploidy or triploidy, crossing-over will only randomize the occurrence of homozygosity with respect to all loci on a single chromosome and will not delay the trend of the entire genome toward homozygosity. The ultimate consequence of participation of homologous chromosomes in bi- valent formation is homozygosity at all loci in the genome. The most sensitive test devised to assess the genetic status of 1971 TETRAPI.OID CNEMIDOPHORUS 11 parthenogenones, namely tissue grafts to determine histocompati- bility (Kallman, 1962; Maslin, 1967), does not discriminate between uniform clonal heierozygosity, which would be preserved in the case of exclusive pairing of sisters, and established clonal homozygosity resulting from independent assortment of bivalents formed from random pairing of both sister and homologous chromosomes. Only the transient period of developing homo- zygosity characteristic of the latter situation would be revealed as frequent failure of parent-to-offspring grafts and sibling-to-sibling grafts. In order to judge if pairing is restricted to sisters, as suggested by MacGregor and Uzzell (1964), one is left with the task of examining directly the composition of bivalents, or of inferring the degree of homo- or heterozygosity that a given parthenogenetic clone might possess. The former possibility, direct determination of bivalent composition, is smple enough in principle. The admin- istration of '^H-thymidine to a parthenogenone at the synthetic phase of mitosis immediately preceding endomitosis would result in the presence of radioactive label in one member of each sister pair arising from endomitotic duplication. Autoradiography of lampbrush bivalents would then show label in half of each bi- valent, should strict sister pairing be the rule. Random pairing of sisters and homologues would result in some bivalents un- labeled, some half labeled, and some wholly labeled. Crossing- over would not complicate interpretation. Although such an ex- periment would clearly resolve the question of bivalent composi- tion, practical problems, such as finding the proper time in the animal's life cycle for •''H-thymidine administration, make this a difficult exercise. Judgement of the degree of homo- or heterozygosity in a par- thenogenetic clone, and hence, inference of the composition of meiotic bivalents, can be based on studies of phenotypic variation. While some good studies of phenotypic variation in partheno- genones and in their sexual relatives have been performed (Zweifel, 1965), results are not easily interpreted in favor of either homo- or heterozygosity. For example, groups of C. tesselatus from a given locality were found to exhibit a range of variation in many characters that approximated half that seen in local populations of the sexual lizard, C. tigris (Zweifel, 1965). Does the relatively 12 BREVIORA No. 381 smaller variation seen in parthenogenones reflect the existence of homozygous clones, or does it indicate the importance of recombi- nation in freeing variation inherent in heterozygous genomes? Complicating this question is a fundamental ignorance of the genetic regulation of most phenotypic expression. One could resort to the doctrine of superior fitness in heterozygotes and argue that parthenogenones that compete successfully with their sexual counterparts must necessarily be heterozygous. This begs an interesting question and ignores an opportunity to test notions of fitness that have become a foundation of evolutionary theory. Furthermore, such reasoning is contradicted by Darevsky's (1966) observations on the cytology of parthenogenesis in Lacerta. Darevsky maintains that failure of second meiotic division is the parthenogenetic mechanism operative in these lizards. Under this mechanism, crossing-over will temporarily maintain some hetero- zygosity, particularly at loci far from the centromere. However, the ultimate tendency is toward complete homozygosity of the genome (Beatty, 1957). Hence, if Darevsky's observations are correct, one must expect the competitively successful Lacerta parthenogenones to exhibit a high degree of homozygosity relative to their sexual counterparts. Assuming that pre-meiotic endomitosis facilitates partheno- genesis in Cnemidophorus, a reliable indication that pairing may be strictly limted to sister chromosomes comes from studies that have deduced genotypes for certain enzymes. Parthenogenetic Ciieiuiclophonis exhibit a striking incidence of heterozygosity at genetic loci determining phenotypes for lactate dehydrogenase, adenosine deaminase, phosphogluconate dehydrogenase, and NADP-dependent malate dehydrogenase (sec Table 1 in Neaves, 1 969 ) . Most impressive are the genotypes for adenosine deami- nase; every parthenogenetic species studied showed heterozygosity at the ada locus. This is clear evidence in favor of fixed hetero- zygosity in parthenogenetic Cnemidophorus, and hence, in favor of strict sister pairing at meiosis. With the question of parthenogenetic mechanisms aired, it ap- pears that pre-meiotic endomitosis provides a basis for under- standing how a triploid, parthenogenetic C. exsanguis could pro- duce an egg whose pronucleus carried an unreduced somatic chromosome complement, and how the union of this pronucleus 1971 TETRAPLOID CNEMIDOPHORUS 13 with a haploid pronucleus from male C. inornatus could rcsull in an offspring carrying the chromosomes seen in Plates 2 and 3. A precedent for these events is established in the occurrence of such a fertile union of pronuclei from parthenogenetic females and sexual males in the genus Poeciiia (Rasch et al., 1965). There remains the question of the absence of tetraploid species of Cneniiclophoriis in spite of the existence of tetraploid hybrids in nature. Probing this question requires some indulgence in speculation, which the novelty of the subject will hopefully excuse. The most likely reason for the absence of tetraploid species may be the failure of tetraploid hybrids to reproduce parthenogeneti- cally. There is no evidence that the tetraploid lizard described in this study was parthenogenetically competent. The fact that it laid fully yolked eggs does not imply that these eggs either, 1 ) carried an unreduced chromosome complement, or 2) were capable of undergoing spontaneous embryonic development. These are two basic criteria that must be met if an interspecific hybrid is to achieve the reproductive success characteristic of existing parthenogenetic species in the genus Cneinidophorus. If one prefers to assume that tetraploid hybrids can reproduce parthenogenetically, then one must account for their absence as species on the grounds that no suitable ecological niche is avail- able to them or that they cannot successfully compete with other species for a mutually suitable niche. However, the similarity of the known tetraploid hybrid to other Cne/nidophorus suggests that it might compete with them for a currently available niche, and the behavior of the tetraploid lizard in captivity suggests that it could be successful in this regard. It seems that one must attempt instead to justify reproductive failure. One possibility is that tetraploidy is incompatible with the mechanism of parthenogenesis operative in Cneinidophorus. How- ever, the suspected mechanism, which is pre-meiotic endomitosis, has the important virtue of theoretically permitting any karyotype, regardless of ploidy, to function normally at meiosis. Given the suspected mechanism of parthenogenesis, tetraploidy itself should not be a barrier to reproduction. The most attractive possibility is that only a small proportion of interspecific hybrids meets the basic requirements of partheno- genetic reproduction. In other words, the genetically determined 14 BREVIORA No. 381 compensatory events, such as pre-meiotic endomitosis and spon- taneous embryonic development, which are presumably needed for parthenogenesis, may be frequently absent in F, hybrids. This possibility is particularly attractive, owing to the apparent genetic uniformity within existing parthenogenetic species of Cnemidophorus. MasUn ( 1967) has demonstrated a pattern of histocompatibility in C. tesselatus that suggests that all diploid members of the species, even when taken from localities hundreds of miles apart, are genetically identical. Similarly, all triploid C. tesselatus are reciprocally histocompatible, and what is more, can accept tissue grafts from the diploid ^ but cannot reciprocate. Biochemical evidence (Neave^, 1969) suggests that diploid C. tesselatus arose from interspecific hybridization between C. tigris and C septemvit- tatus, two sexual species, and that triploid C. tesselatus resulted from the addition of a haploid genome from C. sexUneatus, a third sexual species, to the diploid C. tesselatus genome. Coupled with this evidence, Maslin's (1967) findings suggest that all existing populations of C. tesselatus arose from the offspring of a single hybrid individual representing a cross between C. septeinvitiatus and C. tigris and that all triploid C. tesselatus are derived from a single hybrid lizard representing a cross between C. sexUneatus and diploid C. tesselatus. The genetic uniformity in C. tesselatus could not exist if the species contained offspring of more than one parthenogenetic hybrid, since each individual hybrid resulting from a C. septemvittatus x C. tigris cross will carry a unique recombinant genotype. The genetic uniformity of C. tesselatus points to either one or both of two possibilities, namely that interspecific hybrids are rare or that parthenogenetic competence in an interspecific hybrid is rare. The first possibility cannot be ruled out in the case of C. tesselatus, as no evidence is available that might suggest the fre- quency at which hybrids between either C. septemvittatus and C. tigris or C sexUneatus and diploid C. tesselatus occur in nature. The first possibility can be eliminated in the case of C. "perplexus," the hybrid between C. inornatus and C. neomexicanus. C. "per- plexus" was first collected in New Mexico in 1841 (Maslin et ai, 1958), and since 1962, at least six of these hybrids have been captured at sites where both C. inornatus and C. neomexicanus 1971 TETRAPIOID CNEMIDOPHORUS 15 are sympatric (Wright and Lowe, 1967a). This is a case where hybrids are rather common in nature and where they have oc- curred for at least 1 30 years without developing a parthenogenetic species. This case supports the view that genetic uniformity in existing parthenogenetic species and the absence of tetraploid species are both due to the rarity with which parthenogenetic competence is achieved in interspecific hybrids. The establishment of a parthenogenetic species in Cnemido- phonis may require a lengthy period of experimentation in which thousands or more individual hybrids are synthesized before a re- productively successful hybrid gene combination occurs. Never- theless, the result may still be termed saltatory speciation, in that the divergence of the new species from its progenitors is instantaneous, deriving as it does from a single, reproductively fit individual, rather than from cumulative changes in a population over long periods of time. It seems that both C. "perplexus" and the tetraploid hybrid illustrated here could represent previews of species that might eventually become established in the New Mexico deserts, if a gene combination facilitating parthenogenetic reproduction ultimately occurs in one of these hybrids. SUMMARY A tetraploid lizard resembling C. e.\.\anguis but bearing traits characteristic of C. itioniatus is considered to have resulted from a hybrid mating in which a haploid sperm pronucleus of C. inornatus fused with an egg pronucleus carrying the unreduced somatic chromosome complement of C. exsanguis, a triploid par- thenogenone. Production of such an egg by C. exsanguis may have relied on endomitosis in the oogonium, followed by normal meiosis operating on bivalents composed of paired sister chromo- somes. This modification of oogenesis is compatible with all known aspects of parthenogenesis in Cnemidupliorus, including the existence of apparently fixed heterozygosity within partheno- genetic species. Most cases of parthenogenesis in Cnemidophorus began with the synthesis of interspecific hybrids. However, several considera- tions suggest that many hybrids may be generated before a parthenogenetically competent individual, capable of giving rise to a species, is produced. Among these suggestive considerations 16 BREVIORA No. 381 are: 1) the synthesis of many hybrid individuals in nature, i.e., C. "perplexus," without the appearance of a corresponding species, and 2) the apparent genetic uniformity of parthenogenetic impedes, which indicates their origin from a single hybrid individual. Thus, the appearance of a new parthenogenetic species in Cnenii- dophorus may be preceded by a period of hybridization during which large numbers of reproductively incompetent prototypes of the new species are generated. ACKNOWLEDGMENTS This work was supported by Anatomical Sciences Training Grant GM 406 from the National Institutes of Health. The col- lection of animals was supported by the Evolutionary Biology Committee of Harvard University and by NSF Grant GB 3167. Drs. David Hamilton, Don Fawcett, and Ernest Williams provided generous assistance throughout the course of this study. William P. Hall offered many valuable comments on the manuscript. LITERATURE CITED Beatty, R. a. 1957. Parthenogenesis and Polyploidy in Mammalian Development. London, Cambridge University Press. CuELLAR, O. 1971. Reproduction and the mechanism of meiotic restitu- tion in the parthenogenetic lizard Cnemidophorous iiniparens. J. Morphology, 133(2): 139-165. Darevsky, I. S. 1966. Natural parthenogenesis in a polymorphic group of Caucasian rock lizards related to Lacerta sa.xicola Eversmann. J. Ohio Herp. Soc, 5: 115-152. Hall, W. P. 1970. Three probable cases of parthenogenesis in lizards (Agamidae, Chamaeleontidae, Geivkonidae). Experientia, 26: 1271- 1273. Kallman, K. D. 1962. Gynogenesis in the teleost, Mollienesia formosa (Girard), with a discussion of the detection of parthenogenesis in vertebrates by tissue transplantation. J. Genetics, 58: 7-21. Lowe, C. H., and J. W. Wright. 1966. Evolution of parthenogenetic species of Cnemidophoriis (whiptail lizards) in western North America. J. Arizona Acad. Sci., 4: 81-87. Cole, C. J., and R. L. Bezy. 1970. Natural hybridization between the teiid lizards Cnemidophoriis sonorae (parthenogenetic) and Cnemi- dophoriis tigris (bisexual). Syst. Zool., 19: 114-127. MacGregor, H. C, and T. M. Uzzell. 1964. Gynogenesis in sala- manders related to Ambystoma jeffersonianitm. Science, 143: 1043. Errata for Breviora 381 by William B. Neaves, lines 35-37, page 16: r 1 Cole and R. L. Bezy. 1970. Natural genetic) and Cnemidophorus tigris (bisexuai;. y 1971 TETRAPLOID CNEMIDOPHORUS 17 Maslin, T. p. 1967. Skin grafting in the bisexual teiid lizard Ciicmi- dopJwnis sexUiicatus and in the unisexual C. tcssrldtus. J. Hxp. Zool., 166: 137-150. , R. G. BiiDiJMAN, AND C. H. LowT. 1958. The status o'i the lizard Cucniidophonis perplcxus Baird and Girard (leiidae). Proc. U.S. Nat. Mus., 10«: 331-345. MOORHHAD, P. S.. P. C. NOWEIL, W. J. MtLLMAN. B. M. BaTTIPS, AND D. A. HuNGERFORD. 1960. Chromosome preparations of leucocytes cultured from human peripheral blood. Exptl. Cell Res.. 20: 613-616. N EAVES, W. B. 1969. Adenosine deaminase phenotypes among sexual and parthenogenetic lizards in the genus Cnemidoplioriis (Teiida;). J. Exp. Zool.. 171: 175-184. , AND P. S. Gerald. 1968. Lactate dehydrogenase isozymes in parthenogenetic teiid lizards {Cnciuidophonis) . Science. 160: 1004-1005. Rasch, E. M., R. M. Darnell, K. D. Kallman, and P. Abramoff. 1965. Cytophotometric evidence for triploidy in hybrids of the gynogenetic fish. Puccilia fonuosa. J. Exp. Zool., 160: 155-170. ScHULTZ, R. J. 1967. Gynogenesis and triploidy in the viviparous fish Poeciliopsis. Science, 157: 1564-1567. Taylor, H. L., and P. A. Medica. 1966. Natural hybridization of the bisexual teiid lizard Cnemidophunts inornatus and the unisexual Cnemidophorus perple.xiis in southern New Mexico. Univ. Colorado Studies, ser. biol., 22: 1-9. Vanzolini, p. E. 1970. Unisexual Cnenudophonis Iciuniscatiis in the Amazonas Valley: a preliminary note (Sauria, Teiidae). Papeis Avulsos Zool., S. Paulo. 23: 63-68. Wright, J. VV., AND C. H. Lowe. 1967a. Hybridization in nature be- tween parthenogenetic and bisexual species of the whiptail lizards (genus CneiuidopJwnis). American Mus. Novitates, No. 2286: 1-36. , AND 1967b. Evolution of the alloploid par- thenospecies Cncniidophonis tesselatus (Say). Mamm. Chrom. Newslett., 8: 95-96. Zweifel, R. G. 1965. Variation in and distribution of the unisexual lizard Chemidcphoriis tesselaliis. American Mus. Novitates, No. 2235: 1 49. Table 1 chromosome number 87 88 89 90 91 92 frequency 2 0 113 1 Table 1. Eight apparently complete chromosome spreads from MCZ 101991, the tetraploid hybrid Cnemidophorus. were studied. The fre- quency with which various chromosome numbers were observed in these spreads is indicated in this table. 18 BREVIORA No. 381 1971 TETRAPLOID CNEMIDOPHORUS 19 Plate I Figure 1. A tetraploid hybrid (MCZ 101991) is shown between its suspected parental species. C. inonialiis (IN) and C. exsaiii^iiis (EX). Unlike typical C. exsaiti^iiis, the hybrid has fused paravertebral stripes and blue on its tail and ventral surfaces. C. inoruatiis also has vivid blue on these surfaces. 971 TFTRAPIOID CNI-MIDC^PUORUS 21 Plate II Figure 2. An apparently complete set of 92 chromosomes from the tetraploid hybrid (MCZ 101991) is shown. The technique for obtaining chromosome spreads is described in the text. 22 BREVIORA No. 381 oo f i I I I IT i I t I t 1 € I CO ■c ^ fcO '^'«» t:^ "^^ 971 TETRAPI.Oll) CNEMIDOPHORUS 23 4-" CD 13 ^ c a -» OJ o *». (U C x> V5 !^ V5 C (D J3 c CJ G (U E > rt <4-i 0) j: O "H, V5 c E E o E o n JD •a o E o D. ^ E "5. . ;. o CS c" j: o x: o CJ u j: ."^ 1> r- u V) sz C o H o i; D. c o E ^c (U & o vo E o u. ON o ri •o x: C (U rl ■_o £/J o tjj a. c £ J= " ,-« ^ a; C 1) N . ™ x: ^ c/: v5 £/j ^4- o "O LJU o CJ J2 24 BREVIORA No. 381 f t . / 8 '..**•. •*«/.?> V -■»*;&■' 9 1971 TETRAPLOID CNEMIDOPHORUS 25 Plate IV Figure 4. Courtship behavior between two C. inornutus. The male strokes the subject's neck and head with his jaw. Figure 5. Courtship behavior between a male C. inoniatiis and a parthenogenetic C. e.xsangiiis. The male executes sinuous movements against the dorsum of the female. Figure 6. As in Fig. 5. The male retreats along the female's dorsum, still executing sinuous movements while approaching the copuiatory position. Figure 7. As in Fig. 5. The male wedges his tail between the female's hind limbs and tail and attempts copulation. Figure 8. Courtship behavior between a male C. inornatus and a parthenogenetic C lesselatus. The male strokes the female's neck and head with his jaw. Figure 9. As in Fig. 8. The male executes sinuous movements against the dorsum of the female. >*. ■ ^t ^^ f^l^'S. CGMP. ZOOL UC.r.ARY B R E V I O "ft™l Muiseiiiim of Comparsitive Zoology Cambridge, Mass. 25 February, 1972 Number 382 OSTEOLOGY AND DESCRIPTION OF THRATTIDION NOCTIVAGUS, A MINUTE, NEW FRESHWATER CLUPEID FISH FROM CAMEROON, WITH A DISCUSSION OF PELLONULIN RELATIONSHIPS Tyson R. Roberts^ Abstract. Thrattidion iwctivagiis. a new genus and species of Pellonu- linae, is described from the Sanaga River in Cameroon, and an account given of its osteology. Although it agrees with Congothrissa in having no supramaxillary bone, the two appear not to be closely related. The ecology of African freshwater Pellonulinae is reviewed. Feeding habits range from planktophagous to insectivorous and piscivorous, with piscivores feeding largely on other species of Pellonulinae. Most or all of the species undergo diel migrations, moving upwards or shorewards at nightfall in mixed ag- gregations with other small fishes, chiefly translucent schilbeid catfishes and cyprinids of the subfamily Bariliinae. Sierrathrissa Audenaerde (1969) is evidently the larva of either Pellonula or Cynothrissa. Congothrissa Poll (1964) is referred to the Pellonulinae. Relationships of round herrings (Dussumieriidae) that seem close to Pellonulinae are also discussed; Laevisciitella should be placed in the Pellonulinae. INTRODUCTION During fieldwork on the Sanaga River in Cameroon, I collected a minute clupeid which is clearly a member of the PellonuHnac but defies placement in any described genus. In lacking a supra- maxillary bone in the upper jaw it difl:ers from all other known clupeoids excepting Congothrissa (Poll, 1964) from the Congo basin. Consequently the distinctions between Congothrissa and 1 Museum of Comparative Zoology, Cambridge, Mass. 02138 2 BREVIORA No. 382 other clupeids arc not so great as they at first appeared, and it may now be referred with some confidence to the PellonuHnae. Nevertheless these two minute clupeids are not more closely related to each other than to other forms, and their relationships within the PellonuHnae are unclear. The description of this new form from Cameroon is accompanied by an account of its osteology in the hope that this will contribute to its eventual placement in a phylctic classification. Acknowledgments. I wish to thank Mr. Thomason Newcomb for providing a congenial lodging place during my stay at Edea, and Prof. George S. Myers for reading the manuscript. The field- work was assisted financially by the Office of Environmental Sciences (contract no. RC 717634). THRATTIDION, new genus Type species. Thrattidion noctivagus, new species. Diagnosis. Minute (largest specimen 21.4 mm), largely trans- lucent Pellonulinae whh direct development. Four to seven keel- less prepelvic abdominal scutes and three to five (usually four) strongly keeled postpelvic abdominal scutes. All scutes with well- developed lateral (ascending) arms; pelvic scute with bifurcate or trifurcate lateral arms in larger specimens, Supramaxillary bones absent. Maxillary toothless. Branchi- ostegal rays 5,5 or 5,4. Lower limb of first gill arch with 10-11 and upper limb with four short, widely spaced gill rakers. Anal fin considerably longer than dorsal, with 23-25 rays. Origin of dorsal fin about one-third of distance between verticals through pelvic and anal fin origins. Origin of anal fin on a vertical through base of eighth or ninth dorsal fin ray. Pectoral rays usually ten. Squamation restricted to a patch of about a dozen scales just behind head, a double row of scales on either side of prepelvic abdominal scutes, and a single row of scales on either side of postpelvic abdominal scutes. Vertebrae 45-47, about equally divided between abdominal and caudal (preural centrum 1 counted as last vertebra). Epurals three. Remarks. If the largest specimens in my material are fully grown, Thrattidion is the smallest clupcoid yet described. Next smallest is Congothrissa, with specimens recorded up to 27.7 mm 1972 AFRICAN CLUPEID FISH 3 (here and throughout this paper lengths referred to are standard lengths). Almost all African freshwater clupcids have nine or more prepelvic scutes and at least seven postpelvic scutes; in these forms the prepelvic as well as postpelvic scutes are invariably serrated or keeled. The only exceptions from freshwater in Africa are 1 ) Congothrissa, which lacks abdominal scutes entirely except for a pelvic scute; 2) Laevisciitella, with seven to eight keel-less prepelvic scutes and five to six keeled postpelvic scutes; and 3) Gilchristella, with six to nine unkeeled prepelvic scutes and no postpelvic scutes. The postpelvic scutes in Laevisciitella are unique in lacking lateral arms as adults (lateral arms are, how- ever, absent in the postpelvic scutes of larval Pelloniila). The lateral arms of the pelvic scutes are usually simple; they are bifurcate in Congothrissa (Poll, 1964, fig. 5 on p. 14). All previously described clupeoids have one or two supra- maxillary bones except Congothrissa. Pellonulinae characteris- tically have only one supramaxillary bone (Regan, 1917); its presence has been noted in alevins of Poecilothrissa moeruensis, Stolothrissa tanganicae and Microthrissa stappersii so small that the abdominal scutes are as yet unformed (Poll, 1964: 19). The supramaxillary is reduced in size in some Pellonulinae. The maxillary bone bears teeth in all other African freshwater clupeoids except Congothrissa and Potaniothrissa. Most African freshwater clupeids have 6,6 or more branchiostegal rays; Poecilothrissa, Microthrissa and Potamothrissa have as few as five; Congothrissa has 3,4 (here as elsewhere in this paper, the count on the left side is given first). All other African clupeoids have 16 or more gill rakers on the first gill arch, excepting Congothrissa, which has only seven on the lower limb and four on the upper limb of the first arch. All other African Pellonulinae have 23 or fewer anal rays, except Microthrissa, which has up to 25. Congothrissa has only 15. In Congothrissa the dorsal fin origin is distinctly anterior to a vertical through origin of pelvic fins. In almost all African fresh- water clupeids the anal fin origin is far behind a vertical through base of last ray of anal fin, the exceptions being larval Pelloniila, in the metamorphosis of which the dorsal fin migrates forward, and Microthrissa and Gilchristella, in which the anal fin origin is on or only slightly posterior to a vertical through the last ray or BREVIORA No. 382 two of the dorsal fin. In larval Pellonula the anal fin origin may be as far forward as posterior third of dorsal fin. All other African freshwater clupeoids have 1 1 or more pectoral rays. In all other African clupeoids the body is completely scaled in adults; the highest vertebral count previously recorded is 44, with abdominal vertebrae usually more numerous than caudal (Poll, 1964, table 1). Congothrissa has 21+20 vertebrae. The number of epurals is unrecorded for most Pellonulinae; Congothrissa and larval Pellonula have two epurals. Etymology. Thrattidion, Greek, neuter diminutive of Thrassa (Thratta), feminine, a small, herringlike fish. Thrattidion noctivagus, new species Figures 1 and 2; Table 1 Holotype. MCZ 48161, 21.4 mm, Sanaga River at Edea, Cameroon, 4 May 1971. Paratypes. MCZ 48162, 175 specimens, 9.8-20.7 mm (of which 11 specimens 10.4-20.7 mm cleared and stained), Sanaga River at Edea, 17 April and 4 May 1971. The specimens are in excellent condition. They were fixed in formahn immediately after capture and placed in Nalgene plastic bottles containing only small specimens and filled to the brim to prevent damage by shaking. Consequently, the fins are in perfect condition on almost all specimens, and the squamation completely intact. After three months in formalin they were soaked in water for two days, in 40 percent ethyl alcohol for two more days, then transferred to 70 percent ethyl alcohol for permanent preservation. General characteristics. Snout short and rounded. Jaws equal. Body relatively compressed, depth 3.55-5.0, increasing with size. Figure 1. Thrattidion noctivagus, 19.4-mm paratype. MCZ 48162. 1972 AFRICAN CLUPEID FISH 5 Eye large, 8.5-9.9, its center on or only slightly above longitudinal midline of body (center of eye distinctly above midline in most Pellonulinae, markedly so in OdaxotJirissa and Cynothrissa) . Adipose eyelid well developed. Dorsal fm origin distinctly pos- terior to a vertical through pelvic origin, slightly closer to end of hypural fin than to snout tip. Anal fin long, its origin on a vertical through base of eighth or ninth dorsal fin ray. Pelvic, dorsal, and anal origins on verticals through 15th, 19th, and 25th vertebrae. Pelvic fin relatively large, only slightly smaller than pectoral, its origin closer to pectoral fin origin than to anal fin origin. Dorsal and anal fin margins gently falcate, dorsal height about 1.5 times anal height. Caudal fin deeply forked, upper and lower caudal lobes identical in size and shape. For proportional measurements see Table 1. Fin counts (Table 1). Dorsal fin with 14-16 rays, first three or four simple, last divided to base. Anal fin with 23-25 rays, first four or five simple, last divided to base. Length of first dorsal and anal rays variable, minute in some specimens. Pectoral fins usually with ten rays. Pelvic fins with either seven or eight rays. Principal caudal rays invariably 10+9, upper procurrent rays 12-13 and lower procurrent rays eight to ten. Squamation (Fig. 1). Scales largely absent, restricted to a double row of four to nine scales superficial to lateral arms on each side of prepelvic scutes, a single row of four or five scales superficial to lateral arms on either side of postpelvic scutes, and about a dozen scales in four or five short rows immediately behind head. The latter scales are deeply embedded in tough tissue in which the trunk portion of the cephalic laterosensory system ramifies, and can only be fully discerned in cleared and stained specimens. A 20.7-mm specimen has 12 scales in four rows (three in the uppermost row, five in the second row, one in the third row, and three in the lowermost row); a 19.1-mm specimen 11 scales in five rows (2-4-1-3-1); and an 18.6-mm specimen 11 scales in five rows (1-2-4-2-2). Most of the scales in the longest row are perforated by laterosensory canals. Larval Pellonula of 36-38 mm have the body naked except for two or three scales immediately behind the head and complete squamation on the caudal peduncle. BREVIORA No. 382 •11 -^ o V ~ d r^ ** "* i -r ;. f -r CT- O- O M *n H CD « in ^ o- o o- B - * ~o to -o |l S s t ; 1972 AFRICAN CLUPEID FISH Cephalic laterosensory system (Fig. 2). Cephalic latcrosensory system with supraorbital, infraorbital, mandibulo-preopcrcular, pterotic, opercular, occipital and lateral canals (cf. Wohlfahrt, 1937). A supratemporal branch arises from the supraorbital; in some specimens it appears to enter the occipital canal, but in others it definitely stops short of reaching the occipital canal. Figure 2 is based on a 15.1 -mm specimen that had a drop or two of Ehrlich's hematoxyUn placed on its head. Of several specimens thus treated, its canals showed up the best. The mandibular portion of the mandibulo-preopercular canal could not be dis- tinguished on this specimen, but it appeared in others. The main canals appear to join in the manner illustrated, but the figure should be used with caution. The canals did not take up stain evenly, and comparisons with less well-stained specimens did not ehminate my reservations about the interpretation. I could find no connections between the laterosensory systems of the two sides of the head. In particular the ethmoidal commissure is absent. Dentition. Premaxillary with four to seven small conical teeth soUdly implanted, and two to four replacement teeth in the gum. supraorbital supratemporal occipital lateral infraorbital preopercular opercular Figure 2. Thrattidion noctivagiis, cephalic laterosensory canals of 15.1 -mm paratype. 8 BREVIORA No. 382 Maxillary edentulous. Dentary with a dozen to 15 small conical teeth solidly implanted and several teeth in stages of replacement. Palatine with one to three teeth, roof of mouth otherwise edentu- lous. Fifth ceratobranchial and upper pharyngeal with numerous small teeth, gill arches otherwise edentulous. Abdominal scutes. Four to seven prepelvic scutes, unkeeled, with well-developed lateral arms. Pelvic scute keel-less, with bifurcate or trifurcate lateral arms in largest specimens. Post- pelvic scutes three to five, usually four, with well-developed keels and lateral arms. Specimens as small as 12-14 mm with full complement of scutes (Table 1). Coloration of preserved specimens. Body opaque white, with- out mclanophores, or with only a few melanophores in some specimens at base of anal fin rays and on caudal peduncle. No lineum argenteum. Peritoneum probably opaque white and body otherwise translucent in living specimens. A dense cluster of large melanophores over brain. Specimens were collected only at dusk or at dawn and immediately preserved, so observations were not made on their appearance in life. Additional characteristics. See generic diagnosis and oste- ological account. OSTEOLOGY Neurocranium (Figs. 3-5). Anterior and posterior frontal fontancUes very large. Anterior margin of anterior frontal fonta- nelle formed by diverging posterior arms of mesethmoid. Pos- terior frontal fontanelle incompletely divided into right and left halves, the median extension of supraoccipital failing to reach posteromedial border of frontal bones. Of the types of clupeoid posterior frontal fontanelles figured by Whitehead, those of Spratelloides and Ehirava (Whitehead, 1963a, fig. 1) resemble most closely that of Thrattidion. As in Ehirava, a median exten- sion of the frontals forms a distinct wedge in the posterior frontal fontanelle. Frontals slightly separated at midline of skull. Anterior fontanelle about half as wide as posterior fontanelle. Mesethmoid solidly ossified, with well-developed anterolateral and descending processes, and diverging posterior processes firmly united to anterior ends of frontal bones. Lateral ethmoid with a large cartilaginous portion ventrally. 1972 AFRICAN CLUPEID FISH Pre-epiotic fossa large, an anteroventrally directed process of parietal bone only partially separating it from temporal foramen. Pre-epiotic fossa bordered mainly by frontal and parietal; temporal foramen bordered by parietal, cpiotic and pterotic (Fig. 3). orbitosphenoid mesethmoid \ frontal supraoc :ipital preiuaxillary \ ^^^^r^"! '•■'•'. •;-/Aw«;v =- pterotic maxillary / /Ntv vomer / / ^^^j-^ //iw V basioccipital cartilage / /~'^^^<>--~/ /i%i^^^ \ \ ^ 1, '^.j. ■ - \ V ■ \^ — V" exoccipital lateral ethmoid / / / ~\ V parasphenoid / / sphenotic ^ sagitta pterosphenoid basisphenoid prootic 1 1 mm 1 Figure 3. Cranium of 18.6-mm Thrattidion noctivagiis, lateral view. parasphenoid maxillary premaxillary dentary lateral ethmoid entopterygoid frontal parietal supraoccipital exoccipital basioccipital mesethmoid nasal palatine \ \ ^'^'^--i'-'^—V' epiotic supraorbital orbitosphenoid pterotic ( ^ '""^ I Figure 4. Cranium of 20.0-mm Thrattidion noctivagiis, dorsal view. 10 BREVIORA No. 382 A large foramen bordered by the frontal, pterotic, and prootic opens into the recessus lateralis (Fig. 3). It seems likely that all of the main branches of the cephalic laterosensory system enter the recessus through this one large opening. A laterally directed ridge on prootic extending onto anteroventral portion of pterotic, both bones apparendy contributing to this ridge posteriorly. Intcrcalar bone present, a small, oval bone lying in the interos- seous membrane between exoccipital, pterotic, and prootic, its edges touching each of these bones (Fig. 5). When the pectoral girdle is removed from the skull, the intercalar is invariably carried away with the lower limb of the posttemporal bone. I am unable to find a passageway through which anterior exten- sions of the swim bladder might gain the prootic bulla. A small oval opening in the exoccipital bone opens into a passageway directed dorsoposteriorly, away from the prootic. This opening and its passageway do not have counterparts in the prootic. The opening is flush with the exoccipital, not funnel-shaped as one would expect of an opening for the swim bladder; it probably serves for passage of cranial nerves. A large dorsal foramen of the prootic is readily visible. parasphenoid pterosphenoid orbitosphenoid \ i , . , \ basisphenoid lateral ethmoid vomer cartilage maxillary intercalar sagitta premaxillary nasal mesethmoid supraorbital frontal sphenotic basioccipital exoccipital pterotic prootic 1 mm Figure 5. Cranium of 18.6-mm Thratddion noctivagus, ventral view. 1972 AFRICAN CLUPEID FISH 11 Orbitosphenoid, pterosphenoids, and basisphenoid well de- veloped, with large lateral foramina. Orbitosphenoid with an anteriorly directed process. Vomer edentulous, elongate, and flattened. Parasphenoid uni- formly slender, dividing where it meets base of neurocranium into a pair of posterior processes, the tips of which extend beyond occiput to immediately below second or third vertebral centrum (Fig. 5). Jaws (Figs. 6-7). Premaxillaries separated at midline, each bearing four to seven conical teeth. Maxillary relatively slender, extending to below anterior third of eye, toothless. Supra- maxillaries absent. Dentary with a low coronoid process and a single row of about a dozen to 15 conical teeth. Articular with a posteriorly directed process posterior to quadrate socket. Siispensoriiim (Fig. 7). Palatine bifid at its tip, bearing one to three conical teeth. Suspensorium otherwise edentulous. Ectopterygoid slender anteriorly, moderately expanded where it supraorbital dermosphenotic antorbital opercle premaxillary dentary maxillary subopercle interopercle articular' quadrate symplectic preopercle I ^ ""° 1 Figure 6. Jaws and facial bones of 20.7-mm Thratddion noctivagus, lateral view (1-5 = circumorbital bones). 12 BREVIORA No. 382 premaxillary maxillary palatine opercle dentary coronomeckelian articular angular quadrate j ijiterhyal symplectic subopercle preopercle interopercle 1 mm Figure 7. Jaws, suspensorium, and opercular bones of 20.7-mm Thrat- tidion noctivagus, medial view. attaches to dorsoanterior portion of quadrate bone. Entoptery- goid and metapterygoid well developed. Sphenotic and pterotic articular surfaces of hyomandibular not strongly differentiated from its main body. Hyomandibular with a distinctively shaped lamella dorsoanterior to its symplectic process, the ventral margin of this lamina separated from symplectic process of hyomandibular by a notch. Symplectic bone relatively small, lying midway be- tween hyomandibular and quadrate but in contact with neither. Quadrate with a posterior process directed towards anterior end of symplectic. Facial bones (Fig. 6). Circumorbital series with an antorbital followed by six infraorbitals, a pattern found in generalized mem- bers of many teleostean groups, including Clupeomorpha (Nelson, 1969). Dermosphenotic greatly reduced. Infraorbitals shallow, without extensive ventral laminae. Supraorbital bone large and elongate. Nasal bones widely separated, moderate in size, bearing a tube for laterosensory canal. 1972 AFRICAN CLUPEID FISH 13 dorsal hypohyal ceratohyal epihyal ventral hypohyal branchiostegal rays 1-5 1 mm Figure 8. A. Urohyal bone (upper: dorsal view; lower: left lateral view); and B. hyoid arch (basihyal and interhyal removed), lateral view, of 20.7-mm Thraitidion noctivagus. ventral hypohyal dorsal hypohyal , ceratohyal epihyal interhyal pharyngobranchials 1-3 upper pharyngeal basihyal basibranchials 1-3 hypobranchials 1-3 ceratobranchials 1-5 epibranchials 1-4 1 mm Figure 9. Hyoid arch (branchiostegal rays removed) and branchial arches of left side, of 20.7-mm Thrattidion noctivagus. Dorsal view, upper limb of gill arches unfolded and pinned to the side. 14 BREVIORA No. 382 Opercular series complete; dorsoanterior corner and posterior margin of opcrcle deeply incised. Hyoid and branchial arches (Figs. 8-9). The branchial skele- ton of Thrattidion differs from that in most clupeids in several respects; this is perhaps mainly due to its reduced size. Gill rakers and pharyngeal dentition reduced. Branchial skeleton lacking elongation of individual parts typical of most living clupeoids (and which is at least partly correlated with increased numbers of gill rakers and branchiostegal rays). Basihyal well ossified and tooth- less (usually cartilaginous and tooth-bearing in Pellonulinae, Nelson, 1967: 393). Pharyngeal teeth restricted to upper pharyngeal and ceratobranchial 5. Gill rakers few in number, widely spaced, and relatively short. The 20.7-mm specimen rep- resented in Figure 9 has 15 rakers on the first gill arch, including a few that are rudimentary or unossified. Gill rakers progressively reduced in size and number on arches 2-5. The hyoid arch has paired ventral and dorsal hypohyals, ceratohyal, epihyal and interhyal, and a median basihyal. Branchiostegal rays 5,5 in eight specimens, 5,4 in two; generally one branchiostegal ray attached to epihyal, three to ceratohyal, and one branchiostegal ray in between, not attached to either bone. Individual ossified parts of the branchial arches are median basibranchials 1-3, and paired hypobranchials 1-3, cerato- branchials 1-5, epibranchials 1-3, suspensory pharyngeals or infrapharyngeals 1-3, and upper pharyngeals. Pectoral girdle (Fig. 10). Two slender postcleithra, distal end of second postcleithrum attached by ligament to distal end of first rib. Four proximal radials and about six distal radials (distal radials absent in clupeoids other than Chirocentridae, according to Whitehead, 1963a and Greenwood, 1968). Other bones in pectoral girdle: extrascapular, posttemporal, cleithrum, coracoid, scapula, mesocoracoid. Pelvic girdle (Fig. 11). Pelvic bones relatively elongate and slender. Articular portion of pelvic bone with a deep, elongate notch. Three pelvic radials. Pelvic splint present (pelvic splint absent in clupeoids, according to Whitehead, 1963a and Green- wood, 1968). 1972 AFRICAN CLUPEID FISH 15 posttemporal supracleithrum postcleithra 1-2 \ cleilhrum intercalar extraseapular distal radials proximal radials mesocoracold curacoid Figure lU. Pectoral girdle of 2U.7-mm Thratiidion noctivagus, medial view. Inset, upper right hand corner: extraseapular bone of same speci- men, lateral view. pelvic splint pelvic radials pelvic bone Figure 11. Pelvic girdle of 18.6 mm Thraiiidion noctivagus, dorsal view. Abdominal scutes (Fig. 12). Prepelvic scutes four to seven, keel-less, with slender lateral arms. Pelvic scute enlarged, with bifurcate or trifurcate lateral arms in larger specimens (specimen figured has two branches on one side, three on the other). Post- pelvic scutes usually four, invariably with well-developed keels. Specimens only 12.1 and 14.3 mm have a full complement of pelvic scutes (Table 1). In a 10.4-mm specimen only the pelvic scute is visible. In a 12.1 -mm specimen the scutes are repre- sented only by the lateral arms, which ossify separately. With 16 BREVIORA No. 382 growth they fuse to form a median scute. In the middle of the scutes in larger specimens is sometimes a foramen or gap in the ossification where the two halves are incompletely fused, as in the second three prepelvic scutes in the 20.7-mm specimen figured. Vertebral column and associated bones (Fig. 13). In a series of ten specimens the vertebral count ranges from 45 to 47; abdominal and caudal vertebrae about equal in number. The counts and their frequencies are 22+23 (1); 23+22 (3); 23+23 (3); 23+24 (2); and 24+23 (1). The third vertebra is the first to bear ribs. The tips of ribs 4-7 to 2-9 are bound by connective tissue to the lateral arms of the prepelvic scutes; rib 9 or 10 to the pelvic scute, and ribs 10 or 11 through 15 to the postpelvic scutes. pelvic bone IJelvlc splint postpelvic scutes prepelvic acutos pelvic scute Figure 12. Abdominal scutes of 20.7-mm Thrailiclion noctivagus, ventral view. There are four distinct sets of intermuscular bones: 1) epaxials or epincurals, 2) epicentrals, 3) anterior hypaxials or epipleurals, and 4) posterior hypaxials. The epaxials begin at about the eighth vertebra and continue to the second preural centrum. The epicentrals begin at the first vertebra and continue to the second preural centrum. The epicentrals begin at the first vertebra and extend to about the first caudal vertebra. The anterior hypaxials or epipleurals begin at the second vertebra and extend to the last abdominal vertebra. The posterior hypaxials begin at about the fourth or fifth caudal vertebra and extend to about the second preural centrum. All of the intermuscular bones are relatively simple; in particular the epicentrals, epaxials, and posterior 1972 AFRICAN CLUPEID FISH 17 Figure 13. Axial skeleton of 18.6-mm Thrattidion noctivagits, lateral view. hypaxials are unbranched. The anterior hypaxials are almost as long as the ribs and are expanded near their proximal ends into a triangular lamella. There are ten to twelve supraneural bones. The first supra- neural lies just dorsal to the first neural spine; the last lies over the tenth to twelfth neural spine. Caudal skeleton (Fig. 14). Caudal skeleton with six hypurals, hypural 1 separated from ural centrum 1 by a distinct gap. Preural centrum 1, ural centrum 1 and ural centrum 2 fused. epurals 1-3 urostyle parhypural Figure 14. Caudal skeleton of 20.7-mm Thrattidion noctivagits, lateral view (H^-H^ = hypurals; P^ + U^ + U., = fused preural centrum 1, ural centrum 1 and ural centrum 2; P., = preural centrum 2). 18 BREVIORA No. 382 Hypural 2 fused with ural centrum 1. Parhypural (= hemal spine of preural centrum 1 ) closely articulated with centrum of preural centrum 1, its dorsal margin with a very large, laterally directed flange or hypurapophysis. The single uroneural Ues free against the ural spine and does not extend anteriorly to the ural centrum 1. Epurals three. Caudal fin lobes almost perfectly symmetrical. Principal caudal rays invariably ten in upper lobe and nine in lower lobe. Speci- mens 14 mm upwards have 12 or 13 upper and eight to ten lower procurrent rays, all of which take up alizarin stain. Hypurals 1 and 3 expanded, hypural 2 slender. Hypural 1 articulating with six principal rays, hypural 2 with two, and hypural 3 with five. DISCUSSION Ecology oj African Pellonulimie. Relatively little has been published on the ecology of African Pellonulinae. Poll (1953) and Coulter (1962) gave some information about Limnothrissa and Stolothrissa, the two genera endemic to Lake Tanganyika (Tanganyika is the only East African lake with endemic clupeids). Reynolds (1970; 1971) reported on feeding habits and schoohng and migration of Pellonula and Cynothrissa in man-made Volta Lake in Ghana. The ecology of Pellonulinae in the Congo basin is unknown, excepting brief remarks on food or general habitat by Gosse (1963) and Matthes (1964). Stolothrissa feeds on zooplankton. Juvenile Limnothrissa feed on zooplankton, but adults are macrophagous and evidently feed to some extent on larval Stolothrissa. Stolothrissa are apparently the only food of the endemic centropomid Luciolates (Poll, 1953). Some riverine Pellonulinae are insectivorous to a greater or lesser extent. Mature Povilla (Ephemeroptera) nymphs were the dominant item in the food of Pellonula in Volta Lake, with terrestrial insects and aquatic stages of Diptcra contributing sig- nificantly to the rest of the food taken by them (Reynolds, 1970, table 1 on p. 578). Of the Congo species, Potamothrissa obtusi- rostris feeds largely on aquatic insects, Nannothrissa parva on phytoplankton (unicellular algae, diatoms) and zooplankton (entomostracans), and Odaxothrissa losera on shrimps, Povilla nymphs, and small fish (mainly other clupeids) (Matthes, 1964). 1972 AFRICAN CLUPEID FISH 19 The West African Cynothrissa (which superficially resemble^ Odaxothrissa) is piscivorous, feeding largely on Pellonula in Volta Lake (Reynolds, 1970). Reynolds (1971) found that Pellonula was among several small species in Volta Lake that migrate upwards land shore wards?] at night in mixed aggregations, and compared this migration to that of Stolothrissa and Limnothrissa in Lake Tanganyika. It should be noted that a similar phenomenon occurs in the Pellonu- linac of the Congo basin. Poecilotlirissa, Microthrissa royauxi, Potamothrissa, and Odaxothrissa evidently migrate upwards or shorewards at dusk in aggregation with other small fishes, prin- cipally the cyprinids Chelaethiops, Engraulicypris and Leptocypris, and the schilbeids Parailia and Pareutropiiis. These aggregations apparently do not remain close to the shore all night long, but reform there for another short period around dawn. Reynolds (1971) discusses the selective advantages of such diel migrations, which are evidently primarily improved feeding opportunities, coupled with predator avoidance and possibly a metabolic ad- vantage in living at lower temperatures and feeding at higher ones. Thratiidion was collected only at dusk and dawn, by dipnetting in water two to three feet deep flowing swiftly along the low retaining wall that extends for about a mile above the hydroelectric dam on the Sanaga River at Edea. The only other fishes caught with it were small characids, tentatively identified as juvenile Alestes and Virilia. Its stomach contents included much un- identified material and some terrestrial insects (either Diptera or Hymenoptera) (kindly identified by Dr. David Rentz). The larval stage of Pellonula. The only other species of Clupeidae known or likely to occur in the Sanaga River much above its mouth are Pellonula and Cynothrissa. Although adults of these genera can be distinguished, they evidently are closely related. Distinguishing characters for the juveniles have not been reported. Svensson (1933) described larvae of Pellonula from the Gambia River. I have collected similar juveniles in the Tano River and in the Sanaga River along with adults of Pellonula, and am inclined to believe that they are the same species. They are radically different from Thrattidion. The larvae and juveniles of Cynothrissa are yet to be described; presumably they are similar to those of Pellonula. There can be little doubt that Sierrathrissa 20 BREVioRA No. 382 (Audcnaerde, 1969), from Sierra Leone, is based on the larvae of either Pellonida or Cynothrissa. My specimens of larvae from the Tano and Sanaga rivers agree in almost all respects with the figure and description of Sierrathrissa except that they have more vertebrae (24+17 instead of 19+17) and are incompletely scaled (scales present only immediately behind head and on caudal peduncle). Audenaerdc reported "approximately 37-38 scales in a longitudinal line, about one scale row for each body segment, scales hardly visible on posterior part of body in largest speci- mens." Svensson (1933: 47-48, fig. 16) describes the larva of Pellonida (identified as P. vorax) and its metamorphosis. His description of the larva agrees very well with that of Sierrathrissa. Concerning the metamorphosis, we may quote Svensson: When the fry is about 30 mm. long, an obvious and rapid change takes place in its appearance. The transparency disappears, and pigment com- mences to develop on the back; the head and the anterior part of the body grows thicker and higher, and the dorsal fin advances forwards so that its front part becomes situated above, or slightly behind, the vertical of the origin of the ventral fins. The gill-rakers increase in number to 25 (they are 25-30 in the adult fish, according to Boulenger) and the num- ber of rays in the pectoral and ventral fins, as well as in the branchiostegal membrane, increases to that in the adult. It may be presumed that those changes represent a kind of metamorphosis, corresponding to that in other Cliipeids, and in Elopids, and Albiilids. Svensson found that larval Pellonida have nine to ten pectoral rays, seven pelvic rays, rudimentary abdominal scutes, about 14 gill rakers on lower part of first arch, and only three branchi- ostegal rays visible (in specimens stained with alizarin). He did not mention the presence of scales. Adult Pellonida have 14 pectoral rays, eight pelvic rays, 12-14 prepelvic and eight to nine postpelvic abdominal scutes, all strongly serrate and with lateral arms, 27-32 gill rakers on the first gill arch, and six branchiostegal rays. Despite statements to the contrary (Ridewood, 1904; Poll, 1964, table 1 opp. p. 29), adult as well as larval Pellonida bear numerous fine teeth on the maxillary bone (Regan, 1917; personal observation). The supramaxillary bone is well developed in both adults and larvae. Concerning other African Pellonulinae, Limnothrissa and Stolo- thrissa undergo profound changes in morphology from eight or ten to 25 mm, including considerable forward migration of the 1972 AFRICAN CLUPEID FISH 21 dorsal fin (Poll, 1953). Microthrissa miri, from the Niger and Chad basins, has a slender larva in the metamorphosis of which the dorsal fin position changes relatively little (Daget, 1954: 66-67, fig. 12; Blache, 1964: 58-59, fig. 21 on p. 418). Development is direct or nearly direct in at least some riverine Pellonulinae of the Congo basin. Thrattidion has direct develop- ment: specimens as small as 12 and 14 mm have the main features of the larger specimens, including full or nearly full complements of abdominal scutes, scales, fin rays, branchiostegal rays, gill rakers, and teeth. Fin placement remains unchanged. The most notable change is a regular increase in depth of body from 4.7-5.0 in 10-1 1-mm specimens to 3.55-4.1 in 19-21-mm specimens (Table 1), at which size Thrattidion is deeper bodied than any other African freshwater clupeid. Systematic status of Congothrissa. In erecting a new family for Congothrissa, Poll (1964) stressed its 1) lack of prepclvic and postpclvic abdominal scutes; 2) unkeeled pelvic scute [with branched lateral arms]; 3) terminal mouth; 4) lack of supra- maxillary; 5) short lateral [medial?] branch of supraorbital canal; 6) relatively large scales; and 7) low branchiostegal ray count. Characters 1,4, 6, and 7 are loss or reduction characters such as one commonly finds in teleosts of greatly reduced size. Reliance on such characters leads to classifications in which the smallest representatives of a group are placed in either monotypic or poly- phyletic taxonomic categories. Reduction of scutes has occurred in many lines of clupeoids. Thrattidion has also lost the supra- maxillary. The branchiostegal ray count of Thrattidion (5,5 or 5,4) is intermediate between that of Congothrissa (3,4) and other Pellonulinae. Congothrissa has about 24 scales in a lateral series (Poll, 1964, table 1; fig. 1 opp. p. 8). Judging from the figure of Congothrissa, it may have 25 scales in a lateral series, since there are usually deeply embedded scales immediately behind the head that are not externally visible in Pellonulinae. Some speci- mens of Poecilothrissa have as few as 28 scales in a lateral series (personal observation). An unkeeled pelvic scute with branched lateral arms similar to that in Congothrissa is found in Thrattidion. I find nothing in the nature of the supraorbital canal of Congo- thrissa to distinguish it from other Pellonulinae. A terminal mouth occurs in Thrattidion and in some of the Congo Pellonu- linae. 22 BREVIORA No. 382 In conclusion, Congothrissa does not merit placement in a family by itself. It may either be placed in a separate subfamily, Congothrissinae, or preferably in the Pellonulinae. Its closest relatives are to be sought among the Pellonulinae in the Congo basin. Systematic status of Thrattidion. Several of the distinguishing characteristics of Thrattidion — absence of supramaxillary, tooth- less maxillary, reduced squamation and scutellation, and low branchiostegal and gill raker counts — are loss or reduction characters more or less shared with Congothrissa. On the other hand, Thrattidion has more vertebrae (45-47), anal rays (23-25), and epurals (three) than are recorded for almost any other African Pellonulinae {Microthrissa royauxi has as many as 24 or 25 anal rays), and is relatively deeper bodied than any of the others at comparable size. If Congothrissinae is recognized, then a separate subfamily might also be justified for Thrattidion. It is preferable to place them both in the Pellonulinae. Relationships of Dussumieriidae. Whitehead (1963b) ad- vanced a theory with important implications for clupeoid phylo- geny, namely, that Dussumieriidae are modern representatives of a primitive group of scuteless clupeoid. This theory is strongly contraindicated by the following considerations: 1 ) All Dussumieriidae have at least a pelvic scute that is apparently homologous with the prcpelvic and postpelvic scutes in other clupeoids. 2) Several dussumieriids have either prepelvic or postpelvic scutes or both: Etrumeus (subfamily Dussumieriinae) has a post- pelvic scute immediately behind, and partially overlapping, the pelvic scute (Chapman, 1948, fig. 17 of E. micropus on p. 38; personal observation); Gilchristella and Ehirava mulabaricus (subfamily Spratelloidinae) have prepelvic scutes, unkeeled but with lateral arms, and no postpelvic scutes (Whitehead, 1963b; and Poll, Whitehead, and Hopson, 1965: 286); and Laeviscutella, which Poll, Whitehead, and Hopson placed in the same tribe with Gilchristella and Ehirava, has both prepelvic and postpelvic scutes. In Laeviscutella the prepelvic scutes have lateral arms but are unkeeled, whereas the postpelvic scutes are keeled but lack lateral arms, a condition unlike that in any other clupeoid. If the absence of prepelvic and postpelvic scutes is a primitive 1972 AFRICAN CLUPEID FISH 23 character in those Dussumieriidae hacking them, then such scutes must have originated independently in several lines of clupeoids, including Indo-Pacific Engraulidae and two subfamilies of Dus- sumieriidae. It seems more likely that the ancestors of living clupeoids had a more or less full complement of scutes, and that they have been reduced or lost to varying degrees many times. A strong trend to reduce scutellation is evident in Pellonulinae. So far as known, the pelvic scute is retained in all living clupeoids. Whitehead (1963b) suggested that the pelvic scute of Dussumi- eriidae originated from pelvic splint bones, independently of pre- pelvic and postpelvic scutes. Greenwood (1968: 256) rejected a pelvic spUnt origin for the pelvic scute, largely on the ground that the pelvic scute is morphologically so similar to the other abdominal scutes. In Thrattidion the ontogeny of the pelvic scute is similar to that of the others, and the pelvic splint bone (usually absent in clupeoids) is quite separate. While placing Laeviscutella in the dussumieriid tribe Ehiravini on account of its reduced scutellation, Poll, Whitehead, and Hopson (1965: 286-288) also cited reasons why it might be placed in the Clupeidae, and specifically in Pellonulinae. Daget and litis (1965: 45) placed it in the Clupeidae because "the only known species of Laeviscutella is in reality very close to the forms of Pellonula with which it is easily confused." In my opinion Laeviscutella should be placed in the Pellonulinae. Other genera currently placed in the Ehiravini also may have close relatives in the Pellonulinae. Thus the scuted Pellonulops, from Madagascar and South Africa (cf. Smith, 1954: 90-91, fig. 110), looks as if it might belong to the Pellonulinae and at the same time be closely related to Gikivi Stella and Sauvagella. Whitehead's main reason for considering the Dussumieriidae a primitive type of clupeoid, apart from their lack of scutes, is the high number of branchiostegal rays in the subfamily Dussumi- eriinae, and a belief that reduction in number of branchiostegal rays is a major trend in clupeoid evolution. The notion that the number of branchiostegal rays can be reduced but not increased in teleostean phylogeny is one that had some currency formerly but has lost ground in recent years. Rosen (1964) mentions several phyletic fines in Acanthopterygii in which the number of branchiostegal rays has increased. Increases — some quite 24 BREVIORA No. 382 striking — surely have occurred in various highly specialized lines of catfishcs in South America and India (personal observation), as well as in eels. Reduction of branchiostegal rays, of course, is the more usual trend, very frequently associated with reduction in body size, as it surely is in Congothrissa and Thrattidion. The high number of branchiostegal rays in Dussumeriinae is probably due to specialization rather than retention of a primitive character. The more pellonulin-like Spratelloidinae have only six or seven branchiostegal rays. LITERATURE CITED AuDENAERDE, Thys, V. D. 1969. Description of a new genus and species of clupeoid fish from Sierra Leone. Rev. Zool. Bot. Afr., 80(3-4): 385-390. Blache, J. 1964. Les poissons du Bassin du Tchad et du Bassin adjacent du Mayo-Kebbi. Mem. Off. Rech. Sci. Tech. Outre-mer, Paris: 1-483. Chapman, W. M. 1948. The osteology and relationships of the round herring Etriiineus inicropus Temminck and Schlegel. Proc. California Acad. Sci., 26(2): 25-41. Coulter, G. W. 1962. Limnothrissa miodon and Stolothrissa tanganicae in Lake Tanganyika. In: Research results. Rep. Jt. Fish. Res. Org. N. Rhodesia 1960, 10: 7-30 (not seen). Daget, J. 1954. Les poissons du Niger superieur. Mem. Inst. Franc. Afr. Noire, No. 36: 1-391. , AND A. Iltis. 1965. Poissons de Cote d'lvoire (eaux douces et saumatres). Mem. Inst. Franc. Afr. Noire, No. 74, 1-385, 4 pis. GossE, J. P. 1963. Le milieu aquatique et lecologie des poissons dans la region de Yangambi. Ann. Mus. Roy. Afr. Centr., ser. oct., sci. zool., No. 116: 113-270, pis. 1-10. Greenwood, P. H. 1968. The osteology and relationships of the Denti- cipitidae, a family of clupeomorph fishes. Bull. British Mus. (Nat. Hist.), 16(6): 215-273. Matthes, H. 1964. Les poissons du lac Tumba et de la region de I'lkela. Ann. Mus. Roy. Afr. Centr., ser. oct., sci. zool.. No. 126: 1-204, 1 map, 1 chart, pis. 1-6. Nelson, G. J. 1967. Gill arches of teleostean fishes of the family Clupeidae. Copeia, 1967(2): 389-398. . 1969. Infraorbital bones and their bearing on the phylogeny and geography of osteoglossomorph fishes. American Mus. Novitates, No. 2394: 1-37. 1972 AFRICAN CLUPEID FISH 25 Poll, M. 1953. Poissons non Cichlidae. Exploration hydrobiologique du Lac Tanganika (1946-47), Resiiltats Scientifiques, Vol. 3, part 5a. 251 pp., 11 pis. Bruxelles, Inst. Royal Sci. Nat. Belgique. 1964. Une famille dulcicole noiivelle de poissons africains: les Congothrissidae. Mem. Acad, roy Sci. Outre-mer, cl. sci. nat. med., n.s., 15(2): 1-40, 8 pis. Poll, M., P. J. P. Whitehead, and A. J. Hopson. 1965. A new genus and species of clupeoid fish from West Africa. Bull. Acad. roy. Belgique, cl. sci., ser. 5, 51(3): 277-292. Regan, C. T. 1917. A revision of the clupeid fishes of the genus Pellonula and of related genera in the rivers of Africa. Ann. Mag. Nat. Hist., ser. 8, 19: 198-207. Reynolds, J. D. 1970. Biology of the small pelagic fishes in the new Volta Lake in Ghana. I. The lake and the fish: feeding habits. Hydrobiologia, 35(3-4): 568-603. 1971. Biology of the small pelagic fishes in the new Volta Lake in Ghana. II. Schooling and migrations. Hydrobiologia, 38(1): 79-91. RiDEWooD, W. G. 1904. On the cranial osteology of the clupeoid fishes. Proc. Zool. Soc. London, 1904, 2: 448-493. Rosen, D. E. 1964. The relationships and taxonomic position of the halfbeaks, killifishes, silversides, and their relatives. Bull. American Mus. Nat. Hist., 127(5): 217-268. Smith, J. L. B. 1954. The sea fishes of Southern Africa (2nd ed.). South Africa, Central News Agency, Ltd. 580 pp. Svensson, G. S. O. 1933. Fresh water fishes from the Gambia River. Kungl. Svenska Vetens. Handl., ser. 3, 12(3): 1-102, 8 pis. Whitehead, P. J. P. 1963a. A contribution to the classification of clupeoid fishes. Ann. Mag. Nat. Hist., ser. 13, 5: 737-750. — 1963b. A revision of the Recent round herrings (Pisces: Dussumieriidae). Bull. British Mus. (Nat. Hist.), zool., 10(6): 307-380. Wohlfahrt, T. 1937. Anatomische Untersuchungen iiber die Seitenkanale der Sardine (Cliipea pilchardus Walb.). Zeitschr. Morph. Okol. Tiere, 33(3): 381-411. ML'S. COM?. ZOOU Linr.ARY M^R 1 8 m B R E V I O R A-?. ^..^ ^SITY usemni of Comparative Zoology CAMBRrocE, Mass. 25 February, 1972 Number 383 TWO NEW GENERA OF BEMBIDIINE CARABID BEETLES FROM AUSTRALIA AND SOUTH AMERICA WITH NOTES ON THEIR PHYLOGENETIC AND ZOOGEOGRAPHIC SIGNIFICANCE (COLEOPTERA) ' Terry L. Erwin- Abstract. Two genera and two species are described as new. A key to related species is given and morphological striiciiires are discussed in a phylogenetic perspective with other bembidiine groups. An "elytral chaetotaxy map," which is based on a broad study of tachyine beetles, is included. The distribution of each species is presented by locality records. Illustrations of important morphological characters are given as well as habitus drawings of two species, each representing the new genera described. INTRODUCTION Darlington (1962) pointed out that the members of the hobani group "are certainly Tachys rather than Bembidion by current classification. | although | the species of this group arc anomalous (primitive?) in some ways and should be specially considered by students of bembidiine phylogeny." During the course of my current study of tachyine beetles, I examined sev- eral specimens of the hobani group and found some remarkable characteristics, as well as representatives of two new species. In the analysis of these characteristics and their distribution through- out the carabid beetles, another species with unusual character- istics was uncovered; Jeannel (1962) redescribed and illustrated 1 The types located in the British Museum (Natural History). London, were examined with the support of a grant from the American Philosophical Society, Penrose Fund #5795. - Division of Coleoptera. Department oi Entomology, Smithsonian Institution. Washington, D. C. 20560. 2 BREVIORA No. 383 this species, "Plataphus" reicheelliini Csiki. and assigned it to a "phylogenctic series" of the Tribe Bembidiini .v. str. On the basis of these newly discovered and newly interpreted character- istics, and information from my general tachyine study, I here propose two new genera to contain the groups of species dis- cussed above, and discuss the phylogenetic implications and relict distribution pattern exhibited by the extant species of each group. I have included in this paper an "elytral chaetotaxy map,'" which is the result of the examination of all described and many undescribed groups of tachyine beetles. This was done now to facilitate the descriptions and discussions of these groups. METHODS The methods used here are essentially the same as those I employed elsewhere (Erwin, 1970) with a few exceptions. Measurements were made with a micrometer eyepiece in a Wild M5 stereoscopic microscope at a magnification of 50 diameters. The scale interval was 0.015 mm. Total length given here is in the sense of Lindroth (1961-69), that is, one overall "habitus" measurement of the specimen in a near normal pose, rather than the summation of head, pronotum, and elytron measurements used commonly for variation studies. Illustrations are of two types. Most are India Ink on bristol board, but those with half-tone qualities were made with an "F softness" pencil on bristol board. ACKNOWLEDGEMENTS I wish to thank P. J. Darlington, Jr. and H. B. Leech for the loan of beetles in their charge and C. H. Lindroth for critically reading the manuscript and offering many valuable suggestions. TAXONOMY Tasmanitachoides new genus (Figures 1, 5, 6, 8, 10, 11, 12, 13, 15, 16, 19, 21, 22) Type Species. BemhkHon liolnirli Blackburn 1901: 123. present designa- tion. This genus includes those species Darlington (1962) recognized as the hoharti group of Tachys, as well as two new species described below. !972 SOUTH TEMPERATE CARABID-BEETLES Figure 1. TasmunitachoicU's hoharti (Blackburn), dorsal aspect, male, Queenstown. Tasmania. 4 BREVIORA No. 383 Diagnosis. Head with two parallel and sulcate frontal furrows continuous on clypeus; apical palpal article subulate; ligula 6- setose; anterior coxal cavity uniperf orate; anterior tibia with apex truncate to suboblique; scutellar stria absent externally; sutural stria entire, but not stricdy recurrent (see below); male aedeagus with symmetrical basal lobes. Description. Size: small beetles, 1.6-2.8 mm (upper limit from Darlington). Color: piceous to testaceous, appendages rufous to testaceous, or piceous. Microsculpture: wide, irregu- larly isodiametric meshes almost longitudinally arranged on elytra, more transverse on head and pronotum, very finely im- pressed. Head: broad and moderately depressed, with small but protruding eyes; frontal furrows sulciform and long, continuous on clypeus; clypeus with two setae at each anterior angle; ligula 6-setose (Fig. 10); mentum not foveate, anterior margin acutely and strongly toothed; mouthparts (Figs. 5, 10, 16). Antennal article 2:3 ratio more than 1.0; flagellar articles longer than wide, filiform (Fig. 15). Eyes and head with very short and sparsely spaced setae. Prothorax: pronotum narrowly cordiform with sharp hind angles and broadly lobed base, sides barely reflexed, carina medial to hind angles; two pairs of lateral setae present, the anterior pair far forward even before anterior third, the posterior pair just before hind angles; margins not serrate or setulose; anterior transverse impression obsolete, the posterior impression deeply engraved, interrupted at middle, basal foveae deeply impressed. Prosternum with short, sparsely spaced setae. Coxal cavities uniperforate-separate-closed. Tibia very shallowly oblique apically with outer spine lateral to subapical transverse comb. Male with two dilated basal tarsal articles each with sparse setae beneath. Claws simple. Pterothorax: elytra narrow and elongate, more or less parallel-sided with prominent humeri; sides barely reflexed, without subapical plica; margins setulose-serrate or not; dorsal surface with deeply impressed punctulate striae, at least sutural and base of 5 always present, others various, but 8 absent except at seta EoSa (Fig. 3); recurrent groove appear- ing double because of two elongate and foveate punctures (Fig. 1 ) that are not continuous with sutural stria. Elytral chaetotaxy as in Figure 1. Mesonotum fully winged. Mesocoxae conjunct- confluent. Middle tibia of male with lateroapical brush. Claws simple. Abdomen: segments II, III, IV, and V with a single 1972 SOUTH TEMPERATE CARABID-BEETLES 5 pair of ambulatory setae; segment VI with one pair in male and two pairs in female specimens, those in female parallel to hind margin of segment; venter with many scattered setae as described for eyes and head. Genitalia: female stylus (Figs. 8, 11, 12). Male parameres and median lobe (Figs. 19, 21, 22), the latter with symmetrical basal lobes. Habitat. According to Darlington (1962) these beetles occur "in sand or gravel or under stones by rivers or brooks." Distribution. The known range extends from Tasmania and temperate southeastern Australia north to Katherine in the Northern Territory, but no specimens are known from the area between Rockhampton and Katherine, a distance of more than 1500 miles. Etymology. From Tasmania, the type locality of the type species; from genus Tachys (another group of carabid beetles), to which this group was originally assigned; and from Latin, aides, meaning similar to. Key to species of Tasmanitacho':des new genus (modified from Darlington's 1962 key to the hobarti group) 1. Dorsal elytral striae, except sutural, obliterated. . . liitiis Darlington r. Dorsal striae, at least 5, and usually some others, present on disc 2 2(1'). Clypeus distinctly impressed at middle 3 2'. Clypeus not distinctly impressed at middle 5 3(2). Color piceous, length ca. 1.75 mm niurninibictgensis Slcane 3'. Color: forebody darker, rufo-piceous to rufo-testaceous, than elytra, rufous to testaceous; length ca. 1.95 mm or longer 4 4(3'). Small narrow beetles, ca. 1.95-2.0 mm long, 0.65 mm wide across elytra; forebody rufo-piceous, elytra testaceous arnhemensis n. sp. 4'. Larger and broader beetles, ca. 2.5-2.8 mm long and 0.9-1.1 mm wide; forebody rufo-testaceous, elytra testaceous at apex; male genitalia (Figs. 6, 19) fitzroyi Darlington 5(2'). Larger (2.3-2.7 mm long), almost black beetles with more or less well-defined elytral striae 6 5'. Small ( 1.8 mm or less), pale beetles with weakly impressed elytral striae 7 6(5). Head across eyes subequal to distance across pronotum at widest part; elytral striae well impressed (may include Icai Sloane) liobarii Blackburn 6 BREVIORA No. 383 6'. Head narrower than pronotum; striae, especially 5, shallower wattsense Blackburn 7(5'). Elytral stria 5 well defined to at least middle of elytron; pronotum convex, broad anteriorly, and wider than head across eyes; male genitalia (Fig. 22) kingi Darlington 7'. Elytral stria 5 hardly visible, barely impressed except just behind humerus; pronotum subconvex, narrower than head across eyes katherinei n. sp. Tasmanitachoides arnhemensis n. sp. (Figure 11 ) Type specimens. The holotype female and one female para- type, labelled "Australia, N.T., Edith Falls nr. Katherine, 110 m, X-25-62," and "Collectors E. S. Ross, D. Q. Cavagnaro" are in CAS (California Academy of Sciences, San Francisco). Type locality. Edith Falls on the north branch of the Daly River, near Katherine, Northern Territory, Australia. Diagnosis. Medium-sized beetles, broad, subconvex with rufo- piceous forebody and testaceous elytra; clypeus well impressed medially; striae 1, 2, and 5 well marked beyond middle. Description. Size: 1.95-2.0 mm total length; 0.6-0.7 mm wide (2 specimens measured). Microsculpture: almost perfectly isodiametric on elytra, moderately impressed, surface shiny. Head: broadly transverse, slightly wider across eyes than prono- tum; antennae with middle articles slightly longer than wide; clypeus well impressed medially; frontal furrows deeply sulcate from anterior margin of clypeus to just behind mid-eye level, roughly parallel throughout. Prothorax: subcordate, sides sinuate just before acute hind angles; base broadly lobed, sinuate later- ally inside hind angles; side margins moderately beaded, strongly rounded at apical third; anterior margin truncate; surface micro- punctulate; basal transverse impression deep and coarse laterally, interrupted medially by extended median groove. Pterothorax: wings long; elytra elongate and narrow, sides subparallel; humerus about square, margins subserrate and microsetulose; stria 1 (sutural) deep and entire from base to apex, broadly recurrent at apex to foveate puncture Eo8b; stria 2 moderately impressed from near base to Ed6c; striae 3 and 4 short, weakly impressed from Ed3 to Ed5a; stria 5 sulcate behind recurrent 1972 SOUTH TEMPERATE CARABID-BEETLES 7 humeral margin to level of Ed3, less impressed from Ed3 to mid-elytron; Ed7b foveate, forming inner pseudo-recurrent groove. Abdomen: as described under generic description. Male genitalia unknown. Female stylus (Fig. 11). Notes. This species is most closely related to the more south- ern T. fitzroyi Darlington, but it has members that are consider- ably smaller, both in length and width. Habitat notes. Exact habitat unrecorded, but taken near Edith Falls, so probably as Darlington (1962) and Sloane (1921) re- corded for other species, that is, near water's edge. Distribution. Known only from the type locality. Etymology. From Arnhem Land, the northernmost projection of land in Australia; in reference to the general area where the types were collected. Tasmanitachoides katherinei n. sp. (Figure 1 2) Type specimen. The holotype female labelled "Australia, N.T., Edith Falls nr. Katherine, 110m, X-25-62,'" and "Collec- tors E. S. Ross, D. Q. Cavagnaro" is in CAS. Type locality. Edith Falls on the north branch of the Daly River, near Katherine, Northern Territory, Australia. Diagnosis. Small beetles, narrow, fragile, depressed, with rufo- testaceous forebody and testaceous elytra; clypeus not impressed; striae 2-5 almost obliterated; micropunctulae and microsculpture well marked. Description. Size: 1.6 mm total length; 0.5 mm wide (type). Microsculpture: almost perfectly isodiametric on elytra, moder- ately impressed, surface shiny. Head: broadly transverse, slightly wider across eyes than pronotum and elytra; antennae with mid- dle articles about twice as long as wide; clypeus not impressed medially; frontal furrows shallowly sulcate, poorly defined be- hind level of anterior edge of eye, slightly convergent posteriorly. Prothorax: subcordate; sides sinuate just before slightly obtuse hind angles; base broadly lobed, abruptly sinuate laterally inside hind angles; side margins weakly beaded, moderately rounded at apical third; anterior margin truncate; pronotum barely convex, micropunctulate; basal transverse impression deep and short, 8 BREVIORA No. 383 laterally interrupted well inside hind angle and at middle by extended median groove. Pterothorax: wings long; elytra elongate and narrow, sides parallel; humerus almost square; side margins subserrate and microsetulose; stria 1 deep and entire from base to apex, broadly recurrent at apex to foveate puncture EoSb; outer striae nearly obliterated; Ed7b shallowly foveate, forming an inner pseudo-recurrent groove. Abdomen: as described for genus. Male genitalia unknown. Female stylus (Fig. 12). Notes. This species is most closely related to the more south- ern T. kingi Darlington, but it has members with much less im- pressed striae and a pronotum that is narrower than the head across the eyes. Habitat notes. See preceding species. Distribution. Known only from the type locality. Etymology. From Katherine, the name of the town near which the type was discovered. NOTES ON OTHER SPECIES OF TASMANITACHOIDES Darlington (1962) thought that Bembidion wattsense Black- burn might be a synonym of T. hobarti. I have examined the type in the British Museum and believe that B. wattsense is a valid species and should be included in Tasmanitachoides. I have not seen specimens of T. leai Sloane and prefer to follow Dar- lington's ideas at this time. I have seen the types of all other species listed in the key; those described by Darlington are in the Museum of Comparative Zoology, Cambridge, Massachusetts, and those of Sloane and Blackburn are in the British Museum (Natural History). Bembid arenas new genus (Figures 2, 4, 7, 9, 14, 17, 18, 20, 23) Type Species. Bembidion reicheellum Csiki 1929: 162, present designation. This species was originally described as Bembidiiini reichei Germain (1906: 617). Since reichei was preoccupied (Putzeys, 1846), Csiki (1929: 162) proposed reicheellum. Jeanne! (1962: 653) placed the species in Plataphiis because of the structure of the basal lobes of the aedeagus. However, his drawings (Jeannel 1941: 533) are not ac- curate, and, as Lindroth ( 1963) points out, 'The structure of the basal orifice of penis, used by Jea. (1941, p. 533) for characterizing his 'genus' Plataphiis. is not different from that of related groups." 1972 SOUTH TEMPERATE CARABID-BEETLES Figure 2. Bemhidareiias reicheelliiin (Csiki), dorsal aspect, male, vicinity Punta Arenas, Chile. 10 BREVIORA No. 383 Diagnosis. Apical palpal article subulate, scutellar stria present, anterior coxal cavity uniperforate, anterior tibia with apex truncate, but with transverse subapical comb, clypeus 4- setose, apex of stria 5 deepened, abdomen pubescent and male aedeagus with symmetrical basal lobes. Description. Size: 3.6 to 4.2 mm. Color: piceous with rufescent tibiae. Microsculpture: moderately wide on elytra, almost regularly isodiametric meshes with slight tendency toward transverse arrangement throughout. Head: broad and moderately depressed, with moderate but prominent eyes; frontal furrows double, the inner pair linear, deep and wide from clypeus to just behind eyes at posterior edge, and with a slightly raised tubercle at middle of each furrow, the outer pair short and deep, enclosing the anterior supraorbital setae but not continued on clypeus; clypeus with two setae at end of each frontal furrow; ligula 6-setose (Fig. 9); mentum not foveate, anterior margin with strong, truncated tooth; mouth parts (Figs. 4, 9, 17). Antennal articles 1, 3-5 of subequal length, 1 more robust; flagellar arti- cles and 2 only slightly shorter (Fig. 18). Eyes and gena be- neath eyes with short and scattered setae. Prothorax: pronotum narrowly cordiform with sharp but slightly obtuse hind angles and truncate base, sides slightly reflexed, with deep basal foveae; two pairs of lateral setae present, the anterior pair far forward, even before anterior third, the posterior pair just anterior to hind angles; margin not setulose, anterior transverse impression obso- lete, the basal impression deep and wide. Prostemum glabrous. Coxal cavities unipcrforate-separate-open. Tibia truncate apically but with subapical transverse comb terminated with two spine-like setae. Male with two dilated basal tarsal articles each with parallel rows of sparse modified setae beneath. Claws simple. Pterothorax: elytra moderately narrow, depressed and elongate, with slightly arcuate sides and prominent humeri; sides reflexed, with small subapical plica; margins not setulose; dorsal surface with 5 well-impressed striae, the sixth feebly impressed, seventh obsolete, and eighth wefl impressed behind middle only, absent before middle; scutellar striae well developed; apex of stria 5 deeply impressed and connected with sutural stria (forming "re- current groove"). Elytral chaetotaxy as in Figure 2. Mesonotum fully winged. Mesocoxae conjunct-confluent. Metasternum and 1972 SOUTH TEMPERATE CARABID-BEETLES 1 1 metacoxae with scattered pubescence. Abdomen: segments II, III, IV, V with a single pair of ambulatory setae; segment VI with one pair in male and two pairs in female specimens, those of female arranged in a row transversely across segment; all segments with short, sparse pubescence. Genitalia: female stylus (Fig. 20). Male parameres and median lobe (Fig. 23), the latter with symmetrical basal lobes, but no '"brush sclerite" in the in- ternal sac as in Bcfubiclion members. Notes. I have also seen specimens of a second species of this genus from southern South America in the collection of J. Negre. Negre plans to describe and figure this species in a paper he is preparing on Berubidion of southern South America. The presence of short sparse setae covering the venter of all segments of the abdomen is also found in the subgenus Triclio- plataphus Netolitzky of Bembidion. Setae are also found along only the apical margin of each segment in members of the sub- genus Blepharoplataphiis Netolitzky. This character state must be regarded as convergent in these otherwise quite different groups of beetles. Habitat. According to Darlington {in litt.) this beetle "live(d) in gravel by brooks." Distribution. The known range of this group is southern Chile (Prov. Magellanes ) in the vicinity of Punta Arenas (53°40' S Lat.). Etymology. From Bembidion, another group of closely related beetles, and Punta Arenas, the area in which these beetles are found. DISCUSSION The unexpected discovery of symmetrical basal lobes on the male genitalia of the hobarti group and Bembidarenas led me to investigate this and many other characteristics, as well as their distribution throughout the trechine-bembidiine complex. Some of this investigation has been done on actual specimens, but I have also consulted articles by Jeannel (1926, 1932, 1936, 1941, 1946, 1962), Lindroth (1961, 1963, 1966, 1969), Darhnston (1962), Ball (1960), Bell (1967), Valentine (1932), and bene- fited through correspondence with T. C. Barr and his unpub- lished "Key to Tribes of Subfamily Bembidiinae." 12 BREVIORA No. 383 eol eo2 ed 1 ed2 ed3 ed4 edS ed6 ed7 ed8 Figure 3. Diagrammatic elytron showing all known positions of setae (O) in tachyine beetles. The Eo series is the elytral "umbilicate" series. The Ed series is the elytral disc series. The letters represent the various positions in which these setae are found in different groups, while the short connecting lines are the hypothetical directions of movement from the ancestral condition (A). This drawing is to be used as a map to accompany the descriptions. 1972 SOUTH TEMPERATE CARABID-BEETLES 13 Figures 4-12. Fig. 4. Mandibles of Benibidarenas reicheellum (Csiki), dorsal aspect, male, vicinity Punta Arenas, Chile. Fig. 5. Same of Tasinoni- tachuides hobarti (Blackburn), male, Queenstown, Tasmania. Fig. 6. Geni- talic ring sclerite of Tasmanitachoides fitzroyi (Darlington), dorsal aspect, male, Fitzroy River, North Rockhampton, Queensland, Australia. Fig. 7. Base of genitalic ring sclerite of Bcnihidarcnas reicheellum (Csiki), dorsal aspect, male, vicinity Punta Arenas, Chile. Fig. 8. Genital sclerites of Tasmanitachoides hobarti (Blackburn), ventral aspect, female, Queenstown, Tasmania. Fig. 9. Labium and left palpus of Bembidareiuis reicheellum (Csiki), ventral aspect, male, vicinity Punta Arenas, Chile. Fig. 10. Same of Tasmanitachoides hobarti (Blackburn), male, Queenstown, Tasmania. Fig. 11. Right stylus of female genitalia of Tasmanitachoides arnhemensis n. sp., ventral aspect, Edith Falls, Northern Territory, Australia. Fig. 12. Same of Tasmanitachoides katherinei n. sp., same locality. 14 BREVIORA No. 383 Figures 13-18. Fig. 13. Left anterior leg of Tasinanitachoides hobaitl (Blackburn), lateral aspect, male, Queenstown, Tasmania. Fig. 14. Same of Bembidarenas reicheellum (Csiki), vicinity Punta Arenas, Chile. Fig. 15. Right antenna of Tasmanitachoides hobarti (Blackburn), lateral aspect, male, Queenstown, Tasmania; pubescence and setae not shown. Fig. 16. Left maxilla and palpus of Tasinanitachoides hobarti (Blackburn), dorsal aspect, male, Queenstown, Tasmania. Fig. 17. Same of Bembidarenas rcichceUum (Csiki), vicinity Punta Arenas. Chile. Fig. 18. Same of Bembidarenas reicheelhim (Csiki). vicinity Punta Arenas, Chile; see Fig. 15. 1972 SOUTH TEMPERATE CARABID-BEETLES 15 Figures 19-23. Fig. 19. Male genitalia and apicies of parameres of Tasinanitachoides fitzroyi (Darlington). Fitzroy River. North Rockhamp- ton. Queensland, Australia. Fig. 20. Female stylus of Bemhidarenas reicheelliim (Csiki), ventral aspect, vicinity Punta Arenas, Chile. Fig. 21. Male genitalia and apicies of parameres of Tasmaiiitaclioidcs hobarti (Blackburn), Queenstown, Tasmania. Fig. 22. Same of Tasinanitachoides kingi (Darlington). Queenstown, Tasmania. Fig. 23. Same of Bemhidarenas reicheelliim (Csiki), vicinity Punta Arenas, Chile. 16 BREVIORA No. 383 Traditionally, the trechine-bembidiine complex has been con- sidered as two closely related groups, each at the tribal level (higher levels in the Jeannelian "French School"). Trechini, Bembidiini, and other more loosely associated groups were con- sidered by Jeannel (1941) to constitute the "Styhfera," a "groupement naturel" of the "Caraboidea Limbata" or higher Carabidae. Jeannel's contention that this "groupement naturel, sans cependant avoir etc exactement defini" still holds today. It seems no authors before Ball (1960) and Lindroth (1961-69) worried much that their classifications were based on one or two major character states, as well as one or two minor character- istics, or less. Also these character states were investigated on only one or two specimens of one or two local species in what were in reality unknowingly diverse groups. Even so, the general classification of Carabidae has been in relatively good shape for the last 30 years and many small studies have contributed much to our knowledge at all levels; some of these studies were based on new and novel techniques and approaches. It is hoped that this paper will stimulate just such studies in the "Stylifera" to determine its inner and outer limits and its place among the diversity of the entire Family. DISCUSSION OF BEMBIDARENAS Members of Bembidarenas reicheellum differ from Bembidion at least to the extent that Phrypeiis rickseckeri Hayward does, and under certain analysis considerably more. Superficially, though, B. reicheellum is more conservative than P. rickseckeri and thus has not undergone close scrutiny by students of carabid phylogeny. The symmetrical basal lobes of the male genitalia in B. reicheellum do not occur anywhere in the true Bembidion, but are characteristic of many trechines, all patrobines, Anillina, apotomines, psydrines, Horologion, deltomerines, and Tasmani- tachoides. Nearly symmetrical lobes occur in Phrypeus and the subgenus Pseudolimnaeum of Bembidion, but the reduction of the right lobe is easily observable after dissection. The distribu- tion of symmetrical basal lobes in the Stylifera and other carabid groups indicates to me that this condition is primitive (plesio- morphic). If so, then reduced right basal lobes and the "basal bulb" have arisen several times in diverse groups, and hence must be used with caution in phylogenetic analyses. 1972 SOUTH TEMPERATE CARABID-BEETLES 17 B. reicheelliim has another major difference in the male geni- talia. All known Bembidion species possess a "brush-scleritc" in the internal sac. or remnants of one; in some groups there are species without the structure but it is clearly a secondary loss (Lindroth, personal communication). B. reicheelliim does not have this structure at all. The brush sclerite does not occur in Phrypeus either, but is found in Asaphidion. This structure and many others clearly illustrate the sister group relationship of Bembidion and Asaphidion, regardless of the one extra umbilicate seta (Eo9) in the latter. B. reicheellum has a plurisetose ligula with remnants of para- glossal lobes. The plurisetose ligula is found in all trechines, Asaphidion, and Tasmanitachoides, each with six to eight setae. All other bembidiines, pogonines, patrobines, etc., of the Styli- fera have four or less setae, usually two. The long narrow lateral lobes, probably remnants of the paraglossae, are characteristic of all trechines, but also occur in some Anillina, patrobines, and Horologion. Other members of the Stylifera have small rounded lobes or no lobes. The distribution of these character states indi- cates to me that the multisetiferous ligula is derived (apo- morphic), while the produced lateral lobes are primitive (rem- nants of the paraglossae). At present, any attempt to interpret these two character states into overall phylogenetic analysis leads to utter confusion; new data must be accumulated before further attempts are made. The last characteristic that is now known to be peculiar in B. reicheelliim is the presence of two pairs of setae on the clypeus. This character state also is now known in Tasmanitachoides and a few primitive trechines, e.g., Amblystogenium pacificiim Putzeys. On the basis of gross overall form (e.g., subulate terminal palpal articles, short sutural stria, elytral chaetotaxy, pubescent penultimate palpal articles, etc.) and the analysis of character states above, I think that B. reicheellum is an early off-shoot of the Bembidion lineage which retained some characteristics of the mutual "trechine-bembidiine" stock. It seems more primitive in several characteristics than Phrypeus and probably separated from the main line of evolution before Phrypeus. The habitat of gravel by brooks is considered by Darlington {in litt.) to "be a very old. stable habitat, and that the running water acts as a 18 BREvioRA No. 383 buffer against climatic changes. I think insects in this habitat may persist for very long periods." DISCUSSION OF TASMANITACHOIDES In addition to some of the points discussed above, Tasmani- tachoides members have other peculiar characteristics not gener- ally found in the Bembidiini. Deeply sulcate frontal furrows, ex- tending to about mid-eye level, are elsewhere found only in Phrypeus. The small group of lateral papillae of the lacinia are unique, but some Anillina and Lymnastis have a single spine near this same location. Whether the structures are truly homol- ogous is doubtful, however, because of their orientation. On the basis of gross overall form (e.g., subulate terminal palpal articles, complete sutural stria, elytral chaetotaxy, uniper- forate anterior coxal cavity, pubescent penultimate palpal arti- cles, etc.) and the analysis of character states above, I think the hobarti group is best placed as an early off-shoot of the tachyine lineage which gave rise to the Anillina. All true tachyines (and Lymnastis-Micratopus, which I regard as true tachyines on the basis of many characteristics) have biperforate anterior coxal cavities, and are the only Stylifera that do (aside from Apotomus if this group really belongs to the Stylifera). The distribution of biperforate coxal cavities indicates that this character state is derived (apomorphic), but convergent in distantly related groups, perhaps in strengthening the prothorax under greater demands for digging in partial subterranean life (but see also Erwin, 1970: 168).^'' The characteristics of members of the hobarti group show similarities to the trechines, but as in B. reicheelliim 1 think these characteristics indicate an old lineage surviving in an old but stable habitat, and maintaining certain characteristics of an early "trechine-bembidiine" stock. The relationships within the Stylifera and particularly within the trechine-bembidiine complex will be thoroughly discussed along with the supporting evidence in my current revision of the Tachyina (Erwin, MS) and need not be dealt with at length here. 1972 SOUTH TEMPERATE CARABID-BEETLES 19 LITERATURE CITED Ball. G. E. 1960. Carabidae (Latreille. 1810). //; R. H. Arnett. Jr., The Beetles of the United States. The Washington Catholic Uni- versity of America Press, pp. 55^181. Bell. R. T. 1967. Coxa! cavities and the classification of the Adephaga (Coleoptera). Ann. Ent. Soc. America. 60(1): 101-107. Blackburn, T. 1901. | Australian Bembidiini.] Trans. R. Soc. South Australia. 25: 120-124. CsiKi. E. 1929. Coleopterorum Catalogus, pars 104, Carabidae; Harpalinae. pp. 347-527 Darlington, P. J.. Jr. 1962. Australian Carabid beetles XI. Some Tachys. Psyche. 69( 3 ) : 1 1 7- i 28. Erwin, T. L. 1970. A reclassification of Bombardier Beetles and a taxonomic revision of the North and Middle American species (Cara- bidae: Brachinida). Quaestiones Ent., 6: 4-215. . (MS). A revision of the New World species of Tachyina and a reclassification of the generic components of the world. Germain, P. 1906. Apuntes entomolojicos. Anal. Univ. Chile, 117: 589-653. Jeannel, R. G. 1926. Monographic des Trechinae. L"Abeille, 32: 224-550. 1932. Revision des genre Lininastis (Coleoptera, Cara- bidae). Soc. ent. Fr., Livre der Centenaire, pp. 167-187. 1936. Les Bembidiides Endogcs (Col. Carabidae). Rev. Fr. Ent.. 3: 241-399. 1941. Faune de France 39, Coleopteres Carabiques. Paris, 1. 571 pp. 1946. Coleopteres Carabiques de la Region Malgache, part I, Faune de lEmpire Fran^ais. Paris, 6. Pp. 1-372. 1962. Biologic de TAmerique Australe. Vol. 1. Etudes sur la Faune du Sol. Paris, VII. Pp. 611-655. Lindroth, C. H. 1961-69. The Ground-beetles (excluding Cicindelidae) of Canada and Alaska, parts I-VI. Opusc. ent. Suppls. XX, XXIV, XXIX, XXXIII, XXXIV, XXXV. PuTZEYs. J. 1846. Prcmices entomologiques: Note monographique sur le genre Pasimachus: Nouv. Cicindel et Carabid. Mem. Soc. Sc. Liege, 2: 353-417. Valentine, J. M. 1932. Horologion, a new genus of cave beetles (Fam. Carabidae). Ann. ent. Soc. America. 23(1): 1-11. MUS. CGMP. ZOOL LICRARY * ' » ■• 198; B R E V I O R. A UM!VFRS1TY Meseiuin of Comparative Zoology Cambridge, Mass. 25 February. 1972 Number 384 AN ATTEMPT TO DETERMINE THE SYSTEMATIC POSITION OF ELLOPOSTOMA MEGALOMYCTER, AN ENIGMATIC FRESHWATER FISH FROM BORNEO Tyson R. Roberts^ Abstract. The types and only known specimens of EUopostoma megalomycter (Vaillant) are figured and redescribed. This highly distinc- tive fish may have a Weberian apparatus with osseous swimbladder capsules similar to those in Cobitidae, but it differs from all Cobitidae in several important respects. Resemblance between EUopostoma and Kneriidae evidently is mainly superficial. EUopostoma cannot be classified with assurance because fundamental osteological information about it is lacking, owing to the present condition of the specimens. Its relationships, when worked out on the basis of fresh material, are likely to be of phyletic significance. The present account will permit the identification of new material of EUopostoma with greater facility than the older accounts, because some errors are now corrected and more descriptive details are given. INTRODUCTION The rich fish collections obtained by the Netherlands Borneo Expedition of 1893-97 included specimens of a pecuhar little fish that Vaillant (1902: 141-149, figs. 42-45) described as Aperiop- tus megalomycter and assigned to the Cobitidae. He also proposed a provisional new genus, EUopostoma, for it. Vaillant's figures depict a moderately elongate, small-scaled fish with very large nostrils and eyes, and a most peculiar inferior mouth with a single pair of barbels. No modern ichthyologist has found it possible to assign a firm systematic position to this strange fish. Despite 1 Museum of Comparative Zoology, Cambridge, Mass. 02138. 2 BREVIORA No. 384 Vaillant's placement with the cobitids, later accounts of that family have neglected the species. Its appearance calls to mind the Kneriidae, a family of gonorynchiform fishes known only from fresh water in Africa. Weber and de Beaufort (1916: 237-239, fig. 97) examined the type specimens and doubted that they be- long to the Cobitidae. They reproduced Vaillant's figures and pointed out that the species is not an Aperioptus and should be known as Ellopostoma megalomycter (Vaillant) (see below). No additional specimens have been found and nothing further about it has appeared in print. Dr. M. Boeseman, Curator of Fishes of the Rijksmuseum van Natuurlijke Historic in Leiden, kindly lent his institution's type specimens of Ellopostoma so that I might try to classify it. The following account gives an emended description of Ellopostoma and points out the difficulties in trying to place it on the basis of the available material. There remains a major question as to whether the structures interpreted as a Weberian apparatus by Vaillant are really such. If they are, Ellopostoma' s closest rela- tionship probably is with Cobitidae. If its postcranial bony capsules have some other origin the relationships are more prob- lematic and might lie with Kneriidae. In the description given below its characteristics are compared with those of both Cobiti- dae and Kneriidae. I wish to thank Mr. Andrew Konnerth for preparing radio- graphs and Professor George S. Myers for reading the manu- script, and Dr. M. L. Bauchot for information about the Paris specimen of Ellopostoma. ELLOPOSTOMA Vaillant 1902 Type species. Aperioptus megalomycter Vaillant 1902, by monotypy. Note. In describing Aperioptus megalomycter, Vaillant pro- posed for it the new generic name Ellopostoma in a footnote in case it should prove not congeneric with Aperioptus. Aperioptus Richardson 1848 (type species Aperioptus pictorius Richardson 1848, by monotypy) was based on a rather unsatisfactory draw- ing and notes made before the only two specimens were inad- vertently thrown away, and its identity may never be resolved. The only locality information for the specimens is that they came 1971 BORNEAN FRESHWATER FISH 3 from Borneo. If they were freshwater they may have been cobitids, as implied by Giinther (1868: 371). In any event, Richardson's description and figure (reproduced on pp. 238-239 in Weber and de Beaufort, 1916) indicate a fish differing in several major respects from Vaillant's fish, the proper designation for which is thus Ellopostoma megalomycter, as pointed out by Weber and de Beaufort. Ellopostoma megalomycter (Vaillant) Figure 1 Material. The type series of Ellopostoma consists of four specimens. No holotype was designated and thus they are equiv- alent syntypes. Three specimens, 24.3, 39.5 and 41.4 mm in standard length, were retained by the Rijksmuseum in Leiden (RMNH 7777) and one specimen, 40.5 mm in standard length, was presented to the Museum National d'Histoire Naturelle in Paris (MNHN 03-202). I have examined all four and find them very soft and poorly preserved. The 24.3-mm specimen belongs to the genus Noemacheilus. It has the characteristic three pairs of barbels and 1 1 dorsal fin rays. Its eye is relatively smaller than that of Ellopostoma, but unusually large for a Noemacheilus. The Paris specimen has the mouth, snout, orbit on one side and possibly some gill arches badly damaged. The 41.4-mm speci- men, badly damaged to begin with, was dissected by Vaillant; the posterior gill arches are missing and structures in the anterior region of the vertebral column are missing or badly damaged. The 39.5-mm specimen (RMNH 7777) is intact and is hereby designated lectotype. The following description is based pri- marily on the lectotype and 41.4-mm paralectotype. Vaillant reported that the 41.4-mm specimen contained eggs which seemed ripe or nearly so. It is now completely eviscerated. The specimens arrived in the Netherlands along with other unlabelled material collected by Dr. J. Biittikofer in the vicinity of Sintang, near the middle portion of the Kapuas River (Vaillant, 1902: 149). Coloration. The specimens are now discolored and their original color pattern has largely disappeared. Vaillant (p. 147, fig. 42) gave a rather good color description, to which I can add but little: "La coloration est identique sur les trois grands ex- BREVIORA No. 384 emplaires [i. e., in all but the 24.3-mm specimen, which is a Noemacheilus] et doit se rapporter sans doute a une livree fonda- mentale, plus ou moins masquee peut-etre sur le vivant, mais qui se retabUt lorsque les animaux sont plonges dans la liqueur. Dans I'etat actuel la teinte generale est roux tres pale, devenant blanchatre argente en descendant vers le ventre, tout a fait ar- gentee a la region operculaire, ainsi que sur les cotes et le dessous de la tcte. Une serie de taches plus ou moins en quadrilateres, occupant de 3 a 5 rangees d'ecailles, ornent le dos et les flancs. Sur le premier on en compte 7, les deux anterieures, les plus petites, sont I'une a la region nuchale, I'autre a mi-distance de la dorsale, les trois suivantes, respectivement sous I'origine de cette dorsale, en son milieu, enfin juste en arriere d'elle; les deux dernieres Tune a distance egale de cette tache metepipterique et de la septieme tache, celle-ci placee un peu en avant de I'inser- tion de la caudale. Sur la ligne laterale se voient 7 ou 8 taches, les anterieures peu distinctes; la derniere est au milieu du pedon- cule caudal a I'insertion mcme de Furopterc, les deux precedentes repondcnt aux intervalles clairs des trois dernieres taches dorsales, disposecs par rapport a celles-ci en damier; la quatrieme en procedant toujours d'arriere en avant est sous la partie anterieure de la tache dorsale metepipterique, les 3 ou 4 taches anterieures Figure 1. Ellopostoma megalomycter, 39.5-mm lectotype, RMNH 7777 (camera lucida). First elongate ray of anal and one or two rays of dorsal fin that are broken off restored on basis of 41.4-mm paralectotype. Arrow indicates position of vent. Base of pectoral fin should be slightly more anterior than figured. 1971 BORNE AN FRESHWATER FISH 5 moins developpees, moins distinctes, sont plutot placees au-des- sous de la ligne laterale, la plus avancce scrait juste en avant de I'origine de la dorsale; dans I'espace compris cntre ces taches lateroanterieures et les taches dorsalcs se voicnt 4 macules, nu- ageuses, repondant aux intervalles des cinq taches dorsales anterieures. Enfin il ne faut pas negliger d'attirer Tattention sur une petite tache de forme demi-circulaire, d'un noire bleuatre accentue, qui pourrait bien ctre cerclee de blanc, c'est-a-dire ocellee, laquelle orne la caudale dans sa partie infericure au point d'insertion avec la pedoncule. L'iris parait sombre; un cercle pupillaire argente." The silvery white coloration on the opercle and lower side of the head is still evident, as is the small black spot on the ventral half of the caudal peduncle. A similar spot is present on the caudal peduncle in many Cobitidae and in some Kneriidae. Proportional measurements. Proportional measurements ex- pressed as times in standard length, those of 39.5-mm lectotype followed in parentheses by those of 41.4-mm paralectotype: head 4.4 (4.8); eye 14.9 (13.8); snout 16.5 (17.6); bony inter- orbital 27.3 (26.7); depth 6.5 (5.9); depth of caudal peduncle 11.5 (10.6); snout-tip to dorsal origin 2.5 (2.45); snout-tip to pelvic origin 2.08 (1.91); snout-tip to vent 1.80(1.72); snout- tip to anal origin 1.23 (1.20); length of dorsal base 3.9 (3.75); length of anal base 12.4(14.3); length of caudal peduncle 8.0 (8.6); length of pectoral fin 5.1 (4.9); length of longest dorsal ray 4.5 (broken); length of anal fin 6.7 (6.2). Fins (Fig. 1). Dorsal iii 16, last ray divided to base. Anal iii 5 or iii 6, last branched ray divided to base. Pectoral i 1 1 or i 12. Pelvic i 7 (pelvic splint present). Caudal with nine prin- cipal rays in both lobes, about seven procurrent rays in upper lobe and four in lower lobe. Dorsal fin origin slightly posterior to a vertical midway between insertions of pectoral and pelvic fins. Pelvic insertion shghtly anterior to a vertical through base of seventh segmented ray of dorsal fin. Anal fin set far back, its origin somewhat anterior to a vertical midway between base of last dorsal fin ray and end of hypural fan. Origin of dorsal on a vertical through ninth vertebra. Pelvic insertion on a vertical through 13th or 14th vertebra. Anal origin on a vertical through 25th vertebra. 6 BREVIORA No. 384 Pectoral fin somewhat larger than pelvic fin, none of its rays hypertrophied or otherwise modified, its longest ray reaching al- most to insertion of pelvic fin. Height of dorsal fin almost twice that of anal fin. Dorsal fin base about three and one half times longer than anal fin base. Dorsal fin margin slightly falcate, the first three branched rays sHghtly longer than the rest. Anal fin margin more or less straight. Caudal deeply forked (damaged in both specimens). Membranes between rays in all fins uni- formly thin. Anal fin position variable in Kneriidae, several of them having the anal fin as far posterior as in EUopostoma. In Kneriidae dorsal fin short-based, never with more than 10 rays. Fin counts and placements highly variable in Cobitidae. Relatively few cobitids have the anal fin as far posterior as EUopostoma. Some (e.g., Noemacheilus pavonaceus) have fin placements and number of fin rays near those of EUopostoma. Number of principal caudal rays variable in both Cobitidae (from 17 to 19) and Kneriidae (16 in Grasseichthys, 18-20 in Parakneria). Position of vent (Fig. 1). Vent located between pelvic fins, somewhat closer to insertion of outermost pelvic ray than to tip of longest pelvic ray. In cobitids the vent position varies from midway between pelvic insertion and anal origin (in some Noe- macheilinae) to immediately anterior to anal fin origin (in some Cobitinae and Botiinae). In at least some (all?) kneriids vent slightly or immediately anterior to anal fin origin. Squamation (Fig. 1). Head scaleless. Body entirely scaled except for a small area in front of pectoral base. Isthmus scaled up to attachment of gill membranes. Fin bases scaleless. Ap- proximately 75 scales in a lateral series from upper angle of gill opening to base of caudal fin, 20 scales between supraoccipital spine and dorsal fin origin, 20 scales between dorsal fin origin and pelvic fin insertion (11 rows above and eight rows below lateral line), and 17 scales around caudal peduncle. Scales cycloid, height about 1.25 times width. A scale from side of body above pectoral fin has 12 radii and 22 circuU on its anterior field and 14 radii and 16 circuli on its posterior field. Focus well within anterior half of scale. Vaillant regarded the scales of EUopostoma as belonging to the type (multiradiate) characteristic of Cobitidae, and quite distinct from the type (pauciradiate) characteristic of Cyprinidae. 1971 BORNEAN FRESHWATER FISH Mouth (Figs. 1-4). Mouth inferior. No teeth on jaws or roof of mouth. Upper jaw apparently bordered exclusively by preniaxillaries, which have broad-based ascending processes and are well separated from each other at midline. Approximate shape of preniaxillaries and of portion of dentaries bordering mouth as in Figure 4. Preoral and oral barbels absent. A median fleshy thickening or mentum on lower lip behind symphysis of mandibles (Figs. 2, 3). The name Ellopostoina was given be- cause the mouth resembles that of a sturgeon (Vaillant, 1902: 145). Figure 2. Close-up of head with mouth almost fully open, 39.5-mm lectotype (camera lucida). Arrows indicate extent of gill opening. Vaillant recorded a pair of barbels, one at each side of the mouth, at the junction of the upper and lower jaws, which he supposed were located at the ends of the maxillary bones (I am unable to ascertain the presence of maxillaries, although they may be present). He was able to view the barbels only with difficulty. There is a fold or flap of skin in the lip at or near the rictus of the jaws, but this can hardly be described as a barbel. Certainly there are no structures corresponding to the elongate pair of barbels depicted in Vaillant's figure 43 in either the lecto- type or the 40.5 and 41.4-mm paralectotypes. As noted above, the 24.3-mm paralectotype is actually a specimen of Noemachei- lus, and has three pairs of barbels. 8 BREVIORA No. 384 Figure 3. Ventral view of head, 39.5-mm leciotype (camera lucida). Snout (Figs. 1-3, 5). Snout obliquely truncate and very rigid. Mesethmoid immovably articulated to frontals. Nasal organs oc- cupying a large cavity, its diameter about one-third that of eye. Nostrils separated by a membraneous strip with a small, pos- teriorly directed flap (vestigial nasal barbel?). Dorsal to nostril a slender nasal bone, ventral to nostril a small, broad-based tri- angular shaped bone (antorbital?) with its apex directly ventrally, and posterior to nostril a supraorbital bone and the lateral margin of the lateral ethmoid bone. Lateral ethmoid forming part of anterior rim of orbit, not extending ventrally to orbit, of general- ized morphology. The snout is unlike that in any other teleost I have seen. Figure 4. Ventral view of jawbones bordering mouth, 39.5-mm lecto- type (freehand). 197 BORNEAN FRESHWATER FISH Eye (Figs. 1, 2, 5). The eye is misshapen in all three speci- mens of EUopostoma. Figures 1, 2 and 5 depict its approximate shape. It presumably is oblong in live specimens, not perfectly round as in Vaillant's figures. Orbital rim free. Adipose eyelid absent. The combination of large eye and exceptionally large nostrils found in EUopostoma is unusual. The eye is relatively much smaller in Cobitidae and Kneriidae (in both of which orbital rim usually attached) with the exception of the minute kneriid Grasseichthys, which has relatively small olfactory organs. Laterosensory canals (Figs. 1-3, 5). Laterosensory system with supraorbital, infraorbital, preopercular, temporal, occipital and lateral canals. Supraorbital canal enclosed in a superficial bony tube on frontal bone until anterior margin of eye, then continued unenclosed for a short distance before it is again en- closed by the tubular nasal bone, without ethmoidal commissure or medially-directed branches. Infraorbital canal enclosed in a superficial, segmented bony tube (the infraorbital series?), con- tinued anteriorly onto snout, where it turns upwards abruptly and terminates immediately anterior to nostrils. The appearance of the infraorbital canal is similar to that in Cobitidae (and Kneriidae?). Preopercular canal with two or three short lateral branches terminating in a small pore (Fig. 3). Occipital canal supraorbital canal supraorbital frontal extrascapular nasal lateral ethmoid infraorbital canal temporal canal occipital canal temporal commissure supraoccipital posterior frontal fontanel parietal lateral canal supracleithrum Figure 5. Dorsal view cf head, based mainly on 39.5-mm lectotype (camera lucida). 10 BREVIORA No. 384 with a temporal commissure, the lateral portions enclosed in a superficial bony tube, the median portion unenclosed, traversing the posterior frontal fontanel distinctly anterior to supraoccipital bone (Fig. 5). The temporal commissure is usually (always?) present in Cobitidae and Kneriidae but usually it lies flush with the anterior margin of the supraoccipital bone, sometimes entirely enclosed in a bony tube well within the supraoccipital bone (cf. Ramaswami, 1953; Greenwood et ol., 1966). Lateral canal of Ellopostoma extending along middle of body to base of caudal fin, as in Cobitidae and Kneriidae. Rooj of cranium (Fig. 5). Cranial roof exposed, with only a thin cover of skin, unlike most Cobitidae in which dorsum of head is usually rather fleshy or at least covered with moderately thick skin. No anterior frontal fontanel. Posterior frontal fontanel slightly longer than eye, bordered by frontals, parietals, and supraoccipital, very similar to posterior frontal fontanel in Cobit- idae (cf. Ramaswami, 1953). In adults of the larger kneriids (cf. Greenwood et al., 1966; Giltay, 1934) the cranial roof is complete. Cromeiia (cf. Swinnerton, 1903) and Grasseichthys have a frontal fontanel extending to the snout, quite unlike that in either Cobitidae or Ellopostoma. Posterior region of basicranium (Fig. 6). Parasphenoid ex- tending posteriorly to anterior margin of supraoccipital, with which it is firmly sutured. Parasphenoid with a median sutural process. In Cobitidae and Kneriidae (cf. Ramaswami, 1953; Greenwood et al., 1966, fig. 6 on p. 376) posterior portion of parasphenoid basioccipital 1 J lateral process basioccipital centrum transverse process? firEt vertebral centrum Figure 6. Ventral view of part of basicranium and first vertebral centrum, 41.4-mm paralectotype (freehand). 1971 BORNEAN FRESHWATER FISH 11 parasphenoid divided, the two divisions extending distinctly pos- terior to anterior margin of basioccipital; parasphenoid without median sutural process. Basioccipital centrum of Ellopostoma with anteriorly directed lateral processes, otherwise devoid of processes that might be identical with pharyngopophyses of Cobitidae. Pharyngopophyses of Cobitidae frequently poorly de- veloped or absent; if present, they are posteriorly directed. Pharyngopophyses absent in Kneriidae. Gill cover (Figs. 1, 2). Subopercle relatively large. Shape of opercle and subopercle generalized compared to cobitids in which they are often highly modified. Gill opening wide; gill membranes broadly united to isthmus at a point below middle of subopercle (Fig. 2). Branchiostegal rays apparently three, as Vaillant reported. Cobitidae invariably have three branchiostegal rays as do other cyprinoids. The number of branchiostegal rays is also reduced in Kneriidae, Kneria and Cwmeria with three (Giltay, 1934; Swinnerton, 1903), Grasseichthys with only two (Gery, 1965). The extent of the gill opening, variable in Cobit- idae, is always very narrow in Kneriidae. Gill arches (Fig. 7) The following notes on the gill arches are based on the damaged 41.4-mm paralectotype in which only the first three arches of the left side and the first arch of the right side are present. Suspensory pharyngeals large. First arch of right side with about ten gill rakers on leading edge (none on suspensory pharyngeal) and 15 gill rakers on trailing edge 1 mm I » Figure 7. Dorsal view of left siispen^ory pharyngeal 1 with gill rakers, 41.4-mm paralectotype (camera lucida). (including a few on suspensory pharyngeal). First arch of left side with about 12 rakers on leading edge (none on suspensory pharyngeal) and 18 on trailing edge (of which five are on sus- pensory pharyngeal). The rakers on the first arch are all on the 12 BREVIORA No. 384 uppermost third of the arch. Perhaps the rakers from the lower- most two-thirds have been stripped away. Second and third gill arches with a full complement of gill rakers, at least 25 on both leading and trailing edges. Teeth absent on first three gill arches. Vaillant did not mention the gill arches or rakers; he was unsuccessful in an attempt to remove the pharyngeals intact from the 41.4-mm specimen, and they are now missing. Pectoral skeleton. Extrascapular bone present, bearing junc- tion of temporal, occipital, and lateral branches of laterosensory canal system on its dorsal surface (Fig. 5). Pectoral arch of left side of 41.4-mm specimen detached from cranium and with supracleithrum exposed; posttemporal missing (normally absent?), postcleithrum evidently absent. In Cobitidae the posttemporal is reduced, sometimes absent, and there are no postcleithra (Rendahl, 1930, 1933a; Bacescu- Mester, 1970). The pectoral skeleton of the more typical kneriids, Kneria and Parakneria, is undescribed. Swinnerton (1903: 67, fig. N) described the pectoral skeleton of Cromeria, which has a posttemporal with a very elongate upper hmb articu- lated directly to the supraoccipital (quite unlike the posttemporal of Cobitidae) and a slender postcleithrum. Anterior region of vertebral column, postcranial bony capsules, swimbladder. Radiographs of the lectotype and 41.4-mm para- lectotypc and examination of the latter reveal a pair of large, rounded, bony capsules, one on either side of the vertebral column almost immediately posterior to the cranium. The capsules of the 41.4-mm specimen are broken (their ventral portions missing) and detached from the vertebral column. Their transverse diam- eter is about three-fourths of the eye diameter. Portions of a tough membrane, presumably that of the swimbladder, remain inside the capsules. Vaillant interpreted these structures as homologues of the swimbladder capsules in Cobitidae. As the anterior portion of the vertebral column itself is also damaged and detached, and some parts possibly missing, it is impossible to tell precisely what structures contribute to form the capsules. The appearance of the capsules in radiographs of various Cobitidae is at least superficially similar to that of the capsules of Ellopostoma. The first centrum of Ellopostoma is free and relatively short, and bears a small lateral projection on one side 1971 BORNEAN FRESHWATER FISH 13 (a transverse process?) which is absent (broken off?) on the other side (Fig. 6). Either Ellopostoma has a Weberian apparatus with swimbladder capsules like Cobitidae or else it has complicated bony structures superficially similar and perhaps comparable in function to the swimbladder capsules of Cobitidae. I am inclined toward the first possibility. I am unable to make a detailed comparison of the capsules of Cobitidae with Ellopostoma on the basis of the present material. Nothing comparable to the capsules of Cobitidae or Ellopostoma has been reported in Kneriidae. Vaillant (1902) was unable to find a swimbladder in Ellopostoma and stated (p. 142): "Pour la vessie natatoire je n'ai pu en trouver trace, pas plus que d'enveloppe osseuse pouvant la contenir, I'etat de I'exemplaire [41.4-mm] paraissait cependant devoir permettre d'en constater I'existence." The de- velopment of the swim bladder is variable in Cobitidae. Fre- quently it is reduced to only the left and right halves of the anterior portion enclosed in the bony capsules, but some forms also have a nonenclosed posterior portion, the genus Noema- cheiliis with several species in which it is particularly large (Rendahl, 1933b). The swimbladders of Kneriidae have not been described in detail; Rosen and Greenwood (1970: 11) indicated that Kneriidae possess a swimbladder divided by a distinct constriction into a small anterior and a much larger posterior chamber, as in Chanos and in ostariophysans generalized with respect to their swimbladders. Vertebral counts. Radiographs reveal about 32 vertebrae (ex- cluding hypural centrum) in the lectotype of Ellopostoma and about 31 in the 41.4-mm paralectotype. The following vertebral counts have been recorded for Kneriidae: Grasseichthys 36-37 (Gery, 1965); Parakneria 41-42 (Poll, 1965); and Cromeria 42 and 45 (Swinnerton, 1903). Vertebral counts of Cobitidae range very widely. Caudal skeleton (Fig. 8). Figure 8 is based on a radiograph of the 41.4-mm specimen in which some elements are relatively well defined but others are indistinct, so that it is somewhat interpretive and should be used with caution. Hypurals appar- ently five. Proximal ends of hypurals 1-4 indistinct; impossible to determine from radiographs whether hypural 2 separate or fused with ural centrum. Parhypural evidently united to ural 14 BREVIORA No. 384 epural uroneural ? urostyle Pj.Uj(.U2? parhypurapophysis ? parhypural Figure 8. Caudal skeleton, 41.4-mm paralectotype (camera lucida of radiograph). centrum by a saddlelike joint, and bearing a well-developed par- hypurapophysis. A single epural. Urostyle slender. What ap- pears to be a slender separate ossification (uroneural?) dorsal to urostyle may actually be part of it. Bases of neural and hemal spines of vertebrae 1 and 2 indistinct. The caudal skeleton of Ellopostoma, so far as can be deter- mined, agrees in almost all respects with the caudal skeletons of both Cobitidae (undetermined cobitid; Botia macracanthus) and Kneriidae (Cromeria, Grasseichthys) described and figured by Monod (1968). The cobitids and kneriids lack uroneurals; it is possible that one is present in Ellopostoma. The cobitids have hypural 2 fused to the ural centrum, whereas it is evidently free in the kneriids; the condition of hypural 2 in Ellopostoma is undetermined. All agree in having a single epural and five hypurals, except Botia macracanthus, which has six hypurals. Rosen and Greenwood (1970: 13, fig. 10) found six hypurals and two epurals in Kneria wittei and in Parakneria sp. In K. wittei the first epural is very reduced, but in Parakneria sp. both epurals are large. 1971 BORNEAN FRESHWATER FISH 15 DISCUSSION Ellopostoma differs from all Cobitidac and Kneriidae in the relatively large size of its nostrils and eyes and in the morphology of its snout, mouth, and (so far as known) parasphenoid bone. It also differs from all Cobitidae in lacking barbels, and from all Kneriidae in its possession of postcranial bony capsules, a long- based dorsal fin with relatively numerous rays, and an extensive gill opening, and in the morphology of its posterior frontal fontanel. Ellopostoma cannot be classified with reasonable assurance because important information about the structure of its snout, mouth, pharyngeal bones, and postcranial bony capsules is lack- ing. If the postcranial bony capsules are part of a Weberian apparatus (as seems likely), then it is probably closer to Cobitidae than to any other living group. It does not show the specialized traits (mesethmoid movably articulated to f rentals, lateral ethmoid highly modified as an erectile suborbital spine) of the subfamilies Botiinae and Cobitinae, and differs greatly from them in general appearance. It shares many characters with the Noemacheilinae and is somewhat like them in general appearance, but the resem- blances may be due to convergence and retention of generalized characters. Cobitidae, Kneriidae, and Ellopostoma are similar in many respects, and this is of particular interest in view of evidence that gonorynchiforms and ostariophysans are closely related (Greenwood et ai, 1966; Rosen and Greenwood, 1970). Whatever the relationships of Ellopostoma, it is likely to be of major phyletic significance. LITERATURE CITED Bacescu-M ESTER, L. 1970. The morphological comparative study of the shouldei girdle of the fishes belonging to the Cobitidae family from Rumania. Trav. Mus. Hist. Nat. Grigore Antipa (Bucharest), 10: 251-272. Gery, J. 1965. Poissons du Bassin de I'lvindo. Biologia Gabonica, 1(4): 375-393. GiLTAY, L. 1934. Contribution a I'etude du genre Xenopomaticbthys (Kneriidae). Bull. Mus. roy. Hist. nat. Belgique, 10(44): 1-22. Greenwood, P. H., D. E. Rosen, S. H. Weitzman, and G. S. Myers. 1966. Phyletic studies of teleostean fishes, with a provisional classifi- cation of living forms. Bull. Amer. Mus. Nat. Hist., 131(4): 339- 456, pis. 21-23. 16 BREVIORA 384 GiJNTHER, A. 1868. Catalogue of the Fishes in the British Museum. Vol. 7: XX + 512 pp. MoNOD, T. 1968. Le complexe urophore des poissons teleosteens. Mem. Inst. Fond. Afrique Noire (Dakar), No. 81. 705 pp. Poll, M. 1965. Contribution a 1 etude des Kneriidae et description d'un nouveau genre, le genre Parakneria (Pisces, Kneriidae). Mem. Acad. roy. Belgique, I. Sci, oct., 36(4): 1-28, 13 pis. Ramaswami, L. S. 1953. Skeleton of cyprinoid fishes in relation to phylogenetic studies. 5. The skull and gasbladder capsule of the Cobitidae. Proc. Nat'l. Inst. Sci. India, 19(3): 323-347. Rendahl, H. 1930. Einige Bemerkungen iiber den Schultergiirtel und die Brustflossenmuskulatur einiger Cobitiden. Ark. Zool., 21(16): 1-31. 1933a. Studien iiber innerasiatische Fische. Ark. Zool., 25(11): 1-51. 1933b. Weitere Untersuchungen iiber den Schultergiirtel und die Brustflossenmuskulatur der Cobitiden. Ark. Zool., 25(10): 1-38. Rosen, D. E., and P. H. Greenwood. 1970. Origin of the Weberian apparatus and the relationships of the ostariophysan and gonorynchi- form fishes. Amer. Mus. Novitates, No. 2428: 1-25. SwiNNERTON, H. H. 1903. The osteology of Croineria nilotica and Galaxias attenuatns. Zool. Jahrb. (Jena), anat. abth., 18: 58-70. Vaillant, L. 1902. Resultats zoologiques de Texpedition scientifique neerlandaise au Borneo Central. Poissons. Notes Leyden Mus., 24(1): 1-166, 2 pis. Weber, M., and L. F. de Beaufort. 1916. The fishes of the Indo- Australian Archipelago (Leiden, E. J. Brill), Vol. 3: XV + 455 pp. LIHRARY MAR 1 8 1985 lARVARD B R E V I O R^A MMsemoi of Coioiparative Zoology Cambridge, Mass. 25 February, 1972 Number 385 THE CHANARES (ARGENTINA) TRIASSIC REPTILE FAUNA. XII. THE POSTCRANIAL SKELETON OF THE THECODONT CHANARESUCHUS. Alfred Sherwood Romer Abstract. A description is given, with a reconstruction, of the post- cranial skeleton of the long-snouted thecodont Clutnaresiichus bonapcirtei from the early Middle Triassic Chaiiares Formation of Argentina. The general proportions are comparable to those of crocodilians, and life habits may have been similar. However, there is no positive indication of rela- tionship to either later Crocodilia or Phytosauria, and Chanaresiichiis and its relatives ( Proterochampsidae) appear to be a sterile offshoot of the primitive proterosuchian stock. Dermal armor is little developed; the pelvis, although somewhat advanced in build, shows no trend toward the crocodilian elimination of the pubis from the acetabulum; the pes is of a specialized nature, with emphasis on the inner digits. INTRODUCTION In a recent paper in this series (Romer, 1971), I described the cranial anatomy of two long-snouted thecodonts, Chanaresuchus bonapcirtei and Gualosuchus reigi, from the (? Anisian) Triassic Chaiiares Formation of Argentina. Few postcranial remains of Gualosuchus are available to me. For Chanaresuchus, however, a number of specimens from our 1964-65 expedition are at hand, and Sr. Bonaparte allowed me to study several further specimens in the Instituto Lillo of Tucuman collections. Below, I give a description, with restoration, of the postcranial skeleton of Chanaresuchus. Materials. The following specimens from the La Plata- Harvard expedition include postcranial remains: The holotype: (La Plata Museum 1964-X1-14-12) Skull, partial column, a few limb bones. 2 BREVIORA No. 385 MCZ 4035. Part of the contents of a large nodule, including a column articulated to caudal 3, and considerable girdle and limb material. MCZ 4036. The remainder of material in this nodule, in- cluding incomplete and in part poorly preserved remains of skulls and postcranial materials of two further individuals. MCZ 4037. A skull and presacral column. MCZ 4038. A slab containing a melange of materials; most, however, appear to belong to a nearly completely disarticulated Chanaresuchus skeleton. There is considerable variation in the size of the individuals concerned. Largest is MCZ 4037, in which skull length to the quadrate is 260 mm. In the type skull this measurement is 211 mm. My figures of postcranial material (including restoration) arc based mainly on MCZ 4035, in which postcranial vertebrae and limb bone measurements indicate a size approximately that of the type. The two skulls somewhat imperfectly preserved in MCZ 4036 measure about 230 and 220 mm. The imperfect skull with which the array of postcranial material on MCZ 4038 appears to be associated had a length of approximately 165 mm. Several specimens in the Instituto Lillo collections were studied. Best was a specimen with articulated column extending to the proximal part of the tail, and much of the appendicular skeleton; the skull measured, to quadrate, 175 mm, and the individual was thus considerably smaller than the type. A second specimen in- cluded much of the postcranial skeleton of a moderately large individual; the skull is missing, but comparison of limb bone measurements indicate a size shghtly larger than MCZ 4035. A third Tucuman specimen is an immature individual, in which post- cranial measurements are about half those of the type; preserved is a hind foot exactly duplicating the peculiar structure, and a nearly complete tail. Vertebral column. The atlas-axis complex, best seen in MCZ 4037 (Fig. 1, a, Z?), is of a type readily derivable from that of primitive reptiles. The atlas neural arches are paired. Dorsally a posterior extension was applied to the lateral surface of the axis arch; an anterior extension is flattened ventrally, presumably for articulation with the exoccipital (no pro-atlas has been found). A slight rugosity seen on one specimen indicates a rib attachment. Ventrally the stem of the arch extends stoutly downward to 1972 CHANARESUCHUS Figure 1. Vertebrae and rib of Chanaresuchus bonapartei. Composite, mainly from MCZ 4036. a, lateral view of vertebrae 1-3; b, anterior view of atlas-axis; c, d, lateral and anterior views of an anterior dorsal vertebra; e, lateral view of a posterior dorsal; /, an anterior dorsal rib, in postero- ventral view, x 1. articulate with the atlas intercentrum; internally it is excavated to form the dorsolateral segment of the socket for reception of the skull condyle. The lower part of this socket is formed by the atlas intercentrum, a stoutly developed crescent that articulates broadly on either side with the neural arches and is bevelled 4 BREVIORA No, 385 anteriorly for condyle articulation. Behind the atlas intercentrum lies the axis intercentrum, a thin crescent connecting ventrally the atlas intercentrum and the axis centrum. The atlas centrum is a stout element, which forms the deeper part of the condylar socket (slightly indented centrally by a notochordal pit); dorsally it forms the anterior end of the floor of the neural canal; posteriorly it is firmly apposed to the axis centrum. It is partially obscured in lateral view by the atlas arch and intercentrum and ventrally by the axis intercentrum. The axis neural arch is relatively low but long. As noted, the articulation of the atlas arch is simply a flattened area on the lateral surface of the arch, anteriorly; posteriorly, normal post- zygapophyses are present. A slight lateroventrally directed process, rugose at its tip, is an incipient transverse process for the tubercular articulation of the rib. The parapophysial articulation of the rib capitulum is indicated by a pronounced rugosity far down the side of the centrum and close to its anterior border. In contrast to the intercentrum anterior to it, the axis centrum (like the cervical vertebrae that follow it) is keeled ventrally. Presacral column (Fig. 1, c-e). MCZ 4035 includes a pre- sacral column and sacrum, showing definitely the presence of 23 presacral vertebrae; MCZ 4037 includes, in two segments, the 23 presacral vertebrae, the column having presumably broken off at the sacrum. Most of the description that foflows is drawn from these two specimens; unfortunately, in much of both columns the transverse processes are broken and incomplete. The column is, for the most part, of a typical thecodont character. In MCZ 4037 the length of typical dorsal segments, measured along the centra, averages 16 mm; in MCZ 4035, a somewhat smaller individual, this measurement averages 14 mm. The neural spines are centered somewhat to the back of the mid- length of the centrum; they are broadened anteroposteriorly, expanding in width above the region of the zygapophyses. The anterior spines are relatively tall and narrow; posteriorly, in the lumbar region, they are lower and broader distally. They are narrow in transverse diameter, without noticeable expansion at the top. When the surface is well preserved, faint longitudinal striations are frequently seen. In the larger specimens there appears to have been a late continuation of growth (or rather ossification) of the spines, the neural spines in larger forms being 1972 CHANARESUCHUS 5 taller in relation to other dimensions of the vertebrae. For example, in the large specimen, MCZ 4037, the total height of a posterior cervical (including both arch and centrum) is approxi- mately three times the length of the centrum; in the smaller MCZ 4035, the height is barely above two and one-third times the central length. The posterior zygapophyses are situated directly below the posterior margins of the neural spines; the prezygapophyses, on the contrary, lie well in advance of the anterior margins of the neural spines, on arch processes running upward and forward from the anterior margins of the transverse processes. The articular surfaces of the posterior zygapophyses face somewhat outward as well as downward; the anterior zygapophysial surfaces, in contrast, face inward as well as upward. Beginning with the axis, transverse processes are developed, their bases centering at a point well down on the side of the arch and toward the front. They are directed slightly posteriorly and ventrally rather than directly laterally. The axis process projects but 2 mm or so from the surface of the arch; posteriorly, the lengths increase so that, for example, by vertebra 7 in 4037, the length is close to 10 mm and there is obviously further increase more posteriorly, although, to my regret, the processes of the posterior dorsal and lumbar regions are mostly broken and imperfect. In MCZ 4035 the transverse processes as far back as vertebra 7 terminate in a narrow tip; posteriorly the transverse processes are broadened anteroposteriorly for their entire length, with oval distal facets for tubercular attachment. The anterior margin of the arch curves almost directly down- ward from the process supporting the anterior zygapophysis to the anterior margin of the centrum; posteriorly the lateral margin of the arch is strongly concave in outline, allowing a large gap between arches here for passage of nerves and vessels. The centra are smoothly oval in end contours, with a height considerably greater than the width. The centra are amphicoelous, with thickened rims at either end. Between the two ends of the centrum, the element pinches in somewhat laterally, and in side view the lower margin is somewhat concave in outline. The anterior centra are distinctly keeled ventrally; at about the be- ginning of the dorsal region the keel fades out, and posteriorly the ventral surface is smoothly rounded. . 6 BREVIORA No. 385 In the axis the parapophysis is merely a rugose area far down the anterior margin of the centrum. Posteriorly the parapophysis gradually moves upward along the anterior margin of the centrum and slowly comes to project distinctly from the surface of the bone so that, for example, by presacral 10 it has risen nearly to the top of the front margin of the centrum and projects outward for several millimeters. Posteriorly beyond this point the para- pophysis continues movement upward and backward toward the transverse process and a thin ridge develops connecting the two (as in some primitive reptiles). Regrettably the processes are but imperfectly preserved in the "lumbar" region of materials available to me, but it appears that in the last four or so presacrals the two processes are united. Sacral vertebrae. The two sacral vertebrae arc present in MCZ 4035, and are present also in a Tucuman specimen. They differ mainly from the vertebrae anterior to them in the broad base from both centrum and arch from which the apophyses supporting the sacral ribs arise. Caudal vertebrae. In the material available to me at Harvard, there is little identifiable caudal material; even in specimen MCZ 4035 articulated vertebrae cease at the third caudal. In the Instituto Lillo material, however, one specimen shows twelve caudal vertebrae in articulation with the sacrals and eleven are present in a second specimen. There is little decrease in vertebral length in these series of proximal caudals. Immediately behind the sacrum, however, the neural spines begin to show a backward slant and show a decrease in breadth distally; in the end members of these series, the neural spine is reduced to a small elevation lying above the postzygapophyses. Transverse processes are well developed; broad at their bases, they extend laterally and sHghtly ventrally to become pointed at their tips. In a Tucuman specimen of rather small size (skull length 175 mm), the lengths of the transverse processes of the anterior caudals, measured from the mid-line, are, in sequence, 29, 30, 28+, 33, 29, 25 and 20 mm. In this specimen the first chevron preserved lies between the fifth and sixth centra. That this is truly the first of the series is suggested by the fact that there is little space between the more anterior centra for insertion of a haemal arch, whereas more posteriorly a sufficient ventral gap is present. That the chevrons extended far down the tail is indicated by the fact that in the small 1972 CHANARESUCHUS 7 specimen mentioned below, they are seen in the region of the 20-23rd caudals. In an Instituto Lillo specimen considerably smaller than "adult" specimens of Clumaresuclnis but definitely representing a young individual of this form, 35 articulated caudals arc present, plus a few disarticulated elements beyond. In its present condition, little detail can be made out, but the last elements appear to be subterminal in nature, and a length of 40 or so caudals seems to be indicated. Ribs. Regrettably, there is no available specimen with a good set of articulated ribs. In two instances a single rod-shaped structure that appears to be a single-headed atlantal rib, is present at the anterior end of a column. Back of the atlas, the ribs are distinctly double headed in typical archosaur fashion. In the anterior part of the presacral column, the two heads are widely separated; in correlation with the rapid development of trans- verse processes as we progress backward, the capitular branch is much longer than the tubercular, curving downward and inward from the line of the shaft. A short series of articulated vertebrae in MCZ 4038, representing an individual somewhat smaller than the "adults," bears a sequence of cervical ribs, presumably be- ginning with the atlas; the lengths are: 6, 12, 26, 32, 52, 60, ?, 75 ± and 85 mm. The more anterior, at least, of this series have pointed tips, indicating that they lay anterior to the point of rib connection with the (presumably cartilaginous) sternum. There is no indication in these cervical elements of an anterior "spur" from a point near the head of the rib that is found in crocodilians, and is highly developed in certain other thecodonts. Of ribs in the dorsal region, remains are sparse. A number of rib heads show some variation in the amount of separation of tubercle and capitulum, those in which the two are less widely separated presumably coming from the posterior part of the region, in which transverse processes and parapophyses are approaching one another. In MCZ 4037 a complete rib (Fig. 1,/) is present close behind the eighth vertebra and may belong to that segment. Regrettably, no ribs are preserved in association with vertebrae of the "lumbar" region, in which shortened ribs were presumably present. For most of its extent the rib is slender; proximally, however, there is a slight expansion in the form of a thin flange posterior to the main proximal end of the shaft, which runs to the capitulum. 8 BREVIORA No. 385 1972 CHANARESUCHUS Figure 2. Chanarcsuchus bonapartei, girdle and limb bones. Composite, but mainly from MCZ 4035. a. ventral view of pelvic girdle; b, dorsal view of pelvic girdle and sacral ribs; c, pelvic girdle from right side; d, left humerus, dorsal surface; e, right scapulocoracoid (dorsal end of scapula at left); /, right hind foot, calcaneum incompletely preserved; g, dorsal and lateral views of left femur; /;, extensor surface of left tibia; /, extensor surface of left fibula. X Vz. 10 BREVIORA No. 385 Two sacral ribs are present in MCZ 4035, although some details are obscured. They extend outward from a broad base, including much of the lateral surface of the centrum as well as the lower part of the neural arch. They expand somewhat distally and are blunt-ended; as indicated by several specimens of ilia, they were not too firmly attached to the articular areas for them on the internal iliac surface. Girdles. Several examples of scapulae and coracoids are present in available material; that illustrated (Fig. 2, e) pertains to MCZ 4035. The two elements of the endochondral girdle are not tightly sutured to one another, and were found separated in two instances. The scapula is tall and slender, somewhat broadened at its upper margin, which is a "finished" one in the specimens studied. At its lower end the scapula is somewhat expanded anteriorly and its margin curved outward, presumably for clavicu- lar attachment. The posteroventral portion of the bone is thickened and bears externally the well-defined upper margin of the glenoid cavity, which faces posterolaterally. Scapula and coracoid meet in a long horizontal suture. The latter bone is an oval plate, thickened posterodorsally to bear the lower part of the glenoid, and having the usual coracoid foramen. The re- mainder of the bone is relatively thin and convex in contour externally, the lower portion obviously turning medially across the chest. There is no trend toward posteroventral expansion, such as would be expected in a possible crocodile relative. Presumably clavicles and interclavicles were present, but I have not been able to identify such elements in the material available. A number of specimens include pelvic girdle material, so that the nature of the ilium, ischium, and the anterodorsal portion of the pubis can be readily recognized. In no case, however, is there present a complete, well-preserved pelvis with the elements in proper relations with one another. Closest to this condition is that associated with MCZ 4035, and my figured restoration (Fig. 2, a-c) is based on this specimen. Here the proper position of the ilia can be determined by their relation to the preserved sacral vertebrae and ribs; the ischia are for the most part well preserved; the pubes, however, are imperfect. The ilium is of the primitive type seen in many thecodonts — essentially a simple vertical blade, extending posteriorly in tapering fashion some distance beyond the region of the acetabulum, the 1972 CHANARESUCHUS 11 front margin curving downward not far in advance of the aceta- bulum. On the inner surface are well-marked depressions for attachment of the two sacral ribs. On its outer surface the ilium forms the upper half of the acetabular cavity. This depression is most deeply incised posteriorly, where it reaches its most dorsal position, to be bordered dorsally by a sharp outwardly extending ridge on the ilium; thence the upper border, less sharply defined, slants forward and downward a considerable distance before reaching the pubis. The pubis forms the anteroventral margin of the acetabulum; this cavity is here not as sharply incised into the bone as is true of its iliac and ischiadic margins, and the pubis is not greatly thickened here. The anterior surface of the ilium adjacent to the acetabular border is somewhat thickened; this surface continues downward to form a stout rounded ridge along the anterior margin of the pubis, the ridge turning ventrally in its distal portion. Medial to this ridge the pubis extends as a thin sheet to the mid- line of the body, where it meets its fellow in a long symphysis. Although the condition of the specimens makes the matter some- what uncertain, it would seem that the anterior portion of this broad sheet of bone turns somewhat ventrally, while the posterior part of the symphysial region lies in a horizontal plane. At the posterior end of this ventral expanse of bone, the pubic symphysis terminates and the posterior margins of the two pubes retreat upward and outward toward the acetabular region. The exact nature of the posterior margin of the bone and the distance it retreats dorsally before making contact with the ischium is uncertain. The usual obturator foramen penetrates the pubis not far below the acetabular border. The ischium forms the posteroventral sector of the acetabular border. The dorsal portion of the bone is here greatly thickened, forming a horizontal shelf with a crescentic outer margin, to constitute a prominent segment of the acetabular cavity. Below this area the ischium rapidly decreases in thickness, but increases in anteroposterior breadth ventrally. The posterior margin of the bone, slanting medially, descends nearly straight downward for some distance, then curves posteriorly and, finally, ventrally to meet its fellow in symphysis. This symphysial union extends forward in a nearly horizontal plane to a point somewhat posterior 12 BREVIORA No. 385 to the level of the puboischiadic area of contact below the aceta- bulum. Here, the symphysis terminating, the anterior margin of the ischium ascends nearly straight upward and laterally toward the acetabular region. We thus have, in Chanaresuchus, a definite advance in pubic structure over the proterosuchian condition in the presence of a broad ventral "incision" in the primitive pubo- ischiadic plate and a marked break in the primitively continuous ventral symphysis. Pectoral limb. Little material of the pectoral limb is available. A humerus, 85 mm in length, was found with MCZ 4035 (Fig. 2, d), but is not too well preserved. It is expanded at both ends in primitive reptilian fashion, but shows little of the primitive "twist" of one end on the other. The posterior margin of the "head" segment curves medially to the proximomedial corner of the bone, whence the proximal articular surface, facing dorso- medially, curves laterally around the head of the bone. At the proximal end of the lateral margin a typical deltopectoral crest is present (in crushed condition). The distal end of the bone is distinctly divided, both dorsally and ventrally, into medial and lateral condyles, the medial the stouter, with distal articular surfaces for radius and ulna. Neither of the foramina often present distally in the humerus of early reptiles is present. I cannot with confidence identify either radius or ulna in any of the adult Chanaresuchus specimens studied. They are present, however, in association with a humerus in the small specimen in Tucuman, mentioned above. As preserved in this obviously im- mature individual, they are merely elongate bony cylinders, with some expansion proximally of the element which I take to be the ulna, but without indication of an olecranon. These elements, measuring 32 and 33 mm in length, are associated with a humerus measuring 36 mm in length. If the same proportions were to hold in MCZ 4035, the radius and ulna would be expected to measure 74 and 76 mm in length respectively. Pelvic limb. The series of three major elements of the hind legs — femur, tibia, and fibula — are preserved in articulation in three instances, and a number of isolated hind leg elements are present on MCZ 4038. In the three articulated specimens the lengths of these elements as preserved are 128 mm, 94 mm, and 95 mm in MCZ 4035; 15 1± mm, 129 mm, 91+ mm in the leg 1972 CHANARESUCHUS 13 ^^nr^^^^'^^^ LQ-j yi^\ m "41?: Plate 1. Skeletal restoration of Chanaresiichiis bonapartei. Composite, but skull from the holotype, post-cranial skeleton mainly from MCZ 4035. Manus and dermal shoulder elements unknown; ribs imperfectly preserved and those of "lumbar" region unknown. About V3 size of type and MCZ 4035. 14 BREVIORA No. 385 of MCZ 4036 with articulated foot, 150± mm, 135 mm and 117+ mm for another specimen in the same block. As noted elsewhere, the skeleton of MCZ 4035 is apparently somewhat immature; the limbs in MCZ 4036 appear to be those of essentially mature individuals. In all three articulated examples the femur is definitely longer than the tibia; there is, however, a very consider- able diflerence between the examples, in MCZ 4036 where the femoral length is 1 1 1 percent and 117 percent of the tibia, and MCZ 4035, where the figure is 135 percent. The femur (Fig. 2, g) is of a pattern seen in many thecodonts and preserved little changed in crocodilians. The bone is relatively long and slender, the ends little expanded. In side view, the shape is sigmoid, with the proximal end curved upward and somewhat medially, the distal end curved somewhat downward; the curvature, however, is less extreme than in more advanced thecodonts. The proximal end of the bone is much modified, so that the morpho- logically medial (or anterior) margin is turned upward, the lateral border downward. The curved head of the bone is thickened, the thickening increasing toward the medial (upper) border; preserva- tion in available material, however, is not good enough to clearly outline the articular area. About two-fifths the distance down, the bone is extended ventrally, with, medially, a large oval area that probably (as in crocodilians) lodged the insertion of the long caudifemoral muscle; the projecting lower margin is a trochanter that presumably afforded insertion to the caudifemoralis brevis. The bone broadens somewhat distally, with a partial division into condyles, a shallow intercondylar fossa above, and a ventral popliteal depression. The distal end of the bone is unossified in MCZ 4035, so that details of areas of articulation cannot be made out. Tibia and fibula, again, are of typical primitive archosaur construction (Fig. 2, h, i). The broad head of the tibia is essen- tially triangular in section; there is little development of a cnemial crest, which is represented merely by the top of a ridge that descends much of the length of the bone, separating a lateral- facing surface of the shaft from one facing anteromediaUy. Terminally, there is a broad oval surface, somewhat convex, for articulation with the astragalus. The fibula is, as expected, a relatively slender element, somewhat expanded proximally for appos tion to the femur; distally (in contrast with the situation in 1972 CHANARESUCHUS 15 many reptile groups) there is only a minor expansion in diameter of the bone to accommodate the oval terminal articulation with the calcaneum (and to a much lesser degree with the astragalus). Tarsal elements are present in MCZ 4035 and 4036. In 4035 the astragalus is well preserved (Fig. 2,/). Dorsally it bears a large articular area for the tibia, gently concave, elongate medio- laterally and facing somewhat anteriorly as well as dorsally, and dorsolateral to this a smaller facet for the fibula. No calcaneum was preserved with this specimen. MCZ 4036, on the other hand, lacks the astragalus, but has an incompletely preserved calcaneum. This shows a posterolateral tuber of "crocodilian" type, but the main body of the bone is imperfect. In both specimens mentioned distal tarsals 3 and 4 are present as flattened ovals lying over the heads of the lateral metapodials. There is no trace of more medial distal elements, although they may have been present in cartilage. The structure of the pes is the most distinctive feature of Chanaresiichiis (Fig. 2,/). Much of the right foot is present in MCZ 4035, but digit IV is represented only by a metatarsal splint; a right foot complete except for the ungual of toe IV is present in MCZ 4036, but the proximal ends of the inner toes are obscured by a refractory matrix. Two closely comparable specimens of the Chanaresuchus foot are present in the Tucuman material. In archosaurs generally the trend in foot construction has been for a retention of the primitive phalangeal formula (except for frequent reduction of the fifth toe), but the development of a symmetrical pattern, with toe III the longest, toes II and IV some- what shorter but subequal in length, and toe I short. This is true of the Chanaresuchus foot; but whereas in most archosaur toes II-IV are subequal in development, here there is very strong emphasis on the inner toes, and digit IV, on the contrary, is very slender. Digit I is somewhat shortened, but very stoutly built. Digit II is massive. Digit III is relatively slim. Digit IV is slender, almost sphntlike in structure. Digit V is represented only by a metatarsal spur. In some other archosaurs, such as certain crocodilians, Ticinosuchus, Euparkeria and Stagonolepis, there is a modest trend toward strength in the more medial digits, but never to an extent approaching the condition seen here. A similar trend (but usually a much less extreme one) towards a strengthening of the inner toes is found in mosasaurs and a number of chelonians. These 16 BREVIORA No. 385 are aquatic forms, and this specialization is apparently an "im- provement" in paddle-action — a fact that gives strength to the assumption that Chanaresuchiis was in great measure a water dweller. It has been assumed that Chanaresuchiis and its relatives in the Proterochampsidae might be antecedent to the Crocodilia, but such strong emphasis on the inner toes is not to be expected in an ancestor of the group. As regards possible phytosaur relationship, foot material of that group is rare, and the pes, when restored, is generally based on poor material. I am indebted to Dr. Chatterjee, of the Geological Study Group of the Indian Statistical Institute, for a figure of a well-preserved foot of a phytosaur from the Maleri Formation. This shows almost no trace of emphasis on the inner toes, and relationship of the Proterochampsidae to the phytosaurs seems highly improbable. Dermal armor. No gastralia are present in articulated fashion in any specimen. In contrast to a large fraction of the thecodont assemblage, other body armor appears to have been feebly developed, and consisted merely of a single row of thin scales lying over the neural spines. As preserved in several specimens, scales are definitely present from the axis to the last presacral; there is no evidence as to whether they extended onto the tail region, although this was probably the case. The scales are wedge-shaped as seen from above, narrow anteriorly, broader posteriorly, with a low longitudinal median keel. The anterior end of each scale underlies that anterior to it. There are approxi- mately three scales the length of each vertebra, although there is no apparent relationship between vertebral segmentation and scale arrangement; in one specimen 57 scutes were present above the first 20 vertebrae. Their appearance and arrangement is very similar to that pictured in Ticinosuchiis by Krebs (1965, fig. 8). This author restores the dorsal scales in the presacral region in a double row, in analogy, I suspect, with the pattern seen in various other thecodonts. I doubt if this was the case. The scales pre- served in the presacral region number only about enough to make up a single median row. The presence of scales above the tail vertebrae in Ticinosiichus suggests that, were appropriate materials to be discovered, they would be found to continue along the tail in Chanaresuchiis as well. Restoration. In the accompanying plate I have attempted a skeletal restoration of Chanaresuchus at one third the natural size 1972 CHANARESUCHUS 17 of a typical adult. Much of the skeleton is taken from MCZ 4035. The dermal shoulder elements, the manus. and the posterior ribs are restored, and the tail is imperfectly known. Despite these lacunae and the fact that in ribs and certain other elements the restoration is composite, 1 believe that the restoration gives a fairly accurate picture of the animal's skeleton. In restoring an early archosaur, an important question is whether the pose was that of a biped or a quadruped. In all primitive reptiles the front legs are somewhat shorter and less massive than the hind. In most archosaurs this disproportion in limb lengths is usually much greater and led to a general (but not universal) belief (which I shared) that archosaurs ab initio were more or less bipedal in locomotor tendencies, and that most quadrupedal archosaurs, such as the sauropods and various ornithischians, had relapsed from a bipedal mode of progression back to progressing on all fours. Charig, Attridge and Crompton (1965) have, as a result of a study of probable sauropod history, come to the con- clusion that quadrupedal pose was primitive for archosaurs, and that the development of powerful hind legs and a highly developed tail were, to begin with, adaptations for an amphibious life, and that bipedalism arose later. It is not unreasonable to believe that, as these authors argue, the sauropods developed without passing through a bipedal stage. I think, however, that on present evidence, the primitive orni- thischians were at least partially bipedal, although a majority of the ornithischian subgroups later became quadrupedal. The possibility that the ancestral archosaurs were amphibious leads to speculation as to the general nature of early reptilian history. I have argued (although not without strenuou; oppo- sition ) that the ancestral reptiles, although having acquired an amniote style of development, were still amphibious, and perhaps even mainly aquatic in habits. Of the great group of synapsids, which were dominant in the Permian and earlier Triassic, most became fully terrestrial, but the most primitive (and oldest) synapsids, the ophiacodont pelycosaurs of the Pennsylvanian and early Permian, were still essentially aquatic fish-eaters. Ancestral Permian archosaurs are almost unknown, but they may have followed a similar life pattern, but for a longer period of time. Further, we have no undisputed knowledge of the early ancestors 18 BREVIORA No. 385 of the euryapsid sauropterygians and placodonts or of the ichthyo- saurs; at their earHest appearance these reptiles were aquatic — and perhaps primitively so. Mesosaurus, of the Permo-Carbo- niferous boundary, shows that if the primitive reptiles had be:ome terrestrial, reversion to the water took place at a very early stage of reptilian history. And while I do not want to even suggest any close relationship between archosaurs and Mesosaurus (whose phylogenetic relationships are quite problematical), the similar adaptations in this genus and in the archosaurs in the powerful tail and highly developed hind limbs are strikingly suggestive. But to return from this discussion to the question of limb dis- parity and body pose in archosaurs. Correlation between limb proportions and presumed posture is none too simple a matter. Such obviously amphibious and quadrupedal forms as crocodilians and phytosaurs have front legs nearly as well developed as hind; in a "sample" crocodilian, for example, the humerus plus radius are about 84 percent the length of femur plus tibia, and in a phytosaur described by McGregor (1906), the figure is 87 percent. But forms that seem quite surely quadrupedal may show a con- siderably greater disparity in length between front and hind legs. The heavily armored aetosaurs, for example, are universally con- sidered to be quadrupeds, but in Aetosaurus and Stagonolepis, according to figures given by Walker (1961), the front legs are but 61 percent and 64 percent the length of the hind. Protewsuchus [Chasmatoscmnis] is an early and surely primitive thecodont that is reasonably regarded as amphibious in habits although the available materials suggest that the front legs were but about 66 percent the length of the hind. On the other hand, Euparkeria, Hesperosuchus, and Ornitho- suchus are regarded by those who have studied them (Ewer 1965, Colbert 1952, and Walker 1964) as bipedal; and yet their hmb ratios are as high or higher than some of the presumed quadrupeds, being 70 percent, 67 percent, and 70 to 75 percent, respectively. Obviously limb proportions in themselves are not sufficient to enable one to give a firm conclusion as to pose; body structure in general must be taken into consideration. As regards Chanaresuchus, the Umbs are very markedly dis- proportionate. If the radial length in MCZ 4037 was that esti- mated in relation to the humerus, the front leg would be but 53 percent that of the hind. However, the humerus of this specimen 1972 CHANARESUCHUS 19 may be incomplete distally. If, instead, we take the small specimen from Tucuman in which all four bones conceived are present (with lengths of 39 mm for humerus, and 32, 59, and 59 for radius, femur, and tibia), we get a somewhat higher figure of 61 percent. These figures taken by themselves would seem to strongly suggest bipedality. But in other regards there is considerable reason to believe that Chanaresuchus was amphibious, spending much of its time in the water, where the body would be in a horizontal position; and while the animal could quite probably assume a semi-erect bipedal posture, this would only occur at such times when, ashore, speed seemed imperative. In consequence, I have restored the animal as a quadruped. Relationships. In the preceding paper of this series, I have commented on possible relationships of Chanaresuchus and its relatives in the Proterochampsidae. Reig (1959) and Sill (1967) have argued for Proterochampsa as a pre-crocodile; Walker (1968), on the other hand, maintains that this genus is a pre- phytosaur rather than a pre-crocodilian, but maintains that its "cousin" Cerritosaurus is pre-crocodilian. Study of skull structure led me to believe that there was little in the cranial anatomy of Chanaresuchus to suggest relationships to either Crocodiiia or Phytosauria, and that this group, the Proterochampsidae, was essentially a sterile one, representing a modest advance over the Proterosuchia, which in middle Triassic days occupied a position in the ecology similar to that of the phytosaurs of the late Triassic and the crocodilians of the later Mesozoic. Study of the postcranial anatomy does not lead me to change my previous conclusions. The postcranial skeleton is for the most part of a primitive and generalized archosaurian type; the only advances over the protero- suchian condition lie in the fact that the pelvis has progressed to a more typical archosaurian condition in the "cleavage" between pubes and ischia for much of their height, with interruption of the primitively continuous ventral symphysis, and in the initiation of body armor not found in known proterosuchians. The armor is very lightly developed — much less than we would expect in an Anisian(?) predecessor of either crocodilians or phytosaurs. The pubis shows no indication of a trend toward exclusion from the acetabulum, as one might expect in a crocodile ancestor. The cervical ribs lack the specialization seen in crocodilians (and certain other thecodonts) of an anterior "spur" near the head. 20 BREVIORA No. 385 Still further, the pes is of a highly specialized nature, not of a type antecedent to either crocodilians or phytosaurs. Giialosiichus postcranial material. I may note that a small amount of postcranial material was found with the type skull of Gualosiichus; this included a scapulocoracoid, femur, parts of two tibiae, and a few bones of the pes. The elements preserved are in general comparable, except for larger size, to those of Chanaresu- chus. The anterior border of the scapula is less everted than in Chanaresuchus. The height of the scapulocoracoid is 148 mm, the femoral length 158 mm. The femoral length is close to 50 percent the overall length of the skull, indicating that the Gualosii- chus skull was rather shorter in proportion to body measurements than that of the relatively long-snouted Chanaresuchus. Further postcranial remains of Gualosuchus, which I trust will be described presently by Sr. Bonaparte, are present in the Instituto Lillo col- lections. In one specimen in that collection the humerus is some- what less than two-thirds the length of the femur, thus indicating the same disproportion in limb length as in Chanaresuchus. For aid in the collection, preparation, and publication of the materials described above (as in other numbers in this series), I am deeply indebted to grants GB8171 and 22658 from the National Science Foundation. LITERATURE CITED Charig, a. J., J. Attridge, and a. W. Crompton. 1965. On the origin of sauropods and the classification of the Saurischia. Proc. Linn. Soc. London, 176: 197-221. Colbert, E. H. 1952. A pseudosuchian reptile from Arizona. Bull. Amer. Mus. Nat. Hist., 99: 565-592. Ewer, R. F. 1965. The anatomy of the thecodont reptile Euparkeria capensis Broom. Phil. Trans. Roy. Soc. London, ser. B, 248: 379-435. Krebs, B. 1965. Ticinosuchus ferox nov. gen. nov. sp. Ein neuer Pseudosuchier aus der Trias des Monte San Giorgio. Basel, Birk- hauser Verlag, 140 pp. McGregor, J. H. 1906. The Phytosauria, with especial reference to Mystriosuchus and Rhyddodon. Mem. Amer. Mus. Nat. Hist., 9: 29-101. Reig, O. a. 1959. Primeros datos descriptivos sobre nuevos reptiles arcosaurios del Triasico de Ischigualasto (San Juan, Argentina). Rev. Assoc. Geol. Argentina, 13: 257-270. 1972 CHANARESUCHUS 21 RoMER, A. S. 1971. The Chanares (Argentina) Triassic reptile fauna. XI. Two new long-snouted thecodonts, Chanarcsiichus and Gualo- suchits. Breviora. Mus. Comp. Zool., No. 379: 1-22. Sill, W. D. 1967. Proterochampsa harrionuevoi and the early evolution of the Crocodilia. Bull. Mus. Comp. Zool., 135: 415-446. Walker, A. D. 1961. Triassic reptiles from the Elgin area: Stagonolepis, Dasygnathiis and their allies. Phil. Trans. Roy. Soc. London, ser. B, 244: 103-204. 1964. Triassic reptiles from the Elgin area: Ornitho- suchus and the origin of carnosaurs. Phil. Trans. Roy. Soc. London, ser. B, 248: 53-134. 1968. Protositchus, Proterochampsa, and the origin of phytosaurs and crocodiles. Geol. Mag., 105: 1-14. ^^ f^^L'S. CGMP. ZOOL ^^'^ LinnARY MAR 1 8 1985 B R E V I O ft-^ MmseuiM of Comparative Zoology Cambridge, Mass. 25 February. 1972 Number 386 A REVIEW OF THE AUSTRALIAN SPECIES OF ELAPHROSYRON AND TELOSTEGUS, WITH NOTES ON OTHER GENERA (HYMENOPTERA, POMPIUDAE) Howard E. Evans Abstract. The genera EUiphrosyron and Tcloslci^'iis are reported fiom Australia for the first time, and a brief review of the species is presented. The following new species are described: E. sociiis. T. thomisivoriis. and T. tiinicri; T. nii^rocinerascens (Turner) is also redescribed. Notes are presented on the genera Ponipiliis. Ctcnoitciinis. and Fahriogeuia, and the following new species described: P. hclanloo, C. warragai. F. Canberra, and F. dilga. Notes on the nesting behavior of these and other species will be presented separately. During the summer of 1969-1970, Robert Matthews and I made a number of observations on the nesting behavior of Australian spider wasps (Pompilidae). Since the pompilid fauna of that continent has been little studied, we experienced much difficulty in identifying the species we worked on. Study of the types of Smith's and Turner's species in the British Museum has clarified many of our problems, but there remain several previously unnamed species for which we have field data that we consider worth publishing. Some of our most interesting data pertain to the related genera Elaphrosyron and Telo.stei^iis (neither previously reported from AustraHa), and 1 present here a review of the Australian members of those genera. Notes and descriptions relating to three other genera are also appended. The material considered here will be deposited in the following museums, which are abbreviated in the text as follows: AMS: The Australian Museum, Sydney ANIC: Australian National Insect Collection, Canberra BMNH: British Museum (Natural History), London MCZ: Museum of Comparative Zoology, Cambridge WAMP: Western Australian Museum, Perth 2 BREVIORA No. 386 Genus Elaphrosyron Haupt Eluphrosyron Haupt, 1929, Mitt. Zool. Miis. Berlin, 15: 120 (type-species: E. heinrichi Haupt). — Arnold, 1937, Ann. Transvaal Mus., 19: 40-43 (African species). Protelostegiis Priesner, 1955, Bull. Soc. Ent. Egypt, 39: 167 (type-species: P. anioldi Priesner). New synonymy. Both Haupt and Arnold regarded this genus as closely related to Episyron, with which it shares several important features, most notably the bifid claws and the longitudinal folding of the wings along a vena spuria. However, the postnotum is shaped differently and there is no evidence of squamose pubescence on the body. The genus is poorly known but is apparently confined to warmer parts of the eastern hemisphere. The type species was described from the vicinity of the Caspian Sea, and Arnold recognized one species from central Africa and another from southern Africa. The genus has not previously been reported from Australia. There appears to be only one species on that continent, but it is widely distributed and locally common. Arnold (1937) provided an accurate generic diagnosis, and there seems no need to present one here. The Australian species resembles the African insidiosus Smith closely. Elaphrosyron socius new species Holotype. 9 , Boundary Bend, Victoria, Australia, 25 Feb. 1970 (H. E. Evans and R. W. Matthews) [ANIC]. Description of type female. Length 10 mm; fore wing 9.7 mm. Body and appendages entirely black. Fore wings lightly infuscated except very narrowly darker at basal vein and broadly darker in marginal and outer two submarginal cells, also with a still darker band broadly margining the wing beyond the cells, but not quite reaching the extreme margin, which is whitish; hind wing very lightly infuscated, darker apically. Body clothed with conspicuous silvery pubescence over much of head, thorax, and leg-bases (but pubescence dark on much of mesoscutum and scutellum, meta- pleura, and anterior part of propodeum); abdomen with dark pubescence except most of first segment silvery, also posterior margins of sternite 2 and tergites 2-4. Head and thorax covered with pale erect hair which is especially long and dense on the temples, prothorax, mesopleura, and propodeum; coxae with short, 1972 AUSTRALIAN POMPILIDAE 3 pale hairs, femora weakly hairy; first abdominal tergitc with pale hairs, apex of the abdomen with dark setae above and below. Clypeus 2.4 X as wide as high, somewhat convex, the apical margin weakly concave. Front broad, middle interocular distance .65 X head width; inner eye margins subparallel below, strongly convergent above, upper interocular distance .77 X lower inter- ocular distance; ocelli in a broad, flat triangle; postoccllar line: occllo-ocular hne = 5:4; third antennal segment equal to .70 X upper interocular distance. Pronotum short, broadly subangulate behind. Postnotum smooth, produced backward medially as an obtuse angle which is rounded at its apex. Front basitarsus with three long, weakly spatulate pecten spines, the apical basitarsal spine 1.6 X length of second segment; basitarsus also with two slender accessory spines nearly as long as the pecten spines, the accessory spines located on the inner margin and alternating with the pecten spines. Fore wing with the second submarginal cell 1.4 X as wide as the third, measured below, but of approximately the same width when measured on the radial vein; hind wing with the anal vein meeting media slightly basad of the cubital fork. Allotype. S , same data as type [ANIC]. Description of allotype male. Length 7 mm; fore wing 5.8 mm. Coloration as in female; pubescence and erect hairs also as de- scribed for that sex. Fore wings subhyaline, with a brown band beyond the cells which docs not quite reach the apex of wing; hind wings subhyaline, very slightly darkened apically. Clypeus 2.2 X as wide as high, truncate apically. Middle inter- ocular distance .64 X head width, 1.15 X lower interocular distance; upper interocular distance .95 X lower; oceUi rather large, in a flat triangle; postocellar fine: ocello-ocular line = 4:3. First four antennal segments in a ratio of 10:4:8:9, segment three twice as long as thick, segments three and four together equal to .7 X upper interocular distance. Pronotum broadly angulate behind. Postnotum arcuately produced backward medially. Wing venation as described for female except second submarginal cell much larger than third, 1.6 X as wide measured below, 2.5 X as wide measured along the radial vein. Subgenital plate slender, tapering, its midline strongly elevated, surface with numerous strong setae and margin fringed with short, stiff setae. Genitalia as shown in Figure 1. 4 BREVIORA No. 386 Paratypes. 13 9 9,2^5, same data as type; 1 S , Yaapeet, Victoria, 18-22 Feb. 1970 (Evans & Matthews); 1 c5 , 5-15 miles south of Rainbow, Victoria, 21-22 Feb. 1970 (Evans & Matthews) ; I S , Wyperfeld Nat. Park, 25 miles north of Rainbow, Victoria, 18-23 Feb. 1970 (Evans & Matthews); 1 9, Near Adelaide, South Australia, 7 Jan. 1966 (O. W. Richards); 1 9, 1 S , Dedari, 40 miles west of Coolgardie, Western Australia, 11-21 Jan. 1936 (R. E. Turner); 1 9,55c?, Merredin, West- ern Australia, 13 Dec. 1935 (Turner); I 9 ,2 $ S , Perth, West- ern Austraha, Jan., Feb. 1914, 1936 (Turner); 1 9,1 S, Yanchep, Western Australia, Nov., Dec. 1935 (Turner); 1 $, Mundaring Weir, Western Austraha, 19-23 Feb. 1936 (Turner); 1 9, "New Holland" [ANIC, AMS, BMNH, MCZ, WAMP]. Variation. The females vary in length from 7.5 to 10.5 mm. In some specimens the apical margin of the clypeus is suffused with reddish brown, and the mandibles have a variable amount of this color. Otherwise there is httle variation in color except that the female from Yanchep has slightly darker wings than usual and the silvery pubescent bands on the abdomen are reduced. This specimen also has the pecten spines unusually strongly spatulate. The males vary in length from 5 to 9 mm. In several the second submarginal cell is only slightly larger than the third, and there is some variation in the patterning of silvery pubescence, but other- wise there is no noteworthy variation in this series. Remarks. The females differ from those of the African species insidiosiis Smith in the following particulars: clypeus slightly con- cave apically; third antennal segment considerably shorter; two strong accessory spines present on the front basitarsus in addition to the three pecten spines. I have seen no males of insidiosus, but I judge from Arnold's figures that the genitalia bear a close resemblance to those of socius but have somewhat differently shaped volsellae and parapenial lobes. Genus Telostegus Costa Telostegiis Costa, 1887, Prospetto Imenotteri Italiani, II: 88 (type-species: T. major Costa). — Haiipt, 1930, Mitt. Zool. Mus. Berlin, 16: 703-718 (Paiaearctic spp.). — Arnold, 1937, Ann. Transvaal Mus., 19: 35 (African spp.). — Priesner, 1955, Bull. Soc. Ent. Egypt., 39: 168-183 (Egyptian spp.). 1972 AUSTRALIAN POMPILIDAE 5 This genus is very closely related to Elaphrosyron, the major difference being that the second transverse cubital vein is absent. In all other respects, including the male genitalia, the two genera are scarcely separable. Our field notes suggest that the two genera are much alike ethologically. Telostegus is widely distributed in warmer parts of the Eastern Hemisphere, but it has not previously been recorded from Aus- tralia. Although the specimens from that continent display little morphological diversity, I believe that they represent three species (one known only from males). The species may be separated by the following key: Females Vertex weakly arched above the eye tops; antennae elongate, third segment at least .75 X upper interocular distance; known specimens with a wing length of 7-8 mm iiigrocinerascens (Turner) Vertex strongly arched above tops of eyes; antennae shorter, third segment .50 to .58 X upper interocular distance; known specimens with a wing length of 4.5 to 6 mm thoniisivorus new species Males 1. Third antennal segment about 3 X as long as thick, third and fourth together subequal to upper interocular distance; postocellar line barely exceeding ocello-ocular line; vertex little arched above eye tops; genitalia as in Fig. 3 iiigrocinerascois (Turner) Third antennal segment 2.0-2.7 x as long as thick, third and fourth together equal to from .65 to .80 x upper interocular disiance; postocellar line: ocello-ocular line = 14:11; vertex strongly arched above eye tops 2 2. Clypeus 2.5 x as wide as high; tibial spurs nearly as dark as the bgs; volsellae slender, nearly parallel-sided (Fig. 4). .tiirneri new species Clypeus 2.8 X as wide as high; tibial spurs stramineous, much lighter than legs; volsellae much expanded, somewhat truncate apically (Fig. 2) thoniisivorus new species Telostegus nigrocinerascens (Turner) new combination Aporus nigrocinerascens Turner, 1910, Proc. Zool. Soc. London, 1910: 334 (type: $ , MacKay, Queensland, Australia, summer 1899, Turner Coll. [BMNH]). 6 BREVIORA No. 386 Description of type female. Length 8.6 mm; fore wing 7.5 mm. Black; antennae dark brown; legs black to dark brown, tibial spurs dusky testaceous. Fore wings subhyaline, with a brown cloud in the marginal and second submarginal cells that is partially sepa- rated from a brown band along the outer wing margin (not quite reaching the wingtip, which is hyaline); hind wings subhyaline, slightly darker apically. Body clothed with conspicuous silvery pubescence over much of head, thorax, and leg bases (but pubescence dark on much of mesoscutum, scutellum, and base of propodeum); abdomen silvery-pubescent over first segment and second sternite, also in a narrow posterior band on tergite 2, otherwise dark. Head and thorax covered with pale erect hair that is especially long and dense on the temples, prothorax, meso- pleura, and propodeum; abdomen with sparse, dark setae ventrally and on the apical two segments dorsally. Clypeus 2.7 X as wide as high, its apical margin truncate, polished and slightly elevated. Front broad, middle interocular distance .64 X head width; inner eye margins strongly convergent on the upper half, upper interocular distance .8 X lower; vertex passing nearly straight across between eye tops; ocelli in a flat triangle, postocellar line: ocello-ocular line = 7:6; third antennal segment equal to .8 X upper interocular distance. Pronotum short, broadly subangulate behind. Postnotum arcuately produced backward medially. Front basitarsus with three long pecten spines, the apical one 1.2 X the length of segment two; basitarsus also with two accessory spines ventrally, one quite short and the other nearly as long as the pecten spines. Fore wing with the second submarginal cell 2.5 X as wide as high; hind wing with the anal vein meeting media slightly basad of the cubital fork. Plesiallotype. $ , same data as type [BMNH]. Description of plesiallotype male. Length 6 mm; fore wing 5 mm. Black; antennae and legs, including tibial spurs, dark brown. Fore wings subhyaline, with a broad brown band along the outer margin; hind wings subhyaline. Body pubescence as described for the female. Head, thorax, and abdominal venter witli sparse erect hair, including pale hairs on each side of the propodeum. Clypeus 2.4 X as wide as high, its apical margin truncate. Front of moderate breadth, middle interocular distance .66 X head 1972 AUSTRALIAN POMPILIDAE Figures 1-6. Male genitalia of Pompilidae, ventral aspect (drawn to same scale). Fig. 1, Elaphrosyron sociiis n. sp. Fig. 2, Telostegus thomisivorus n. sp. Fig. 3, T. nigrocinerascens (Turner). Fig. 4, T. turneri n. sp. Fig. 5, Pompilus belardoo n. sp. Fig. 6, Fabriogenia Canberra n. sp. 8 BREVIORA No. 386 width; upper interocular distance .95 X lower; vertex weakly arched above eye tops; postocellar line: ocello-ocular line = 11:10. First four antennal segments in a ratio of 15:8:20:19, segment three 3 X as long as thick, segments three and four together equal to the upper interocular distance. Pronotum angulate behind. Postnotum arcuately produced backward medially. Venational features as in female. Subgenital plate elevated along the midline, pointed apically, the margin beset with strong spines. Genitalia as shown in Figure 3. Other specimens examined. 4 9 9,5 S S , same data as type [BMNH]; 1 9, Yeppoon, Queensland, 3-6 Feb. 1970 (H. E. Evans) [ANIC]; 1 9, Ku-ring-gai Chase, 20 miles north of Sydney, New South Wales, 4-8 Jan. 1970 (H. E. Evans) [ANIC]. Variation. The females from Yeppoon and from Ku-ring-gai Chase resemble one another and differ from the MacKay series in the following respects: middle and hind tibial spurs white; abdominal tergites 3-6 with silvery pubescence apically; posterior margin of pronotum rather sharply angular; front basitarsus with both accessory spines rather long. These females resemble those from MacKay in all other particulars, and I feel it probable that all are conspecific. Telostegus thomisivorus nev\^ species Holotype. 9 , Nilemah Station, 50 miles south of Denham, Western Australia, 8-9 October 1969 (R. W. Matthews, note no. AM22) [ANIC]. Description of type female. Length 7 mm; fore wing 6 mm. Black, except anterior margin of clypeus and much of mandibles dull ferruginous, tibial spurs dusky testaceous. Fore wings lightly infuscated, more heavily clouded at the basal vein and in the marginal and second submarginal cells, also with a brown sub- apical band, the extreme outer wing margin hyaline; hind wings subhyaline, shghtly darker apically. Body clothed with silvery pubescence over much of head, thorax, and leg-bases (but pubescence dark on mesoscutum, scutellum, and base of pro- podeum); abdomen with dark pubescence except silvery on first segment, second sternite, and apical margin of second tcrgite. Head and thorax covered with pale hair that is especially dense and long on the temples and propodeum; abdomen with sparse. 1972 AUSTRALIAN POMPILIDAE 9 dark setae ventrally and on the apical two segments dorsally. Clypeus 2.8 X as wide as high, its apical margin truncate, polished and slightly elevated. Front broad, middle interocular distance .68 X head width; upper interocular distance .87 X lower; vertex forming a strong, even arc above tops of eyes; post- occllar line: ocello-ocular line = 7:6; antennae rather short, third segment equal to only .55 X upper interocular distance. Pro- notuni broadly angulate behind. Postnotum arcuately produced backward medially. Front basitarsus with three slender pecten spines, the apical one 1.3 X the length of segment two; basitarsus also with two slender accessory spines, as described for nigro- cinerascens. Fore wing with the second submarginal cell 2.5 X as wide as high; hind wing with the anal and cubital veins inter- stitial. Allotype. $ , same data as type except taken in a Malaise trap (H. E. Evans & R. W. Matthews) [ANIC]. Description of allotype male. Length 5.5 mm; fore wing 5 mm. Color of body and of tibial spurs as in female; pubescence as in that sex except abdominal dorsum with brownish pubescence beyond segment one. Fore wing lightly infuscated, slightly darker in marginal cell and in a subapical band; hind wing subhyaline, slightly darker apically. Sparse, pale hairs present on head, pro- thorax, pleura, and propodeum; abdominal venter with a few dark setae, especially toward the apex. Clypeus 2.8 X as wide as high, its apical margin truncate. Front broad, middle interocular distance .7 X head width; upper interocular distance .9 X lower; vertex forming a strong, even arc above eye tops; postocellar line: ocello-ocular line = 14:11. First four antennal segments in a ratio of 8:3:7:7, segment three about twice as long as thick, segments three and four together equal to .66 X upper interocular distance. Pronotum angulate behind. Postnotum arcuately produced backward medially. Wing venation as in female except anal vein reaching media slightly before the cubital fork. Subgenital plate slender and tapering, its midline strongly elevated, surface with strong setae. GenitaHa differing from those of nigrocinerascens chiefly in having the vohellae more abruptly truncate apically (Fig. 2), Paratypes. 1 9,1 $ , same data as allotype; 1 9 , 6 $ $ , 2 miles west of Coorow, Western Australia, 12 Oct. 1969 (Evans & Matthews); 1 9,1 5 , 13 miles SW of Carnemah, Western 10 BREVIORA No. 386 Australia, 12 Oct. 1969 (Evans & Matthews); 1 9,1 $ ,21 miles north of Northampton, Western Australia, 10 Oct. 1969 (Evans & Matthews); 1 9 , Miaboolya Beach, 9 miles north of Carnarvon, Western Australia, 4 Oct. 1969 (Evans & Matthews) [AMS, ANIC, BMNH, MCZ, WAMP]. Variation. The females vary slightly in size but are consistently smaller than the known females of nigrocinerascens (length 4.5- 7.0 mm). The fore wings vary from hyaline to moderately infus- cated basally, and the abdomen often lacks silvery bands beyond the first segment. The middle interocular distance varies from .67 to .70 X the head width, the third antennal segment from .50 to .58 X the upper interocular distance. The males vary in length from 4 to 6 mm. Some of the smaller specimens have more silvery pubescence on the abdomen than described for the allotype, and in some of these specimens the upper interocular distance exceeds the lower interocular distance slightly. Telostegus turneri new species Holotype. $ , YaUingup, Western Austraha, 1-12 Dec. 1913 (R. E. Turner) [BMNH]. Description of type male. Length 7 mm; fore wing 5.3 mm. Body black; antennae and legs dark brown except tibial spurs stramineous. Wings very lightly infuscated except fore wing with a darker band along the outer margin, hind wing slightly darker apically. Pubescence brownish except conspicuously silvery on front, temples, prothorax, coxae, lower pleura, posterior part of propodeum, first abdominal tergite, and first two stemites. Head, thorax, and abdominal venter sparsely setose, the temples, pro- pleura, and propodeum with an abundance of pale hair. Clypeus 2.5 X as wide as high, truncate apically. Front broad, middle interocular distance .69 X head width; upper interocular distance .95 X lower; vertex forming a strong, even arc above eye tops; postocellar line: ocello-ocular line = 14:11. First four antennal segments in a ratio of 16:7:19:17, segment three 2.7 X as long as thick, three and four together equal to .80 X the upper interocular distance. Pronotum angulate behind. Postnotum subangularly produced backward medially. Fore wing with the second submarginal cell 2.5 X as wide as high; hind wing with 1972 AUSTRALIAN POMPILIDAE 1 1 the anal and cubital veins interstitial. Subgenital plate strongly elevated medially, pointed apically, surface with several strong setae and margin with a row of stout spines. Genitalia very similar to those of nigrocinerascens, but the volsellae more slender and parallel-sided (Fig. 4). Paratypes. 2 $ $ , same data as type except one dated Nov. 1913 [ANIC, BMNH]. Variation. Both paratypes are smaller than the type (fore wing 3.8-4.3 mm). In the smaller specimen, the upper and lower interocular distances are subequal and antennal segments three and four only .65 X the upper interocular distance. In the larger male these measurements approximate those of the type. Genus Pompilus Fabricius Pompilus Fabricius, 1798, Suppl. Ent. Syst., p. 212 (type-species: P. pulcher Fabricius). The precise limits of this large, cosmopolitan genus have never been defined. Many generic and subgeneric names are available, but for the most part these are difficult to apply when the genus is considered from a world point-of-view. Pompilus-Vikc wasps are abundantly represented in Australia, and they are somewhat diverse morphologically, but there seems little hope of fitting them neatly into accepted subgenera or closely related genera at this time. Hence, I shall use the generic name in its broad sense and when discussing species simply attempt to point out the closest known relatives within the genus. Pompilus cinereus (Fabricius) Sphex cinerea Fabricius, 1775, System. Ent., p. 350. Sphex pliimbea Fabricius, 1787, Mant. Insect., I: 278. New synonymy. Pompilus pulcher Fabricius, 1798, Suppl. Ent. Syst., p. 249. New synonymy. This is a very common wasp in eastern Austraha, its typical habitat being sea beaches and sand banks along streams. I have studied the Fabricius specimen in the Banks collection at the British Museum, probably the type, and my interpretation of the species is based on that specimen. Australian specimens compare very favorably with specimens of pliimbeiis and pulcher from India, Africa, and Europe, even to minor details of the male 12 BREVIORA No. 386 genitalia. Evidently this one species ranges in suitable habitats throughout the warmer parts of the eastern hemisphere. Arnold (1937, Ann. Transvaal Mus., 19: 47) recorded it from China, and one assumes that it also occurs in the East Indies. There is an extensive literature on this species, mainly under the name Pompilus plumbeus. Our field observations, so far as they go, agree well with the accounts of various European authors. Arnold (1937) placed three additional species names in synonymy with plumbeus. Pompilus belardoo new species Holotype. 9 , Rottncst Island, Western AustraUa, 21-22 Oct. 1969 (H. E. Evans & R. W. Matthews) [ANIC]. Description of type female. Length 14 mm; fore wing 11 mm. Body and appendages entirely black; wings moderately infuscated, broadly darker along outer margins. Pubescence in large part dark, with a blue-green sheen in certain lights, but conspicuously silvery as follows: much of scape, clypeus, front, and temples; anterior third and posterior margin of pronotum; sides of scutellum and sides and central part of metanotum; much of upper surface of tibiae; narrow posterior margins of abdominal segments one to four, these bands narrowly interrupted mid-dorsally and broadly interrupted ventrally. Body clothed with rather long, dark hairs over much of the head and thorax, including the scape, pro- podcum, coxae, and to some extent the femora; abdomen with dark hairs ventrally and on the apical two tergites. Clypeus 2.3 X as wide as high, truncate apically; malar space, at its minimum, about half the width of the anterior ocellus. Front rather narrow, middle interocular distance .56 X head width; upper interocular distance .82 X lower; vertex passing nearly straight across between eye tops; postocellar line: ocello- ocular hne = 6:5; third antennal segment subequal in length to upper interocular distance. Pronotum broadly subangulate be- hind. Propodeum with a median sulcus anteriorly, posterior third with a flat declivity. Front basitarsus with a strong pecten con- sisting of five slender spines, the apical one 1.7 X as long as the second segment, the basitarsus also with two accessory spines nearly as long as the pecten spines. Fore wing with the marginal cell removed from the wing tip by approximately its own length, 1972 AUSTRALIAN POMPILIDAE 13 the radial vein somewhat angled at the third transverse cubital vein; third submarginal cell about as wide below as the second, but more strongly nai rowed above. Allotype. S, same data as type |ANIC]. Description of male allotype. Length 1 1 mm; fore wing 9.7 mm. Color of body and wings as in female, but silvery pubescence less extensive than in that sex, restricted to the base of the mandibles, sides of the clypeus and front, temples, and interrupted apical bands on abdominal tergites 1 and 2 and sternite 2. Head and thorax with an abundance of long, dark hairs, including some on the scape and on the coxae and to some extent the femora; abdomen sparsely setose ventrally. Clypeus convex, twice as wide as high, its apical margin trun- cate; malar space rather long, nearly equal to width of anterior ocellus. Middle interocular distance .58 X head width; upper interocular distance .97 X lower; vertex passing nearly straight across between eye tops; postocellar line very slightly exceeding ocello-ocular line. Third antennal segment 2.4 X as long as thick, third and fourth together subequal to upper interocular distance. Pronotum subangulate behind. Propodeum rounded, abruptly dechvous on posterior fourth. Apical segment of front tarsus symmetrical, unmodified, but the inner claw strongly curved, bifid. Venation as in female. Subgenital plate elevated along the mid- line, broadly rounded apically. GenitaHa with the basal hooklets absent, the aedeagus with an abrupt apical expansion that is turned sharply downward (Fig. 5). Paratypes. 6 9 9 , 5 <5 $ , same data as type [AMC, ANIC, BMNH, MCZ, WAMP]. Variation. The females vary in length from 11 to 14 mm. In the majority the third antennal segment is very slightly shorter than the upper interocular distance (.93-1.00 X this distance) and in three the silvery pubescence on the posterior margin of the pronotum and on abdominal tergite 4 is weakly developed. The males vary in length from 8.5 to 11 mm. Three of them have slightly more silvery pubescence than the allotype, including some on the pronotum and sides of the scutellum; however, the banding of the abdomen is relatively constant. Remarks. This species is closely related to several other Australian species, such as semiluctuosus Smith, but the patterning 14 BREVIORA No. 386 of silvery pubescence, the male genitalia, and the nature of the pecten spines of the female are distinctive. Belardoo is an abo- riginal word from Western Australia referring to coastal sand dunes. Genus Ctenostegus Haupt Ctenostegus Haupt, 1930, Mitt. Zool. Mus. Berlin, 16: 685 (type-species: Sphex cingiilata Fabricius). This is a dominant genus of Pompilidae in Australia and on adjacent islands; in fact it may well be the largest genus of the family in that zoogeographic region. It is closely related to Pompiliis and probably a derivative of that genus. Only a few of the species have been described. One of the commoner species, for which we have nesting data, is described below and compared with the type species. Ctenostegus warragai new species Holotype. 9 , 3 miles west of Wentworth, New South Wales, 27 Nov. 1969 (R. W. Matthews, note no. AM92) | ANIC]. Description of type female. Length 12 mm; fore wing 9.5 mm. Body and appendages entirely black. Wings rather heavily infus- cated, broadly darker along outer margin. Pubescence wholly dark, on the abdomen with dark blue-green reflections in certain lights. Body with short, sparse, rather dark hairs as follows: front, vertex, temples, propleura, front coxae, propodeum, and tip of abdomen. Clypeus 2.4 X as wide as high, apical margin narrowly polished, weakly concave; malar space well developed, at its minimum nearly as long as width of anterior ocellus. Front narrow, middle interocular distance .55 X head width; upper interocular distance .92 X lower; postocellar line: occllo-ocular hne = 6:5; third antennal segment equal to .77 X upper interocular distance. Posterior margin of pronotum angulate. Propodeum rounded, with a flat, oblique declivity on the posterior third. Front basi- tarsus with four rather broad pecten spines, the apical one 1.7 X as long as the second segment. Second submarginal cell of ap- proximately the same width as the marginal cell, measuring 2.2 X as wide as high, narrowed by two-thirds above. 1972 AUSTRALIAN POMPILIDAE 15 Paratypes. 10 9 9, same data as type except collected by H. E. Evans & R. W. Matthews; 6 9 9,4 miles east of Wilcannia, N.S.W., 1-2, 20-21 Nov. 1969 (Evans & Matthews); 1 9, 5 miles west of Wilcannia, N.S.W., 1 Nov. 1969 (Evans & Matthews); 4 9 9, Packsaddle, 111 miles north of Broken Hill, N.S.W., 31 Oct., 21-26 Nov. 1969 (Evans & Matthews); 1 9, Port Germein, South Australia, 28 Oct. 1969 (Evans & Matthews) [AMS, ANIC, BMNH, MCZ]. Variation. The paratypes vary in length from 8 to 14 mm; the middle interocular distance varies from .52 to .57 X the head width, the third antennal segment from .75 to .85 X the upper interocular distance. Several of the specimens have a small amount of silvery pubescence on the sides of the clypeus and/or lower front. Remarks. This species is closely related to the type species, cingulatus Fabricius, but the pubescence is wholly or almost wholly dark, the propodeum more hairy, and the pecten spines longer. I have not been able to associate any males with these females with any certainty. All specimens were collected in areas of extensive sand dunes or ridges. The species name warragai is an aboriginal word from New South Wales meaning "plenty of sand." Genus Fabriogenia Banks Fabriogenia Banks, 1941, Occ. Papers B.P. Bishop Mus., Honolulu, 16: 240 (type-species: F. incompta Banks). This is the dominant genus of the tribe Auplopodini in Aus- tralia. Turner (1910, Proc. Zool. Soc. London, 1910: 310) presented a key to several of the species (under the name Pseiida- genia), but there are many additional species. Both Banks (1941) and Townes (1957, Bull. U.S. Nat. Mus., 209: 141) pointed out the similarity of Fabriogenia to Phanagenia, but I regard them as generically distinct. Not all of the species have a well-developed malar space and lateral spines beneath the apical tarsal segments, as described by Banks for the type species, but they agree in having a central polished area on the apical tergite of the female (approaching the condition in Auplopiis) , as well as a group of stout bristles on the mentum. The generic classifica- tion of the Auplopodini i:, in a deplorable state, but for the present 16 BREVIORA No. 386 it seems sufficient to assign the majority of the Australian species to Fabriogenia. Fabriogenia Canberra new species Holotype. 9 , Canberra, A.C.T., Feb. 1970 (R. W. Matthews, note no. AM191) [ANIC]. Description of type female. Length 9 mm; fore wing 8 mm. Body and legs black; antennae bright orange except scape and apical segment very weakly infuscated. Wings clear hyaline except fore wings with a narrow brown band at the basal and transverse median veins and a somewhat wider band through the base of the marginal cell, across the second submarginal and base of the third submarginal cell, and into the outer discoidal cell; tip of wing also darkened. Body covered with fine, silvery pubescence which is darker on the vertex and mesoscutum. Pale, erect hairs are present over much of the head, prothorax, front coxae, mesopleura, and propodeum; abdomen sparsely setose ventrally and densely so on the apical tergite. Clypeus convex, weakly produced medioapically, measuring 1 .9 X as wide as high. Middle interocular distance .64 X head width; upper interocular distance .95 X lower; postocellar line: ocello-ocular line = 7:9; vertex forming a low, even arc above eye tops. Antennae moderately thick, third segment 3.5 X as long as thick, exceeding the fourth segment as 10:9, equal to .70 X the upper interocular distance. Pronotum broadly subangulate behind. Postnotum, along the midline, about two-thirds the length of the metanotum. Propodeum evenly rounded in lateral view. Hind tibiae with numerous short spines laterally, smooth above. Second submarginal cell twice as wide as high, receiving the first recurrent vein slightly before the middle; third submarginal cell 1 .7 X as wide as second, receiving the second recurrent vein .3 X the distance from the base. Allotype. S , same data as type [ANIC]. Description of allotype male. Length 7.5 mm; fore wing 6.8 mm. Black, except as follows: mandibles with a cream band about halfway from the base, the tips rufous; clypeus cream laterally and apically; sides of lower front narrowly cream; front legs suffused with light brown on their inner sides; hind tibial spurs white, other spurs more or less infuscated; antennae orange except 1972 AUSTRALIAN POMPILIDAE 17 basal two segments weakly infuscated, apical four segments strongly infuscated. Wings clear hyaline, unhanded, fore wings slightly darkened at apex. Body clothed with fine, silvery pubescence. Head and thorax extensively covered with pale, erect hairs, including the propodeum and front coxae. Clypeus 2.2 X as wide as high, truncate apically. Middle interocular distance .64 X head width; upper intcrocular distance 1.1 X lower; postocellar line: ocello-ocular line = 3:4; vertex forming a smooth arc above tops of eyes. Antennae elongate, third segment subequal to fourth, 3.4 X as long as thick, .67 X upper interocular distance. Pronotum broadly subangulate behind. Postnotum medially only slightly shorter than metanotum. Vena- tion as in female. Abdominal sternite 6 with a pair of toothlike elevations bordering a median, flat area. Subgenital plate broad, outer apical margins rounded, extreme apex truncate; surface somewhat concave, margins fringed with setae. Genitalia as shown in Figure 6. Paratypes. 4 9 $ , same data as type [ANIC, BMNH, MCZ]. Variation. The four paratypes range in size from 7 to 9.5 mm and resemble the type closely in all details; the upper interocular distance varies from .93 to 1.0 X the lower, antennal segment three from .65 to .72 X the upper interocular distance. Remarks. This species runs to the final couplet in the key to Pseudagenia provided by Turner (1910, Proc. Zool. Soc. London, 1910: 310). It is a smaller species than australis Cameron, and the male lacks banding on the fore wings and has the sixth sternite somewhat differently modified. The size is comparable to that of jusijormis Saussure, but the latter species is much less hairy and has a longer pronotum. Fabriogenia dilga new species Holotype. 9, Canberra, A.C.T., 31 January 1970 (H. E. Evans, note no. A118) [ANIC]. Description of type female. Length 9.5 mm; fore wing 8.8 mm. Coloration of body appendages, wings, and pubescence exactly as described for Canberra. Head and pronotum, including the front coxae, with numerous pale, erect hairs; propodeum with a few pale hairs on each side, but the pleura weakly hairy; abdo- men sparsely setose ventrally and densely so on the apical tergite. 18 BREVIORA No. 386 Clypeus convex, strongly and subangularly produced apically, measuring 1.6 X as wide as high. Middle interocular distance .61 X head width; upper and lower interocular distances subequal; postocellar line: ocello-ocular line = 2:3; vertex forming a low, even arc above eye tops. Antennae elongate, third segment 4 X as long as thick, exceeding the fourth segment as 10:8, equal to .82 X the upper interocular distance. Pronotum very broadly angulate behind. Postnotum along the midline about half the length of the metanotum. Propodeum evenly rounded, with a weak median sulcus. Hind tibiae weakly spinose laterally, almost without spines above. Second submarginal cell 1.7 X as wide as high, receiving the first recurrent vein slightly beyond the middle; third submarginal cell nearly twice as wide as the second, receiving the second recurrent vein .38 X the distance from the base. Paratype. 9, same data as type [ANIC]. Variation. The paratype is smaller (7.5 mm, fore wing 7 mm) and the third antennal segment equal to only .73 X the upper interocular distance. Otherwise it is very similar to the type. Remarks. This species is exceedingly similar to the preceding, but differs in the shape of the clypeus, the more elongate antennae, the shorter postnotum, and minor details of wing venation. Only the female is known. Dilga is an aboriginal word from New South Wales, meaning "a stick of wood." Two females were reared from a trap-nest, so I assume the species typically nests in hollow twigs or other cavities in wood. ML'S. CO MP. ZOOL LIHRARY •1 <^ ,._ B R E V I O Miaseitaini of Comparative Zoology ' HARVARD Cambridge. Mass. 15 March. 1972 Number 387 GROWTH CHANGES IN AMPHILIMNA OLIVACEA (LYMAN) AND THE SYSTEMATIC STATUS OF AMPHITARSUS SPINIFER SCHOENERi Lowell P. Thomas- and Amy Schoener^ Abstract. Analysis of the developmental sequence in the brittlestar Aiuphiliniiui olivacea (Lyman) has revealed the similarity between its young stages and those of adult Amphitarsiis spiiiifer Schoener. Evidence is provided to justify synon\mizing the latter species with the former. INTRODUCTION Two recent papers published almost simultaneously have dealt with the enigmatic brittlestar genus Ainphitarsus. Schoener (1967a) described two new species of this previously monotypic genus and discussed its possible family affinities, while Thomas (1967: 126) pointed out the similarities between Ainphitarsus mirahilis H. L. Clark 1941 and Amphilimua olivacea (Lyman, 1869). The information contained in these two papers has prompted a re-examination of Ainphitarsus spinijer and a study of the growth stages of Amphilimua olivacea. As a result of our studies we synonymize A. spinijer as a junior synonym of Amphilimua olivacea and discuss changes that take place during the growth of this species. ' Contribution No. 1454 from the University of Miami. Rosenstie! School of Marine and Atmospheric Science - Rosenstiel School of Marine and Atmospheric Science, University of Miami. Id Rickenbacker Causeway, Miami, Fla. 33149 •^ Museum of Comparative Zoology, Harvard University, Cambridge, Mass. 02138 2 BREVIORA No. 387 Amphilimna olivacea (Lyman, 1869) Ophiocnida olivacea Lyman. 1869, BulL Mus. Comp. ZooL, 1(10): 340 (off Alligator Reef, Caribbean, 79 fms. ). Amphitarsus spinifer Schoener, 1967a, Breviora No. 267: 6, fig. 2 (NW Atlantic, 200 m). See Thomas (1967) for complete synonymy of A. olivacea. Diagnosis. Adult specimens (see below for discussion of young animals). Jaws with two, occasionally three, infradental papillae at apex, two or three slender oral papillae bordering each adoral plate; two oral tentacle scales, resembling papillae, in each oral slit. Arms slender, six times disc diameter; tentacle pores large; eight to ten proximal ventral arm plates with two tentacle scales on each side, innermost attenuated; beyond tenth ventral arm plate a single tentacle scale on each side; seven to nine slender arm spines, ventralmost largest; arm spines of arm segments under disc greatly flattened, fused together to form peculiar flanges occupying genital slits; dorsal arm plates slightly wider than long. Disc scales studded with slender spines; primary plates present, often in- conspicuous. Radial shields narrow, joined proximally, slightly separated distally; disc deeply notched at each pair of radial shields. Color variable, disc gray, tan, or brown, arms pink or orange. SYSTEMATIC DISCUSSION Although this species has been known for over a century, it has been only in the last nine years that the peculiar fused arm spines under the disc have been described. Early references to A. olivacea, in addition to omitting mention of this important char- acter, are generally brief and often unillustrated. Lyman's only figures (1871, pi. 1, figs. 7, 8) show primitive plates, although only indistinctly, and a view of the ventral arm and disc surfaces, omitting the fused arm spines. The only other illustrations prior to 1962 are VerrilFs (1899; pi. 42, figs. 1, la) stylized figures of the ventral disc surface and one row of arm spines. Again, the fused arm spines are omitted. Finally, Cherbonnier (1962) described and illustrated the fused arm spines, erroneously referring to them as "ecailles genitales." They were described and properly identified by Thomas (1967), who also considered the similar "winglike flanges" of Amphitarsus niirabilis to be fused arm spines. If the latter observation is correct, it is almost certain, in view of 1972 AMPHILIMNA OLIVACEA AND AMPHITARSUS SPINIFER 3 the other similarities discussed by Schocner (1967a) and Thomas (ibid.), that Amphitarsus mirabilis and Amphilimna olivacea are congeneric. External features and dissection of oral and dental plates indicate that Amp/iilimna olivacea belongs in the family Ophiacanthidae, but material of Amphitarsus mirabilis is not available for dissection. GROWTH CHANGES In the following section small (presumably young) and large (presumably adult) specimens of Amphilimna olivacea are figured in dorsal and ventral aspect (Figs. lA, B & 2 A, B). Synopses of the growth changes are given, purposely written so as to trace the growth sequence in reverse, going from the larger to the smaller individual. This was done so that specimens of the even smaller species, Amphitarsus spinijer (Figs. IC, D & 2C, D), could be viewed as initial stages of a growth series terminating with large specimens of A. olivacea. Growth changes on the dorsal surface Amphilimna olivacea. In the larger specimen (9.7 mm disc diameter) important systematic characters of the dorsal surface (Figure lA) appear as follows: (1) Only the central plate of the six primary plates is conspicuous. (2) The radial shields are contiguous for much of their length and are greatly attenuated. (3) Most of the disc is covered by fine overlapping scales, each of which frequently bears a spine. (4) The dorsal arm plates are rectangular or with a convex outer edge. In a smaller specimen (4.5 mm disc diameter): (1) All six of the primary plates are conspicuous and occupy a greater proportion of the disk than in the larger specimen. (2) The radial shields, which are less attenuated, are not always contiguous. (3) The fine scale covering of the disc is comprised of fewer scales than in the larger specimen, and the scales less frequently bear spines. (4) The dorsal arm plates have convex distal edges and are con- cave laterally. Amphitarsus spinijer. The holotype of this species (Fig. IC) has a disc 3.8 mm in diameter, and is therefore comparable to the specimen of A. olivacea illustrated in Figure IB. Here: (1) The six primary plates are conspicuous, although in this case they are BREVIORA No. 387 Figure 1. Dorsal views of specimens arranged in order of decreasing size. A. Amphilimna olivacea (9.7 mm); B. A. olivacea (4,5 mm); C Amphitarsus spinifer (3.8 mm); D. A. spinifer (ca. 3 mm). 1972 AMPHILIMNA OLIVACEA AND AMPHITARSUS SPINIFER 5 less perfectly arranged and other large irregular scales are present on the disc. (2) The radial shields are separated for their entire length and, although longer than wide, do not appear greatly attenuated. (3) The fine disc scales bear spines only occasionally. (4) The dorsal arm plates are nearly identical to those figured for A. olivacea of this size (Fig. IB). Figure ID illustrates a paratypc of A. spinifer whose disc diameter is smaller (ca. 3 mm) than that of the holotype. Here: ( 1 ) The primary plates, of which one central and three radial are figured, are large relative to the size of the disc, occupying a sizable fraction of the dorsal surface. (2) The radial shields are much shorter, relative to their width, and are not contiguous. (3) The scalation of the disc consists of even fewer fine overlapping scales, which bear scale spines only occasionally. Here again, several larger scales, particularly one in each interradial area, are quite noticeable, although other large scales are present around the central disc area. (4) The dorsal arm plates, which are basically rectangular, show some convexity at the distal edge and slight indentations laterally. Growth changes on the ventral surface Amphilimna olivacea. In the specimen with a disc 9.7 mm in diameter: (1) Seven sets of flanges of fused arm spines are borne by those arm segments overlain by the disc. (2) The oral shield is basically triangular with a distal edge that forms a slight outward bulge toward the middle. (3) On the arm plates overlain by the disk two tentacle scales are usually present. (4) Eight or nine arm spines are present on each side arm plate at the point where the arm becomes free of the disc. (5) The ventral surface of the disc is covered by many small overlapping scales bearing spines. Figure 2B shows the smaller specimen of this species: ( 1 ) There are only three sets of flanges per arm. (2) The oral shield, with an even more convex distal side, almost forms a rhombus. (3) Two tentacle scales are often present on the arm plates overlain by the disc. (4) The arm bases have fewer arm spines (five or six) than in the above specimen. (5) The ventral surface of the disc is covered by fewer overlapping scales than in the above specimen. Some of these bear single scale spines. Am phi tars us spinifer. The specimen illustrated in Figure 2C (the holotype) is almost the same size as the A. olivacea shown BREVIORA No. 387 Figure 2. Ventral views of specimens arranged in order of decreasing size. A. Amphilimna olivacca (9.7 mm); B. A. olivacea (4.5 mm); C. Amphitarsiis spinijer (3.8 mm); D. A. spinifcr (ca. 3 mm). 1972 AMPHILIMNA OLIVACEA AND AMPHITARSUS SPINIFER 7 in Figure 2B. In the characters considered, except perhaps for the length of the disc scale spines, it scarcely differs from A. oUvacea. Figure 2D shows the ventral aspect of the smaller specimen of A . spinijer: ( 1 ) The number of flanges in the genital area is reduced still further to two sets per arm. (2) The distal side of the oral shield is more rounded. (3) Two tentacle scales usually are present on the arm plates overlain by the disc. (4) The number of arm spines near the arm base (four or five) is less than that in the larger A. spinijer (above). (5) The ventral surface of the disc is covered by still fewer scales, which overlap less, and only a few of which bear spines. CONCLUSION In order to conclude on the basis of evidence from growth sequences that two supposed species, the larger Amphilimna olivacca and the smaller Amphitarsiis spinijer, are in reality only one species, two criteria must be satisfied. First, for specimens of each species that overlap in size, one must show that the variation in important characteristics is negligible. Second, one must be able to offer a logical progression of growth stages from one to the other. The first point is readily satisfied upon examination of specimens of the same size range. This has been done (see Figs. IB, C & 2B, C) and the specimens are found to be very similar. The second point, that growth sequences should generally agree within reasonable bounds with those of other species investigated, is also satisfied. Superficially the following sequences are con- sistent with knowledge of developmental series for other ophiuroid species (Schoener, 1967b, 1969). These points include the following: (1) The six primary plates of the dorsal disc surface may become less conspicuous as the specimens in a series increase in size. (2) The radial shields are initially small and may elongate with an increase in the size of the specimen. (3) More arm spines are added as the adult condition is approached. (4) There may be an increase in the number of specialized elements (e.g., flanges in the genital area) as the adult condition is reached. (5) There will be an increase in the number of scales covering the disc if 8 BREVIORA No. 387 their absolute size remains constant while the size of the specimen increases. Several other characters remained fairly constant in this series. These were the deep notching of the disc at the distal ends of the radial shields, the shape of the dorsal and ventral arm plates, and the number of infradental papillae at the jaw apex. Points on which no judgment is presently made concern the variation in the shape of the oral shield, which in any case seems slight, and the fact that there is no documented sequence of development in which radial shields in the smallest specimens are initially separated but later become contiguous. However, since our knowledge in this area is just being expanded, this may indeed occur in other species. Based on the above evidence, it is concluded that the smaller species A mphitarsus spinifer Schoener is part of the developmental series of Amphilimna olivacea (Lyman), the latter name having priority. This research has been supported by National Science Founda- tion Grant GB- 16556 and a grant from the Radcliffe Institute, for which grateful acknowledgment is made. We also thank Claire Ulanoff, who illustrated our specimens. LITERATURE CITED Cherbonnier, G. 1962. Ophiurides. Exped. Oceanogr. Beige Eaux Cot. Afr. Atlant. Sud., 3(8): 1-24, pis. 1-7. Clark, H. L. 1941. Reports of the scientific results of the Atlantic expeditions to the West Indies, under the joint auspices of the University of Havana and Harvard University. The echinoderms (other than holothurians). Mem. Soc. Cubana Hist. Nat., 15(1): 1-54, pis 1-10. Lyman, T. 1869. //; Pourtales, L. F. de. Preliminary report on the Ophiuridae and Astrophytidae dredged in deep water between Cuba and the Florida reef . . . Bull. Mus. Comp. Zool., 1(10): 309-354. . 1871. Illustrated catalogue of the Museum of Comparative Zoology at Harvard College. No. 6. Supplement to the Ophiuridae and Astrophytidae. Mem. Mus. Comp. Zool., 2(6): 1-18, pis. 1-2. Schoener, A. 1967a. Two new species of Ainphitarsiis (Ophiuroidea) from the western North Atlantic. Breviora, No. 269: 1-9, figs. 1-3. 1967b. Post-larval development of five deep-sea ophiuroids. Deep-Sea Res., 14: 645-660, figs. 1-9. 1972 AMPHILIMNA OLIVACEA AND AMPHITARSUS SPINIFER 1969. Atlantic ophiuroids: some post-larval forms. Deep- Sea Res., 16: 127-140. Thomas, L. 1967. The systematic position of Amphilimna (Echinoder- mata; Ophiuroidea). Proc. Biol. Soc. Washington, 80: 123-130. figs. 1-9. Verrill, a. E. 1899. North American Ophiuroidea. I. Revision of certain families and genera of the West Indian ophiurans. II. A faunal catalogue of the known species of West Indian ophiurans. Trans. Connecticut Acad. Arts. Sci., 10(7): 301-385, pis. 42-43. * f^L'S. CrOMP. 200U' B R E V I O'WM U.\JJVGRSiTY Rluseuni of Comparataye Zoology Cambridge, Mass. x^pril, 1972 Number 388 INTENSE LOW-FREQUENCY SOUNDS FROM AN ANTARCTIC MINKE WHALE, BALAENOPTERA ACUTOROSTRATA William E. Schevill and William A. Watkins Abstract. Intense low-freqnencv underwater sounds, somewhat similar to those heard from other species of Btihwitoptoa, have been recorded from ininke whales, Balaenoptera acuturostrata Lacepede 1804, in tlie Ross Sea, Antarctica. The small Balaenoptera f ininke whales) of the Antarctic had for many vears been identified as B. acutorostrata Lacepede I8U4, until \Villiamson 1959 and 1961 ) indicated that some of them might be referable to B. bonaerensis Burmeister 1867, which van Utrecht and \an dcr Spoel 1962) considered no more than a \ariety. Since our Ross Sea whales showed us only the top of the back and the part of the head from the blowholes forward . Fig. 1 ) , we could not judge whether they were this form or the t\pical acutorostrata, so we refrained from reporting the sounds till the relationship of these Antarctic minke whales to those in other parts of the world was more clearly defined. Ohsunii, Masaki, and Kawamura (1970) have now compared the southern and northern forms and concluded p. 116) that an\ differences were minor and that the Antarctic minke whale was not taxonomically separable from the typical northern Balaenoptera acutorostrata, and this conclusion we happily accept. Our recordings were made from the edge of the Ross ice shelf 2 km east of Cape Crozier, Ross Island, on 22 Noxember 1964. A whale had been sighted earlier in the 4-km stretch of open water between the ice shelf and the loo.se pack-ice further out, but it was too far away for identification. A light northerly wind e\entually clo.sed this open water and dro\e the pack-ice against the ice shelf. Large (hunks of ice were forced on edge, 2 BREVIORA No. 388 and the loose ice was pushed together against the shelf, foiining a solid co\er as far as the eye could see. ^\'e had been listening underwater for ice sounds as the pack came in and had forgotten about the whale sighting, when we were startled by the characteristic sound of a whale blow in air. A minke whale had thrust its head out through a hole in the ice far enough to breathe (Fig. 1 ). In an 8-m whale, this means nearly 1.5 m. There were three holes nearby, apparently kept open by emperor penguins, Aptenodytes forsteri G. R. Gray 1844, in relatively thin ice that had formed behind a projecting tongue of the ice shelf. We were using the nearest for our hy- drophone, and the whales appeared in one about 5 m further away. The whale sounds had been noted, but not identified, in the underwater ambient before the whales' appearance. There were two minke whales, though at first we saw only one at a time. Later both were \isible at once as they blew in adjacent holes. The whales came from the direction of the ice pack and would return again in that direction after a series of blows. Five to ten minutes elapsed between series of four to seven blows. Examination of the recordinos reveals that initially there were two whales producing sounds, one relatively close by and a sec- ond at a distance. Sounds produced when the whale was close by often took us by surprise and massi\ely overloaded the sound equipment. The loud sounds were heard when a whale was near the breathing holes, and therefore near the h)drophone, either just before or just after a series of blows. The blows were barely audible underwater. The l)ackground ambient on the day of these recordings was filled with a wide variety and range of sounds. Though most of these remain imidentified, occasionally some could be matched to ice movement and also to a single \isit from a leopard seal, Hydruroa leplonyx (Blain\'ille 1820). Others were recognized as sounds from Weddell seals, Leptonychotes weddelli (Lesson 1826). We had been working for some weeks in McMurdo Sound in an acoustically pure culture of Leptonychotes and we were confident that we could recognize most of their vocaliza- tions. Throughout this entire day's listening, these seals were heard quite often as they moxed from crack to crack under the more solid ice co\tr. They were audible except during the period of the approach of the minke whales. During this time, a period of about 30 minutes, the Weddell seals were silent. 1972 SOUXI1S 1 Rrnr ax antarctic minkk whale 3 EQUIPMENT AND METHODS The sounds were recorclcd with an LC-34 (Athmtic Research) hydrophone ancl a ^\'H()I-huilt ampHfier and sprin^"-dri\en re- corder i^\\'aikins, 19()jj. AnaKsis pla\ljack was on Clrown 800 series) recorders. The combined frequency response was flat wiihiii Vl> dR from M) to 30,000 Hz. Spectograms were made on a Kay l^lectric 7029A Sound Spectograph. The h\ch'ophone was thro\vn fiom the ice shelf across thin ice through holes used by emperor penguins, Aptcnodytes jor- sleri. The hydrophone was suspended 3 meters or more in the water from the ice edge, but since the cable was operating at low impedance, no noise was generated by motion of the cable arainst the ice. \->" CHARACTERISTICS OF THE WHALE SOUNDS The minke whale '>;ounds were intense. The loudest of these in undistorted recording are 60 to 65 dB above the local rela- ti\ely high background aml:)ient, \vhich a\-eraged about 0 dB re 1 d\ ne cm". Of course, we do not know how near the whales were, I^ut we assume they were quite near, since the loudest sounds occurred either a few seconds before or after a whale was seen in a breathing hole. Thus, 65 dB re 1 dyne/cm- ma\ be nearly the maximum 1 m) signal strength of these whale sounds. The sounds were composed of a single downward sweep in frequency, starting at 130 to 115 Hz and sw^eeping to about 60 Hz see Fig. 2). Since the frequency sweep continued through- out the sound, the rate of drop in frequency varied with both the span of frequencies and the duration of the call. The sweep rate was fairly regular throughout each sound. Sounds that were recorded at a low enough le\-el to be free from distortion had no harmonic structure; they appear to ha\-e l^een composed of nearly sinusoidal wa\'es. The second and third traces \isible with each sound in Figure 2 are from reflections off ice walls or the bottom. The minke whale sounds began with gradualK increasing in- tensitv for the first few cycles and ended with a gradual lethic- tion in intensity for the last few cycles. Thus the sounds appeared to rise out of background and disappear back into it. The endings of the sounds also were further obscured b)' reflections and re\er- berations. Measurement of the duration of a sound depended 4 BREVIORA No. 388 on its relati\'e intensity oxer ambient. Indi\idual minke whale sounds lasted from 0.2 to 0.3 seconds. No repetitive pattern was evident in sequential sounds either from one individual or from both. Of nine sounds presumed to be from one whale and recorded on one continuous tape, the intervals ]:)etween sounds (in seconds) were approximately 19, 8, 13, 97, 35, 10, 89, and 12. The sounds were not all identical, but had a basic similarity in duration, frequency sweep, and intensity. Differences could not be attributed easily to characteristics of indi\idual whales, though certainly that possibility exists. DISCUSSION Intense low-frequency sounds have been recorded in the pres- ence of other species of Balaenoptera. Schevill and Watkins (1962) reported a 75- to 40-Hz sound from B. physalus (Lin- naeus 1758), and Schevill, Watkins, and Backus (1964) iden- tified a 20-Hz (23-18 Hz) sound also with B. physalus. More recently, other low-frequency sounds have been reported from B. physalus, 20 Hz to 100 Hz, and B. edeni Anderson 1879, with average frequency of 124 Hz, by Thompson and Cummings (1969), and from B. musculus (Linnaeus 1758), with most energy below 50 Hz, by Cummings and Thompson ( 1971 ) . The intensity of most of these low frequencv whale sounds has been estimated to be 60 dB or more re 1 dyne/cm" at 1 m. Other sorts of sounds also have been reported from Balaeno- ptera. These include pulses at about 25 kHz from B. musculus by Beamish and Mitchell (1971), and chirps and whistles at 1500 to 2000 Hz from B. physalus by Perkins ( 1966) . Our gear is capable of receiving such sounds, and with it we have listened to a few hundred P>alaenoptera over many years, and yet we have recorded only lower-frequency sounds from them. The minke whale sound is similar in most respects to both the 75- to 40-Hz and the 23- to 18-Hz sounds in our recordings of the larger fin whale, B. physalus. The one from B. acutorostrata and these two from B. physalus (1 ) are relatively intense, (2) are composed of low frequencies, (3^ have a downward sweep in frequency, (4) are nonharmonic. nearly sine-wave, (5) are made UD of about the same number of cycles duration at the same relative intensity ( about 20 cycles at 40 dB above background ) , and (6) begin with gradually increasing intensity and end with 1972 SOUNDS FROM AN AXTARCTU: MINKK WHALE 5 dropping intcnsit\ . The differences in these sounds are mainly those related to freqiienc\-. This \\ould seem to indicate a com- mon method of sound production and similar acoustic structures. Neither the 75- to 40-H7 finback whale sound nor the 120- to 60-Hz sound of the minl!mim(mi^iasm^Hntitnmfimmniimm)iesti&^ Seconds T* 6 Figure 2. Fom miiikr wluilc sounds are analyzed witliout intervening inter\als — up to iwo niihules elapsed between sounds. J he high level ot the sounds relali\e lo background peiniiis .uiaKsis withniu oiivions iiuer- fcrence from anihitnt sound. Ilie seiond and iliird iiaces ac(oinpan\ nig Ihe niinke whale .sounds aie from lellections oil the bottom or otf ice walls. The slightly beaded ajjpearanee of liie sound traces probably is a result of multiple-path sound tiansnusMon with constructive and destructive rein- forcement of the sound as the wa\e length (from about 11.5 to 25 m) varies with the sweep in frequency. The elfective analyzing idter bandwidth was II Hz. 8 BREVIORA No. 388 LITERATURE CITED BE.\^rISH, P.. AND E. Mitchell. 1971. lUlrasonic sounds recorded in the presence of a blue whale Balaenoptera inusculus. Deep-Sea Research, 18: 803-809. CuMMiNGS, VV. C, AND P. O. THOMPSON. 1971. Underwater sounds from the blue whale. Jour. Acoust. Soc. Anier.. 50, 4: 1193-1198. Ohslimi, S.. V. Masaki. and A. Kawamura. 1970. Stock of the Antarctic minke whale. Sci. Rep. Whales Res. Inst., 22: 75-125, Appendix 1, 3 pis. Perkins, P. J. 1966. Communication sounds of linback whales. Norsk Hvalfangst-Tidende, 55: 199-200. Schevill. W. E., and W. A. Watkins. 1962. Whale and porpoise voices, a phonograph record. Woods Hole, ^Voods Hole Oceanographic Institu- tion. 24 pp. anil phonograph disc. , , and R. H. Backis. 1964. The 20-cycle signals and Balatnoptera (fin wiiales) . Pp. 147-152, in Marine Bio- acoustics, W. N. Tavolga (ed.) , Oxford, Pergamon Press. 413 pp. Thompson, P. O., and W. C. C(iMMiNf;s. 1969. Sound production of the finback whale, Balaenoptera pliysalus, and Eden's whale, B. edeni, m the Gulf of California (.\bstract) . P. 109. Proceedings of the 6th an- nual conference on biological sonar and di\ing mannnals. Stanford, Calif., Stanford Research Institute. Utrechp, W. L. a AN. AND S. VAN DER Spoel. 1962. Observations on a minke whale (Mammalia, Cetacea) from the Antarctic. Ztschr. fiir Saugctier- kundc, 27: 217-221. AVatkins, W. A. 1963. Portable underwater recording system. Undersea Technology, 4 (9) : 23-24. ■\V^iLLiAMsoN, G. R. 1959. Three unusual rorqual whales from the .Antarctic. Proc. Zool. Soc. (London) , 133: 135-144. . 1961. Two kinds of minke whale in the Antarctic. Norsk Hvalfangst-Tidende, 50: 133-141. LIHRARY B R E V I O R IMiuiseiuijii of Co:miparsitive Zoology Cambridge, Mass. August 11, 1972 Number 389 THE CHANARES (ARGENTINA) TRIASSIG REPTILE FAUNA. XIII. AN EARLY ORNITHOSUGHID PSEUDOSUGHIAN, GRACILISUCHUS STIPANICICORUM, GEN. ET SP. NOV. Alfred Sherwood Romer Abstract. A description is given of the skull and skeleton of a small ornithosuchid thccoilont. Gidcilisuchns stipcniicicorum gen. et sp. nov., from the Triassic ( ? Anisian) Chanares Formation of Argentina. The skull is of an advanced pseiidosiichian character, with a large antorbital opening, a very large orbit and a lateral temporal opening of an advanced type. The front limbs are short, being about 3/5ths the length of the hind; the tibia is nearly as long as the femur. Armor consisted of a double row of dorsal scutes. INTRODUCTION Work in the Triassic of Argentina and southern Brazil in re- cent decades has resulted in the discovery of a considerable number of new thecodonts; some 21 genera have now been described from the Triassic of this area. Most are known from incomplete remains; howexer, in the last paper in this series (Romer, 1972), I have been able to give a fairly complete skeletal restoration of the long-snouted Chanaresuchus, and be- low I give a description of the nearly complete skull and skeleton of a small ornithosuchid. Gracilisuchus stipanicicoruvt, gen. et sp. nov. Cotnbined generic and specific description. A small ornitho- suchid, with a skull length on the order of 95 mm and a pre- sacral column of 23 segments, with a length of about 21 cm. Skull of advanced pseudosuchian type. Premaxillae extending upward back of nares, excluding the maxillae from that open- ing. Antorbital fenestra large, included in a recessed area of 2 BREVIORA No. 389 maxilla and lacrimal; antorbital bar moderately narrow. No pineal opening- ; a tiny postparietal bone present. V-shape of lateral temporal opening so pronounced that the upper part of the opening is closed by apposition of squamosal to postorbital and jugal. Basicranial kineticism lost; the pterygoids meet me- dially beneath the basisphenoid region. Lateral flanges of ptery- goids highly developed, extending direcdy outward the entire width between lower jaws. Normal stance possibly bipedal; femur and tibia combined about 1 1/2 times skull length and about 1 2/3 the length of humerus plus radius; tibia and fibula somewhat shorter than femur. Dorsal scutes are about 1/2 ver- tebral length and paired, each element having a \ertical lateral portion and a horizontal median flange that overlaps its mate. The generic name refers to the obviously graceful build of the little reptile. The specific name is in honor of Drs. Pedro and Maria Stipanicic, able students of Triassic stratigraphy and palaeobotany. I am indebted to National Science Foundation grant GB-2454 for aid in the collecting of the specimen and grant GB-22658 for preparation and publication. Holotype. La Plata Museum No. 64-XI-14-11 (Field no. 146). A slab (Fig. 1) exhibiting a skull in dorsal view, much of the presacral column and scattered limb and girdle material. Also present on the slab is the type material of Lagosuchus talam- payensis (Romer, 1971), and originally a gomphodont skull was likewise present. Collected from the Chaiiares Formation in La Rioja Province, Argentina, about 2 km north of the Rio Cha- fiares. Other yyiaterial. A number of further Chaiiares specimens include remains of the prescHt animal. These are: MCZ' 4117 (Field no. 153, partim). A nearly complete skull and jaws from the same locality as the holotype. MCZ 4118 (Field no. 153, partim). In the same nodule as the last, but separated from it by a short distance, was a speci- men including \entral elements of the skull, the jaws and a par- tial postcranial skeleton, including a well-preserved cer\-ical re- gion. MCZ 4116 (Field no. 174). A slab including a crushed skull and jaws and considerable postcranial material, part of it pertaining to a smaller reptile. From the holotype locality. ^Museum of Comparative Zoology. 1972 GRACILISUCHUS STIPANICICORUM 3 Further Gracilisuchus material, collected by Sr. Jose Bona- parte, is in the Instituto Lillo, Tucuman. Notable is a specimen with an incomplete skull and the greater part of a skeleton that as^rees with Gracilisuchus in all identifiable features; the individ- ual is about 20 percent larger than the holotype. CRANIUM (Figures 1-4) Cranial materials are present in all four specimens listed above. The skull in MCZ 41 17 is nearly completely preserved, except for the palate, and is uncrushed. In the holotype the skull roof and right side of the face are seen on the upper side of the slab; Neu- trally there is present part of the palate and the disarticulated left maxilla and jugal. In MCZ 4118 the ventral margins of the skull and complete lower jaws are preserved. In MCZ 4116 the right aspect of the skull, crushed and elements disarticulated, is seen on one surface; on the opposite surface are disarticulated elements of the left side. Skull length to the posterior end of the table is 85 mm in MCZ 4117, and appears to have been similar in the holotype and in 4116 and 4118. The general proportions are those to be ex- pected in a moderately ad\anced pseudosuchian. As viewed dorsally, the shape is essentially triangular, expanding gradually from a slender snout to a greatest width across the temporal re- gion. Anteriorly the skull outline curves up sharply abo\e the external nares, but from this point backward there is little fur- ther increase in height, the roof being essentially flat and the height nowhere great. The external nares are moderately large. The facial length is not excessixe, the distance from snout to orbital margin being somewhat less than half the total skull length. The antorbital xacuities are large, and set in an oval recess, which is bounded anteriorh b\ a pronounced curved line running upward along the maxilla; it is deeply overhung dor- sally by the prominent lateral edge of the skull roof. The orbit is verv large, its diameter being about 1/3 the skull length. It not only occupies nearly the entire height of the face, but also extends medially across much of the dorsal surface of the skull. The temporal region, in contrast, is short. The superior temporal fenestra is triangular in shape, narrow anteroposteriorly but ex- tending broadly outward behind the postorbital bar. The lateral temporal opening is of unusual structure. .\s in advanced theco- donts generally, the posterior border is V-shaped, the apex of BREVIORA No. 389 popr. Figure 1. The skull in dorsal view. This and the following skull figures are based mainly on ISICZ 4117 XI. Abbreviations for Figs. 1 to 5: a, articular; an, angular; c, coronoid; d, dcntary; ec, ectopterygoid; /, frontal; /, jugal; I, lacrimal; m, maxilla; n, nasal; p, parietal; part, prearticular; pf, postfrontal; pw, premaxilla; po, postorbital; popr, paroccipital process; pp, postparietal; prj, prefrontal; pt, pterygoid; q, quadrate; qj, quadratojugal; san, surangular; soc, supraoccipital; sp, splenial; sq, squamosal. the V pointing anteriorly. Here, howe\er, in contrast to normal advanced forms, the forward push of the V is so pronounced that the upper limb of the V is in contact with the postorbital and jugal; as a consequence the upper half of the fenestra is completely closed, the opening remaining being a ventral triangu- lar area. A similar situation is present in the aetosaurs. The premaxillae are thickened ventrally, with accommodation for tooth roots. Anteriorly each element sends a slender process upward to meet the nasal medial to the narial openings. Pos- teriorly the premaxilla sends a stout process upward to form part of the posterior border of the naris. A pronounced ridge extends backward on either side of the skull roof along the upper margin of the antorbital region toward the upper anterodorsal margin of the orbit; this pair of ridges, formed anteriorly by the nasals, sharply separates the flattened dorsal surface of the skull from the essentially \ertical lateral surfaces. Below this rids^e the nasal extends downward to form the upper boundary of the naris. This flange is in contact with the premaxilla ventrally, both in front of and behind the naris. On the dorsal surface the nasals extend well backward, broad- 1972 GRACILISUCHUS STIPANICICORUM Figure 2. The skull in lateral view. X 1. ening posteriorly. The lateral ridges form their lateral bound- aries for most of this distance; part way along the upper margin of the facial region, however, the width of the nasals is some- what constricted by the presence of the lacrimals. Nasals and frontals meet in a broad trans\erse suture, not readily discern- ible, a short distance anterior to the anterior margin of the orbits. The frontals are elongate, their length being roughly that of the large orbits. Broad anteriorly, they are constricted in width by the strong medial curx-ature of the dorsal orbital margins, the prefrontals being interposed for some distance between them and the frontals. Posteriori)- the frontals are bounded laterally by the postfrontals, and terminate at an irregular transverse su- ture with the parietals. The last elements are short in longitudinal extent, occupying, roughly, only the length of skull roof opposite the short temporal region. There is no pineal opening; posteriorly the median su- ture is absent. Laterally the parietals send out a flange on either side, anterior to the superior temporal fenestra and behind the postfrontal, to gain a postorbital contact. Posterior to the upper fenestra the parietals send out on either side strong flanges that form the upper margins of the occipital surface and, along the posterior margins of the fenestrae, overlap posteriorly ascending flanges of the squamosals. The postparietal, seen on MCZ 4117, persists as a small trian- gular wedge of bone that projects strongly backward from the middle point of the dorsal occipital surface; it is essentiallv the most anterior portion of the dorsal armor that continues down the back. The maxilla is prominently de\eloped. Its major ramus is a long strip of bone that carries the tooth bases and extends back- ward from a premaxillary contact to a point not far forward of 6 BREVIORA No. 389 the posterior border of the orbit. Anteriorly it forms the lower margin of the antorbital fenestra. Near the posterior hmit of the fenestra the anterior end of the jugal is found above the maxilla, and from this point back the maxilla is reduced, to be- come a slender splint. Anteriorly the maxilla fails to reach the external naris. It sends upward a broad process, bounded pos- teriorly by the antorbital vacuity and anteriorly by the premax- illa, from which it is separated by a very prominent incised suture. Curving upward along the surface of this process is a well-marked ridge separating the proper outer surface from the depressed area in which the antorbital vacuity develops. Dor- sally this maxillary process extends backward below the lateral skull ridge to form part of the upper margin of the antorbital vacuity. The lacrimal forms most of the vertical bar of bone between orbit and antorbital \acuity. This bar is of limited width; it has a conspicuous ridge along its anterior border, and a less de- veloped ridge posteriorly. Below, the lacrimal meets the jugal. Abo\e, the lacrimal forms the projecting process between the orbital rim and the lateral ridge abo\'e the facial region, and extends forward above the antorbital opening. It has (unusual- ly) a modest exposure on the dorsal surface, lateral to frontal and nasal. The prefrontal is a small wedge-shaped element, lat- eral to the frontal and posterior to the dorsal exposure of the lacrimal; it forms the anterior half of the curved dorsal margin of the orbit. The postfrontal is a triangular element of modest size, lying aboxe the posterodorsal margin of the orbit. Medially it is in contact with the frontal, posteriorly with a lateral flange of the parietal; laterally it is barely in contact with the postorbital. The postorbital is centered at the point of meeting of the postorbital bar with the bar of bone separating upper and lower temporal openings. The bone is here thickened, with a promi- nent external knob, posterior and dorsal to which it is gently concave externally. A \entral flange forms much of the poste- rior margin of the orbit, oxerlapping the jugal anteriorly. Above, the bone is in contact with postfrontal and parietal along the upper part of the postorbital bar. A posterior flange joins with the squamosal in the formation of the bar separating the tem- poral openings. The jugal is of typical construction, its main ramus lying be- low the orbit and extending from a point anterior to the orbit backward to a contact with the quadratojugal below the lateral 1972 GRACILISUCHUS STIPAXICICORUM 7 temporal fenestra. Behind the orbit a ramus ascends, shmming as it goes, behind the lower branch of the postorl)ital. The squamosal is a complex element. Its structural center lies at the posterior end of the bar separating the two temporal fenestrae. Anterodorsally a thin but deep flange extends forward to overlap laterally the posterolateral extension of the parietal. Anterolateralh a ramus extends for\\ard to form, with the post- orbital, the bar between the temporal openings. A backward extension of the bone from its center affords a broad area of articulation for the paroccipital process. Beneath this region the concave \entral surface of the bone supports the head of the quadrate. Forward and somewhat ventrally from this region a broad flange of bone forms a firm union with the dorsal end of the quadratojugal. In many pseudosuchians this flange forms the upper half of a \^-shaped posterior border of the lateral tem- poral fenestra. Here, however, (as previously noted) an unusual condition exists. This flange turns so sharply forward that the upper half of the "normal" lateral opening is obliterated, and the flange is apposed to the postorbital, which forms the anterior margin of the upper part of the lateral opening under "normal" conditions. The quadratojugal is a well-developed element. It forms the posterior part of the skull rim below the lateral temporal fenestra, between jugal and quadrate. A broad ramus extends upward and forward posterior to the lateral fenestra to terminate be- neath the anteroventral ramus of the squamosal. The quadrate is highly de\eloped. It presents a convex, trans\ersely broad- ened, articular surface for the lower jaw on the under side of its posterior termination. From the medial edge of the articular area a sharply defined ridge runs upward and somewhat forward to terminate beneath the squamosal. The quadrate is broadly developed lateral to this ridge for nearly the entire height of the bone; this area is somewhat concave in transverse section. The lateral margin of the quadrate is in contact with the quadrato- jugal for most of the extent of the two bones. Not far abo\-e their \entral margins, howe\er, there appears to be a small fora- men, as in many early tetrapods, between the two bones, and quadrate and quadratojugal separate dorsally to reach their dif- fering termini. Medial to the major vertical ridge on the quad- rate there is seen a thin sheet of bone runnino- mediallv for some distance in the position appropriate for the quadrate ranuis of the pterygoid; at the bottom of this sheet is a cur\ed ridge, run- ning upward and anteromedially from the articular region of the 8 BREVIORA No. 389 pop Figure 3. Palatal surface of the skull; the anterior portion of the palate is not preserved in available material. X 1. quadrate. I am unable to determine how much of this bone per- tains to the quadrate and how much to the pterygoid. The upper rim of the occipital surface is formed by the two posterior flanges of the parietals, which curve backward, outward and somewhat downward to their lateral points of termination near the posterior angles of the squamosals. Between the two flanges, as noted earlier, is the small projecting postparietal ; be- neath, on either side, are the posttemporal fenestrae. Below and beneath the central part of these flanges is a broad plate of bone, the supraoccipital. It is essentially flat, but with a slight dorso- ventral swelling in the mid-line. Fused with the supraoccipital on either side are the paroccipital processes, which extend out- ward to form a broad (but not tightly sutured) union with the squamosals. The paroccipital processes are relatively narrow proximally, expanding somewhat in vertical breadth distally and, except at their bases, thin anteroposteriorly; they are roughly oar-shaped. Exoccipitals and basioccipitals have been lost on MCZ 4118. They appear to be present in the crushed remains of the occiput in the holotype, but little detail can be made out. The palatal surface is poorly preserved in available specimens. Posteriorly is a pair of well-developed basisphenoidal tubera. I ha\'e obtained no data regarding the lateral walls of the brain- case in the otic region. No stapes has been discovered. The two quadrate rami of the conjoined quadrates and pterygoids slant strongly inward anteriorly, so that the two pterygoids are 1972 GRACILISUCHUS STIPANICICORUM 9 Figure 4. Occipital view of the skull. X 1. apposed in the midline below the basisphenoid; it seems reason- ably certain that moxement of pterygoids on the braincase had been lost. A short backward median projection is present on each pterygoid at its point of apposition. Transverse pterygoid flanges are highly developed, extending straight out laterally and somewhat \entrally to occupy the total space available between the lower jaws when in occlusion. A pair of ridges extends out- ward along the under surface of each flange. Anterior to the flange, a curxed sheet of bone, on which there appears to be a suture between pterygoid and ectopterygoid, slants upward an- teriorly for a short distance. No data are available regarding the anterior portion of the palate. The jaws (Figs. 2, 5) are slender. The symphysis is moderate- ly elongate but shallow and is formed entirely by the dentaries. The dentary occupies practically the entire outer surface of the jaw for more than half its length. Posteriorly the dentary has a V-shaped suture with the surangular dorsally and a diagonal suture with the angular \entrallv. Between dentarv and surangu- lar abo\e and angular below is a typical archosaur lateral man- dibular fenestra. There is a well-de\eloped retroarticular process behind the broadly concave articular cavity. A splenial is well developed, occupying a considerable area on the inner surface of the jaw, but is barely visible externally. The prearticular forms a buttress at the anterior marsfin of the articular surface and thence extends forward and downward below the adductor fossa. The jaw is strongly compressed mediolaterally, and hence the large fossa looks medially rather than dorsally. The dorsal rim of the fossa is marked by a well-developed longitudinal ridge along the upper margin of the surangular. I am not sure whether or not a coronoid was present at the anterior margin of the fossa; the material is imperfect but suggests that a thin sheet of coronoid may have been present anterior to the adductor fossa. 10 BREVIORA No. 389 son Figure 5. Internal surface of the lower jaw. Sutures in the presinned coronoid region are uncertain. X 1. The teeth of Gracilisuchus are of the typical thecodont type common among carnixorous archosaurs — somewhat flattened mediolaterally, sharp-pointed and curxed somewhat posteriorly toward their tips. Only two small teeth are definitely preserved in the only specimen in which the premaxillae are present, but the space available suggests the presence of a third. A disarticu- lated maxilla of the txpe skull has fortunately preserved a nearly complete series of maxillary teeth, not fully thecodont. About 14 appear to have been present. From a small first tooth, there is a steady increase in size to the fourth, following which there is a steady reduction to small elements for the last half dozen of the series. The lower teeth are not fully preserved, but the evi- dence suggests that none were of large size, and that there was a rather e\en row of small teeth, spaced about 2.5 mm apart, to a total of 16 or so. AXIAL SKELETON A considerable amount of xertebral material of Gracilisuchus is present in the collection. The holotype when entombed pos- sessed a complete articulated presacral series ( Fig. 6 ) . However, the nodule in which the specimen was preserved had undergone considerable damage before collection. A spht had occurred in the nodule that slanted back down most of the length of the series of dorsal vertebrae and, with the loss of a large chip adja- cent to it, caused the complete loss of the posterior cervical and anterior dorsal vertebrae and damage to the remainder of the dorsals. The second and third vertebrae from the end of the series are obviously sacrals. Most of the ribs of the right side are preserved in their original position. This aids greatly in de- termining the spacing of the missing vertebrae, as does, further, a calculation, from known lengths of cervical and posterior dor- sals, of the number of vertebrae contained in the missing seg- ment of the column. As a result, it seems rather certain that 23 presacrals were present in life — a reasonable thecodont count. 1972 CKACII.lSLCll us S'lIPANICICOKL'.M 11 X u ^ h tc be 'iZ 12 BREVIORA No. 389 In MCZ 4118 a number of short sequences of articulated vertebrae are present: ( 1 ) A series of cervicals articulated with the fragmentary skull. (2) Six articulated centra that appear to lie in the cervical- dorsal transition. (3) Six vertebrae, three well preserved, lying close to the skull-cervical series; they are probably anterior dorsals. (4) Five poorly preserved vertebrae, presumably dorsals. (5) Nine vertebrae, probably posterior dorsals, sacrals and proximal caudals. (6) T.WO poorly preserved dorsals, with ribs. (7) Ten caudals, probably from the middle of the tail. These series will be referred to below by number. Excluding' no. 6, 35 vertebrae are present, most of which are surely presa- crals. It is obvious that not all belong to one individual; some, apart from Jios. 1 and 3, may have been associated with MCZ 4117, a skull found in the same nodule. The Tucuman specimen mentioned aboxe has present much of the cohunn; in all observed respects it agrees well with the material from the holotype and no. 4118. MCZ 4116 contains a melange of vertebral material, some of which appears to belong to Gracilisiiclms (as does the skull included in this slab). The cerx'icals are best seen in the series (no. 1 ) connected with the skull remains in 4118 (Fig. 7); a similar but less well-pre- served series is present in the holot\pe. The atlas is partially concealed by other materials, but as far as can be seen, has a typical archosaur structure — a well-developed intercentrum, above which arc paired neural arches, and back of this articular ring a small atlas centrum and a small axis intercentrum. The axis has a well-developed centrum and a neural arch and spine of stout construction. The vertebrae in the cervical region pos- terior to the axis are similar in pattern, but with a less expanded neural spine. In these vertebrae the spines slant forward dor- sally; in typical members of this series they are, as preserved, covered by dermal armor at their tips; isolated posterior cervi- cals and anterior dorsals show that their tips were flattened, obviously for close apposition of armor scutes. Two well-de- veloped apophyses for rib articulation are seen on each centrum from the axis backward. The capitular articulation is a short parapophysis developed low down anteriorly on the centrum and terminating in a round articular area. 'Fhe area for tubercular attachment is a short transverse process slanting sharply down- 1972 (iRACILI.SUCHUS STIPANIGICORUM 13 Plate 1. Skeleton of Gracilisiirhiis stipanicicoiiini, restored. Dermal slioiililer elements, manus. pubis, and distal part of tail unknown. X 1/2 14 BREVIORA No. 389 Figure 7. a. Dorsal scales of the cervical region, seen lioni above, b, The cervical region in side view (the atlas is not induded) . r. Incomplete rihs ol ihe riglil side, hansilional l)el\veen cervical and th)rsal series, d, A doisal vertel)ra in hiteial and anterior views. All from MCZ 4118. X 1. ward fioni a point near the anterior edge of the neural arch. By the tinie the seventh \'ertebra is reached, the diapophysis has moved to a point somewhat higher on the arch and is di- rected more laterally. The parapophysis, lunvever, appears to have been persistently ventral in position in the cervical region. No intercentra are preserxed ])ack of the atlas. In MCZ 4118 cervicals two to six are completely preserved, the seventh is in- complete; in the holotype six vertebrae are similarly preserved. In MCZ 4118 the typical cervicals average 10.5 mm in length and in the holotype the length is approximately the same. From the axis back all centra are keeled \entrally. Of the posterior cervicals and anterior dorsals, nothing can be made out on the holotype. Associated, however, with 4118 are two series of vertebrae that appear to represent this area. One such series (no. 2) includes six articulated centra and a fraction of the next posterior centrum; central lengths here average 8.5 mm. Neural arches and spines are not preserved, but ribs are present on four, the first two well preserved. Since the first rib of the series resembles that which (as noted later) appears to be associated with vertebra 7 in the holotype, it is reasonable to assu!ue that this series includes \crtebrae 7—12 as well as part of centruiu 13. In presumed xertebrae 7-9 the capitulum still ar- ticulates well down the side of the centrum, and the build of 1972 GRACILISUCHUS STIPAXICICORUM 15 tubercular process indicates that in these segments the transverse process still slanted strongly downward. Further vertebrae from the anterior part of the column are present in series no. 3, thiee in articulation and well preser\-ed. ^Ve are dealing here with typical dorsals (Fig. 7). \>rtebral lengths a\erage close to 8 mm. Rib attachments are markedly different from those of the series described abo\e. The diapophy- ses now extend directh out from the arches. In the first of the three complete \ertebrae the parapophysis lies about at the boundary between centrum and arch at the anterior edge of the %-ertebra ; in the third member of the series, the parapophysis is well up on the side of the arch, not far hdow and anterior to the base of the diapophysis. The neural spines are broad antero- posteriorly, but low, with a height of but 7 mm above the zyga- pophysis. Their upper ends are broadly o\al, for apposition to the dorsal armor. In the arch region the surface is poorly pre- ser\-ed, so that little can be seen regarding rib articulations. Series 4, 5 and 6 are poorly preserved and crushed, so that it is difficult to determine the nature of the apophyses, and no ribs are attached, except in no. 6 where the vertebrae show little). An ilium lies close to the posterior end of series 5, sug- gesting that this sequence may include proximal caudals and sacrals as well as posterior dorsals. No chevrons are present, and the condition of preser^•ation of the lateral surfaces of the verte- brae is such that it is impossible to determine whether sacrals are included. It seems clear that, as in archosaurs generally, the capitular articulation has joined the tubercular one in originat- ing from the transverse process; details, however, are not clear. In no. 5 central lengths are about 8 mm. In series 6 the mean central length is a little o\er 7 mm; the neural spines are broad anteroposteriorly but low, rising only about 5 mm abo\"e the zygapophyses, and the total height of a vertebra is about 13 mm. As preserved, the centra in series 5 and 6 are prominently keeled \-entrally and thin from side to side, but this may be due to post-mortem compression. One further series of \-ertebrae on the 4118 slab, no. 7, in- cludes 10 articulated caudal vertebrae, evidently from the mid- portion of the tail. Central length of an anterior member of this series is 6 mm. that of posterior members 5.5 mm. Chevrons are present, the most anterior one being 10.5 mm in length; in contrast, that between the sixth and se\enth of the series is but 7 mm long. The vertebrae are still fairly tall, an anterior mem- ber measuring 8 mm from bottom of centrum to zygapoph\"sis, 16 BREVIORA No. 389 while at the posterior end of the series this height is somewhat reduced. The first two members of the series have well-developed spines, slanting backward and rising to a \ertical height above the zygapophyses of 6 mm. Posteriorly the spines are reduced, and at the end of the series are merely low triangles above the zygapophyses. As noted earlier, sacral vertebrae are present, although poorly preserved, in the type. The Tucuman specimen (Fig. 8) in- cludes well-preserved sacrals in articulation with ilia. Beyond them are 16 articulated caudal vertebrae. Anterior members of this series have a central length of about 9 mm, posterior seg- ments about 7.5 mm. In the anterior members of the series the neural spines are narrow anteroposteriorly, but moderately high, with heights of 10 mm or so above the zygapophyses; they are tilted posteriorly. In the posterior members of this series the spines are much reduced in height. The first chevron is present between vertebrae 3 and 4. Most chexrons are imperfect dis- tally, but that between vertebrae 8 and 9 is 1 7 mm long. If we compare this series with that of no. 7 in MCZ 4118 — and take into account the larger size of the Tucuman specimen — it would seem that the 4118 series includes segments comparable to the posterior part of the Tucuman series. Vertebrae present on the holotype slab, although disconnected from the main specimen, appear to represent much of the length of the tail. One series of seven \ertebrae, and part of an eighth, is articulated and for the most part well preserved. Each meas- ures 8 mm in length. The neural spines are low, and capped by dermal plates to give a total height of about 8 mm above the le\el of the zygapophyses. The transverse processes are incom- pletely preser\ed. There then follows a series of about 14 par- tially disarticulated and poorly preserved vertebrae which, as indicated by long transverse processes, are definitely caudals. Of several that are moderately well preserved, the length of the cen- trum is, again, about 8 mm; the width over the transverse proc- esses is 15 mm. Beginning near the end of this series is a further series of about 15 poorly preser\ed caudals of smaller size, with lengths of 5.5 to 6 mm in the distal members. There is thus defi- nite evidence of 39 caudals making up a considerable part of the presumed length of the tail. Cervical ribs (Fig. 7) are well seen in both 4118 and the holotype. There is no evidence as to the presence or absence of an atlantal rib. Typical members have the highly specialized pattern developed among certain early archosaurs and retained 1972 GRACILISUCHUS STIPAXICICORUM 17 todav in crocodilians. Each rib is essentially plow-shaped, the two "handles"" represented b\ conxerging rami running outward from parapophysis and diapophysis. The "blade" of the plow, formed distally by the union of the two "handles," includes a short point anteriorly and a long posterior extension. In typical cer\icals each "blade"' overlaps its more posterior neighbor to make a continuous rib series from the axis as far back, at least, as vertebra 7. As in the case of the vertebrae, the rib transition to the dorsal series is somewhat uncertain. In series 2 of no. 4118 (Fig. 7), four incomplete ribs are present on the left, and one on the right. The first three ribs extend strongly forward from the point of union of capitulum and tuberculum to form a triangular sheet of bone, the front end of which underlies the next anterior rib. The main shaft of the rib extends outward and downward in line with the tuberculum, rather than turning backward as in typical cervicals; shaft lengths are uncertain. The width of the proximal rib expansion presumably associated with serratus muscles) decreases from the first to the third of the series, and rib 4 appears to lack any expansion. In the holotype most of the dorsal ribs are preserved. Mid- dorsals have a\"erage lengths of 62 mm. There is, as expected, a diminution in length toward the sacrum. The fourth presacral rib measures but about 46 mm, the three following, as preser\ed, 39, 18 and 10 mm. The main shaft of typical dorsal ribs aver- ages but about 1 mm in diameter for most of their length; they thicken somewhat toward their heads. The direct proximal course of each rib is toward the tubercular attachment; the capit- ular head slants downward medially from the course of the shaft to extend se\eral millimeters further than the tuberculum. The ribs show considerable curxature proximally, little distally, thus suggesting (reasonably) that the trunk was relati\ely high and narrow in its proportions. DERMAL ARMOR As in manv other pseudosuchians, Gracilisuchus was armored dorsallv. Best preserved is the armor of the cervical region (Fig. 7). The plates, thin but highly sculptured, are paired, and ap- proximately two pairs are present for each vertebral segment. When articulated, the series of plates form a dorsal shield with a flat area, about 5 mm wide, running longitudinally down the column over the neural spines and with, on either side, a ver- tical sheathing about 5 mm in height. A sharp ridge separates 18 BREVIORA No. 389 dorsal and lateral portions of each plate laterally; each plate oxerlaps its posterior neighbor. The plates of either side join to form the dorsal area, each member of a pair overlapping dor- sally (as laterally) its posterior neighbor, and with members of each pair overlapping its partner; in 4118 the left plates overlap the right. Anteriorly, over the atlas region, the plates appear to narrow dorsally as a pointed terminus, but details are uncertain. The plating of the dorsal region is imperfectly preserved in available material. Only small plate fragments are present in the isolated dorsal series found with 4118. In the badly pre- served dorsal series of the holotype, a lateral plate covering is seen in the area of the third to fifth presacral neural spines, in- dicating that the type of plating seen in the cervical region was continued down the back. In the holotype the series of proxi- mal caudals described definitely carry dorsal scutes, although details are somewhat obscure. No articulated series of abdominal ribs is preserved, but in the holotype a scattered series of typical archosaur gastraHa are present in the area near the shoulder girdle and front legs de- scribed above. Those gastralia that are completely preserved measure about 35 mm in length. They are essentially straight for most of their length, but gently curved toward their presumed medial ends. APPENDICULAR SKELETON Much of the girdles and appendages are preserved, although partially disarticulated, in the holotype; a number of elements are preserved in MCZ 4118; disarticulated elements are to be found on the MCZ 4116 slab. In Gracilisuchus the front limbs are much shorter than the hind, and hence, as expected, the shoulder girdle is of small size. A right scapulocoracoid (Fig. 8) is present in the type (as well as an incomplete left scapula ) . The height of the scapula is 24 mm. The structure is typically thecodont; the scapular blade is slender, but ventrally the anteroposterior width increases to 1 1 mm before the anterior margin retreats to the clearly marked scapulocoracoid suture. The back margin ventrally is strongly bevelled off for a prominent area of articulation for the humerus, facing diagonally outward and backward. A some- what less marked articular area is present below on the coracoid. This latter element, with rounded borders, is much broader an- teroposteriorly than dorsoventrally, the measurements concerned 1972 GRACILISUCHUS STIPANICICORUM 19 Figure 8. a, Right shoulder girdle, humerus, radius, and ulna. From the holotype. b. Left ilium and ischium, c. Sacral ribs and ilia, seen from above. From a specimen in the Instituto Lillo, Tucuman. All X 1. beino- 1 7 mm and 1 1 mm ; there is thus no suggestion of croco- diloid build. Just below the suture with the scapula the coracoid is slightly notched anteriorly; below this point the bone expands anteriorly to a considerable degree. A coracoid foramen is pres- ent anteroventral to the articular area. A similar scapulocora- coid is present on MCZ 4116. Of dermal girdle elements, none are identifiable with certainty in the material studied. Of pehic girdle elements ,Fig. 8), iha are present in the holotype, in MCZ 4116, MCZ 4118, and in the Tucuman specimen men- tioned. The element is of generalized thecodont proportions, with an iliac blade extending only a short distance anterior to the acetabulum, but much better developed posteriorly. The upper edge of the blade is thin; below, howe\er, it swells con- \exly on the inner surface to allow for exca\ation of the aceta- bulum externally. The acetabular margin, semicircular in out- line, is well defined ; it is deeply incised into the bone, particularly anteriorh , where the margin de\elops as an oxerhanging shelf. The lower margin of the bone is convex in outline, with some differentiation of pubic and ischiadic contacts. It is ob\ious that the pelvis was imperforate. 20 BREVxORA No. 389 Little evidence of the ventral elements of the girdle can be identified in the material. I have found nothing in the available specimens that I can identify with confidence as pertaining to the pubis. Imperfectly seen ischia are present in the Tucuman specimen and a pair of conjoined ischia are present in MCZ 4116. Unfortunately, the front margins of the ischia are imper- fectly preserved, so that only a fraction of the acetabular margin is present and nothing can be said concerning the relations of ischium and pubis. The bones are bladelike, tapering posteriorly and having a long median contact between the two elements. Of the short front legs, the long bones of both sides are present in semi-articulated fashion close to the right scapulocoracoid of the holotype (Fig. 8). A humerus is present in 4118, and two incompletely preserved specimens are present in 4116. Except for the last, the bones are exposed from the \'entral surface. The structure is that typical of primiti\'e archosaurs generally — hour- glass shaped, moderately expanded at either end, and constricted at mid-length of the shaft. There is a well-developed deltopec- toral crest. Distally, there is a circular convex area for articula- tion with the head of the radius; lateral to this the bone is somewhat notched for reception of the olecranon. The right radius and ulna ( Fig. 8 ) are present and articulated with the humerus in the holotype; the left radius and ulna are incomplete distally. I have not been able to identify these ele- ments in other specimens. As preserved, they show little char- acter; both are slender elongate cylinders. The ulna is some- what expanded proximally, but there is no olecranon ossification. Regrettabh there are no identifiable remains of the manus preserved. Femora (Fig. 9a) are present in the holotype, nos. 4116, 4118, and the Tucuman specimen mentioned earlier; in the 4116 slab there are four femora of appropriate size and shape, indicat- ing the presence of two indi\iduals of Gracilisuchus. The bone has the typical sigmoid curvature of a proper archosaur. The head is turned somewhat medially from the shaft, but this curva- ture is less pronounced and the distinction between head and shaft less marked than in Lagosuchus, for example. There is no evidence of the presence of a "fourth trochanter." Tibia and fibula are present in articulation with the femur on the right side of the holotype, and these elements are present also in nos. 4116, 4118, and the Tucuman specimen (Fig. 9b and c). These elements are long and slender Ijut where associated are never- theless somewhat shorter than the femur. As always, the tibia is 1972 GRACILISUCHUS STIPANICICORUM 21 d Figure 9. a, b, c, Left femur, tibia, and fibula, d. Left pes, composite. All X I. much .stronger than the fibula, with an expanded triangular head, a well-developed cnemial crest, and a broadh o\al distal area for astragalar articulation. The fibula is moderately broad but flattened throughout, with a gently sigmoid cur\ature. The pes (Fig. 9d) is incompletely known. In the foot per- taining to the right leg of the holot\pe the usual two small distal tarsals are present. But little is preser\ed of the two proximal tarsals, and nothing worthy of description can be made out from tarsal remains associated with tibia and fibula in MCZ 4118. Astragalus and calcaneum are, howe\'er, moderately well pre- served in the Tucuman skeleton mentioned earlier. The astraga- lus is a stout element, broad mediolaterally and moderately deep proximodistally; the anterior surface is somewhat conca\-e. At its lateral margin the astragalus, as articulated, is in contact with the fibula, and below this area a diagonal surface of contact with the calcaneum is present. The calcaneum is less developed proximodistally. Details of the astragalocalcanear articulation cannot be determined; the calcaneum, howex'er, had a well-de- \eloped "'heel"" in the fashion of crocodilians and manv pseudo- suchians. 22 BREVIORA No. 389 Data on the digits of the pes are available only in the holo- type and the Tucuman skeleton; the foot of the former is of the right side, of the latter, the left. Neither is complete distally. In the holotype metatarsals I-IV are present, but metatarsal IV is incomplete. Metatarsal IV is notably more slender than I-III. Metatarsal lengths as preserved are: 12 +, 23, 28 and 24 + mm. In the Tucuman specimen all five metatarsals are present, with lengths of 24, 28, 33, 32 and 18 mm; metatarsal V is of the "hooked" type, pointed distally. On the holotype no phalanges are present on toe I, but on digit II all three phalanges are pres- ent, with lengths of 7, 6, and 8 mm. On toe III a single phalanx is imperfectly preserved ; no digits are present with metatarsal IV. In the Tucuman specimen, the two phalanges of digit I are present with lengths of 6 and 8 mm. On digit II all three phalanges are present, the third incomplete, the first two with lengths of 7 and 4 mm. On digit III three phalanges are present, but only the first, with a length of 10 mm, is well preserved. With digit IV there is found only a single phalanx, 7 mm long. Allowing for a disparity in size of about 20 percent, the data from the two specimens agree well and allow a complete con- struction of the foot except for the more distal phalanges of toes III and IV. Assuming that as regards these phalanges, the pattern is that found in other advanced thecodonts and primitive saurischians, the restoration shown in Figure 9d cannot be far from the actual condition. It may be noted that in the specimens with articulated foot material, the toes are closely appressed to one another, as in my figure, with no trace of the fan-shaped spreading seen in many reptiles. RESTORATION Between the various available specimens, nearly the entire skeletal structure of Gracilisuchus is identifiable, and hence a skeletal restoration is justified ( Plate 1 ) . As usual in archosaurs in which the front limbs are notably shorter than the hind, the question arises as to whether a bipedal or quadrupedal pose is suitable. In the case of Chanaresuchus (Romer, 1972) I re- stored the animal as a quadruped, despite considerable disparity in limb lengths, because of the probably amphibious, crocodile- like nature of the animal. Gracilisuchus, as already mentioned (and discussed below), is quite surely a relative of Ornithosuchus, and I have followed Walker's restoration of that reptile in re- storing Gracilisuchus as a biped. It is possible that the normal 1972 GRACII.ISUCHUS STIPAXICICORUM 23 pose of GrncUisuchus was a qiiadrupechil one; however, I feel sure that, if pressed, this animal was able to run in the l)i])ccl:il manner in \vhich I have restored it. RELATIONSHIPS It is quite clear, I think, that Gracilisuchus is a relati\-e of Ornilhosuchus of the later Elgin beds of Scotland, ably described by \Valker in 1964. The skull structure is closely comparable in most resrards, as are \arious postcranial features. Walker has suggested certain late Triassic saurischian genera as possible Orni- thosuchus relatives which might belong with this genus in a common family Ornithosuchidae, and Bonaparte (1969b) has recognized two Argentinian genera (Venalicosuchus and Rio- jdsuchus ) that seem quite sureh to belong in this family. Gracili- suchus, from the Chafiares beds — quite probably Anisian in age — is the oldest (and smallest) of forms that may pertain to this apparently common and perhaps widespread Triassic family. In a fe\v features (such as the partial closure of the lateral tem- poral fenestra) the genus Gracilisuchus is perhaps slightly aber- rant, l:!ut it seems quite surely close to the base of this stock. \\'alker argues further that Ornithosuchus is a carnosaur — a proper dinosaur rather than a thecodont ancestor of dinosaurs. I provisionally adopted this interpretation when I published my 1966 edition of Vertebrate Paleontology. I confess, however, to now having reservations on this assignment (cf. Bonaparte, 1969a). Certainly the ornithosuchids show a number of features that might be expected in a carnosaur ancestor. But in certain features Gracilisuchus surely is below a proper dinosaur ''grade" in structure: the apparent complete closure of the acetabulum, for example. Again, saurischians are completely devoid of armor, and I would be loath to believe that the dorsal armor seen in Gracilisuchus and Ornithosuchus would ha\e been developed and secondarily lost. I do not wish to enter the controversy over the evolution of tarsal structure, but the presence in ornitho- suchians of a "crocodilian"' type of tarsus is, to say the least, an argument against placing the family in the Carnosauria, although not necessarily debarring the group from an ancestral position. Certainly the ornithosuchids show a trend in de\elopment that is in manv wavs similar to that which led to the carnosaurs. But for the time, it is, I think, better to regard them as forms related to and paralleling the line leading to the carnosaurs rather than members of that group. 24 BREVIORA No. 389 As our knowledge of thecodonts increases (as it is currently doing at a rapid rate), it seems clear that while crocodilians, pterosaurs, bird ancestors and ornithischians have struck off on a variety of "tangents," the saurischians show merely an im- provement on structural patterns already evident among the thecodonts. It has been generallv held that the Saurischia are a "natural"' order, monophyletic in origin. But recently it has been advocated (Charig et al., 1965) that the sauropods are only distantly related to the "theropods" and may have arisen independently from ancestral thecodonts, and it is not impossible that among "theropods," carnosaurs and coelurosaurs may ha\e had independent origins (cf. Bonaparte, 1969a). I am consti- tutionally allergic to unnecessary ad\'ocacy of polyphyletism, but I fear that we are as yet far from a solution to questions of re- lationships between various thecodont and saurischian groups. TABLE I Lensth of limb bones, in nun Holotype MCZ 4118 MCZ 4116 Hiuncius 38 44 44 Radius 30 — — Ulna 31 — — Femur 58,60 68 64,62,61,60 Tibia 56 59, 64 61,56 Fibula 55 64 47 + Tucuman specimen 81 73 65 REFERENCES Bonaparte, J. F. 1969a. Comments on early saurischians. Zool. J. Linn. Soc, 48: 471-4S0. 1969b. Dos nuevos "faunas" de reptiles triasicos de Ar- gentina. Gondwana Stratigraphy, lUGS Symposium 1967: 283-306. Charig, A. J., Attridce, J., and Crompton, A. W. 1965. On the origin of the sauropods and the classification of the Saurischia Proc. Linn. Soc. London, 176: 197-220. RoMER, A. S. 1966. Vertebrate Paleontology. Chicago, 111.: Univ. of Chicago Press. 1971. The Chanares (Argentina) Triassic reptile fauna. X. Two new but incompletely known long-limbed pseudosuchians. Breviora, No. 378: 1-10. 1972. The Chanares (Argentina) Triassic reptile fauna. XIL The postcranial skeleton of the thecodont Clianarcsuchus. Breviora, No. 385: 1-21. Walker, A. D. 1964. Triassic reptiles from the Elgin area: Ornithosuchus and the origin of carnosaurs. Philos. Trans. R. Soc. London (B), 248: 53-134. LIHRARY MAR 1 8 1985 HARVARD BIREVIORA Rliuiseom of Coixiparatave Zoology Cambridge. Mass. August 11, 1972 Number 390 THE GHANARES (ARGENTINA) TRIASSIG REPTILE FAUNA. XIV. LEWISUCHUS ADMIXTUS, GEN. ET SP. NOV., A FURTHER THECODONT FROM THE GHANARES BEDS Alfred Sherwood Romer Abstract. Incomplete remains of a new thecodont from the Chanares Formation. Leicisuchus admixtus. are described and figured. Incomplete skull remains indicate that the posterior border of the lateral temporal opening was nearly vertical; the basal articulation with the palate was freely movable. A maxilla indicates elongation of the snout. Much of the presacral column is present; the cervical vertebrae are somewhat elongate, the ribs unspecialized. Scapulocoracoids are preserved, but no pelvic mate- rial; limb material is incomplete and disarticulated; femur and tibia are slender and of approximately equal length; the pes is long and slender. A single row of thin dorsal scales is present. Lewisuchus is a relatively primitive pseudosuchian that may be related to coelurosaur ancestry. INTRODUCTION After much preparation and a general survey of the Chanares collection, it became apparent that six thecodonts of \arious sorts were present in the material; these ha\-e been described in pre- \ious papers in this series. Recently, ho\ve\er, Mr. Arnold Lewis, in preparing a concretion containing a mixed assortment of reptilian remains, discovered that in addition to parts of a gom- phodont skeleton and miscellaneous materials of small thecodonts, there was present a considerable fraction of a skeleton and skull of a thecodont of relati\ely good size which was obviously new, and is described below. I am indebted to the National Science Foundation grant GB- 2454 for aid in the collecting of the specimen and grant GB- 22658 for preparation and publication. BREVIORA No. 390 Figure 1. Lateral view of the skull, restored, an, angular; d, dentary; ;', jugal; m, maxilla; po, postorbital; q, quadrate; san, surangular; sq, squa- mosal. The jugal-quadratojugal suture is obscure. X 2/3. Lewisuchus adtnixttis, gen. et sp. nov. Holotype. Museo de La Plata 1964-XI-14-14, consisting of much of the presacral column, part of the skull and jaws, scapu- lororacoids and some limb material, contained in a concretion including also remains of a gomphodont and smaller thecodonts. The specimen was collected from the Ghanares Formation of La Rioja Province, Argentina, about 4 km north of the mouth of the Chaiiares River. The generic name is in honor of Chief Preparator .\rnold D. Lewis, who discovered the remains during preparation of the nodule containing them. The specific name refers to the mixture of materials in the nodule. DESCRIPTION Skull and jaws (Figs. 1-5 ) . In one portion of the nodule were found the "cheek" region of the skull in articulation ^vith the back end of the mandible, and close by, the occipital plate and basicranium, as well as a limited amount of other skull material. Separately were found an incomplete maxilla with dentition, an appropriate dentary, and a small segment of another maxilla and dentary. Of the skull remains, the region of the left lateral temporal opening is well preserved ( Fig. 1 ) . The fenestra is subquadrate in outline, relatively narrow anteroposteriorly, the posterior mar- gin descending \ertically to the region of the jaw articulation. The jugal forms most of the straight lower margin of the skull in this area, the portion preser\'ed running back from the area 1972 LEWISUCHUS ADMIXTUS O below the orbit to a contact with the quadratojugal and includ- ing a triangular process ascending between orbit and lateral temporal opening to a diagonal articulation with the postorbital. The hitter bone extends down nearly to the posteroventral angle of the orbit, narrowing distally; this \'entral poition of the bone has a vertical groove at mid-width. At the level of the top of the lateral fenestra the postorbital is thickened, with a pro- nounced transverse ridge on its outer surface. The bone is broken off shortly abo\-e this point. The quadratojugal completes posteriorly the ventrolateral skull margin. Posteriorly, adjacent to the quadrate, it expands in triangular fashion and sends a slender process, cur\dng some- what forward, up along the anterior margin of the quadrate. This process, as preser\-ed, terminates about half-way up the posterior margin of the lateral fenestra. The quadrate is nearly complete. Below, it presents a broad area for jaw articulation ; this area, howe\'er, is not ^vell preser\-ed and the surface was perhaps cartilaginous in life. Its external ramus sweeps far upward behind the lateral fenestra, with a mildly con\-ex anterior border, and %vith a conca\-e cross section. At the posterior margin of this ramus there is, as in thecodonts generally, a sharp ridge, internal to which is the ramus articulat- ing ^vith the pterygoid, little de\eloped dorsally but of consider- able extent further \entrally. Lying above the head of the quadrate is a triangular piece of bone, quite surely broken off dorsally, with its apex directed downward along the anterior edge of the quadrate, its posterior margin following the cur\ed upper edge of the quadrate. Its cur\ed anterior margin appar- ently formed the posterodorsal angle of tlie lateral fenestra. This is surely a fragment of the squamosal. It is probal)le that in life squamosal and quadratojugal were in contact along the anterior margin of the quadrate. Abox'e the pre«er\ed portions of the postorbital and squamosal are se\'eral bone fragments of indeter- minate nature which ma\" ha\'e pertained to the missing skull roof. Internal to the temporal area described above, but not articu- lated with it, is a nearly complete liraincase (Figs. 2—4) mostly in a good state of preservation. No sutures are apparent. The occipital plate is complete. Above is a broad supraoccipital area, with a median \ertical ridge dorsally. On either side the upper margins cur\'e for^■vard into the otic res^ion of the braincase: below this, on either side, paroccipital processes, broadened at their tips, run outward and somewhat downward and posteriori) . 4 BREVIORA No. 390 aiz^ Figure 2. Posterior view of the skull; roof restored, pt, pterygoid; q, quadrate; qj, quadratojugal; sq, squamosal; tbsph, basispheuoidal tubera. X 1. Figure 3. Ventral view of posterior part of skull; quadrate and dermal bones shown on right side of figure; ventral view of braincase on left. bptp, basipterygoid process of basisphenoid; /, jugal; popr, paroccipital process; ps, cultriform process of parasphenoid; pt, pterygoid; q, quadrate; qj, quadratojugal; st, stapes; XII, openings for hypoglossal nerve. X 1. bptp^ popr { ] Y \ yn.jx.x.fo tbsph Figure 4. Lateral view of braincase. hptp. basipterygoid process; fo, foramen ovalis; popr, paroccipital process, cut olT at base; ps, cultriform process of parasphenoid; tbsph, basisphenoidal tubera. Roman numerals indicate presumed region of nerve exits. X 1. 1972 LEWISUCHUS ADMIXTUS Below the large foramen magnum is the occipital condyle. Its surface is broad and somewhat subdivided posteriorly, and faces as much ventrally as posteriorly, suggesting a head posture ap- propriate to a possible bipedal pose. The basicranial region is well preserved. Just anterior to the condyle are highly de\eloped basisphenoidal tubera. Broad posteriorly, they diminish in size anteriorly as they con\erge, with a deep longitudinal median grooxe between them. Anterior to the rea^ion of the bases of the tubera are highly developed basipterygoid processes, which extend strongly downward, outward, and somewhat anteriorly. It is obvious that at their curved articular termini there was free mo\-ement between the processes and the pterygoids. Extending forward bet\veen the base of the basipterygoid processes is the slender cultriform process of the parasphenoid, incomplete an- teriorly. The upper margins of the braincase stop at a point where presumably the roof of the brain cavity was continued by the dermal roofing bones. Laterally (Fig. 4), the upper margins curve forward on either side. The upper part of the lateral brain- case walls slants medially; below this, a prominent if rounded ridge runs forward from the anterior surface of the paroccipital process, separating from the upper part of the wall a lateral surface ^vhich lies in a \'ertical plane and ventrally turns some- what medially. Three bars or "struts" constitute the lateral braincase wall connecting the upper portion of the braincase with the basi- cranial region. The most posterior, adjacent to the foramen magnum, is formed by the exoccipital, presumably reinforced anteriorly by opisthotic ossification. Two foramina are present in this strut. At least one was obxiously occupied by ner\e XII; possibly the other served for passage of the \agus ner\'e, but the opening is small and I tend to belie^•e that this was a second hypoglossal foramen, and that the vagus emerged anterior to this strut. Between the posterior and middle struts there is a large un- ossified area in the side wall of the otic region; presumably in this area lav the exits of nerve VII, the \agus foramen and the fenestra ovalis. The upper portion of the inter\al between pos- terior and middle struts is ossified, the ossification lying deep to the struts concerned. The middle strut is presumably formed by a prootic ossification, with ner\e V emerging anteriorlv be- neath it. The most anterior strut, presumably a laterosphenoid ossification, descends to meet the basisphenoid at the base of 6 BREVIORA No. 390 the basipterygoid process. Dorsally the presumed laterosphenoid bifurcates, leaving an opening (? a fenestra epioptica) between its branches and the dorsal taenia marginalis. No more anterior ossification is preserved in the braincase. The dorsal surface of the cultriform process is grooved, presumably for reception of a cartilaginous sphenethmoidal braincase segment. Much of the left pterygoid is preserved, although it is some- what displaced. The quadrate ramus is present. Its ventral border is strongly ridged, continuing a ridge present more pos- teriorly on the quadrate. Anteriorly this ridge curves upward, separating a thinner posterior portion of the ramus from a thicker anterior area; presumably this marks the boundary of the tym- panic cavity. The area for articulation with the basipterygoid process is a recess on the medial border of the bone at the junc- tion of quadrate and palatal rami. The posterior portion of the palatal ramus is preserved; there is no evidence of palatal teeth. A well-developed lateral flange appears to have extended di- rectly laterally, with a slight ventral curvature distally. A ridged medial border of the ramus is preserved for a distance. From the tip of the lateral flange the lateral, ridged, margin of the palatal plate curves forward and somewhat medially and dor- sally; for the short distance preserved, the lateral border twists sharply outward and gains contact with the jugal. It is probable that this region pertains to an ectopterygoid, and there is some evidence of a line of suture between this element and the ptery- goid. On the right side a slender splint of bone, about 7 mm long, extends outward from the braincase region anterior to the paroc- cipital process. This is reasonably identified as a stapes. No more anterior portions of the skull are present in articula- tion ^^•ith, or near, the posterior skull elements so far described. Probably pertaining to this skull are tooth-bearing elements that are certainly thecodont in nature and of appropriate size. A tooth-bearing strip of bone, 73 mm in length, is surely an incom- plete right maxilla ( Fig. 1 ) . The teeth are incompletely pre- served, but some 18 teeth or alveoli can be counted. A maximum size appears to be de\eloped not far from the anterior end, pos- terior to which there is a gradual diminution in tooth size. Ex- ternally, near the anterior end there is a curved ridge that presumably marks the border of the depression containing the antorbital fenestra. Part of the upper margin of the maxilla appears to be a finished surface bounding the fenestra. The in- ternal surface of the bone is considerably swollen in the area in which the bases of the larger teeth were contained ; above this 1972 LEWISUCHUS ADMIXTUS Figure 5. Inner view of lower jaw, restored, (iti, angular; d, dentary; san, surangular; sp, splenial. X 2/3. is a thin area that is obviously part of the maxillary extension upward anterior to the antorbital fenestra. It may be noted that the tooth row is essentially straight, without the ventral con- vexity noticeable in carnosaurs and e\en in ornithosuchids. A fragment that pertains to a left maxilla shows a series of teeth clearly decreasing in size posteriorly. As in thecodonts generally, the teeth are somewhat compressed mediolaterally, conical, sharp-pointed, and recur^•ed posteriorly toward their tips. The posterior portion of the left mandible is present in essen- tially natural relations with the cheek elements and quadrate (Figs. 1, 5). The anterior end of the preserved portion lies in the region of the external mandibular foramen, showing alcove this opening the surangular and, below^ the angular and poste- rior end of the splenial. Internally the jaw portion preserved shows the suture outline of the mandibular fossa. It is probable that in life the outer surface of the ramus was tilted strongly medially, but, even so, it would appear that the fossa faced as much medially as dorsally. Abo^•e the fossa the jaw, as pre- served, appears to show a broad horizontal shelf along the course of the surangular. The articular surface of the mandible is large, broadened lateromedially, and divided into a smaller anterior and larger posterior portion; the articulation is so oriented that the inner margin is somewhat more anteriorly placed. There is a well-developed retroarticular process and, in addition, a strong flange directed N'entromedially behind the articular area. Presumably belonging to this specimen is an isolated tooth- bearing element, obviously a dentary, 62 mm in length, which bears about 20 teeth or alveoli ( Figs. 1,5). As frequently in thecodont jaws, there is no great regional difference in tooth size along the series. It is probable, from the contours of the bone, that little is absent anteriorly. Two small foramina for blood \essels are present close to the front end ; internally a longi- tudinal meckelian groove is present, above which the bone is thickened for tooth roots. 8 BREVIORA No. 390 In Figure 1 I ha\e attempted to restore the skull in lateral view. Because of the relative length of the maxilla, it is ob\'ious that the "snout" was much longer, relatively, than in many thecodonts. Axial skeleton. Not connected with the cranial remains first described, but reasonably associated because of thecodont na- ture and proper size, is a series of 17 articulated xertebrae, be- ginning with the axis; part of this series is shown in Figure 6. Anterior to the axis are imperfect remains that appear to repre- sent the atlas centrum and axis intercentrum (no intercentra are present more posteriorly). The axis is well developed, with a relatively low but long neural spine, with a curved upper mar- gin. The cervical vertebrae are elongate as compared with the rest of the column; the axis centrum is 14 mm in length, and the more posterior cervicals are approximately similar, compared to an a\erage of 1 1 mm in dorsal members of the series. Neural spines are not well preserved in the cervical region (the axis apart). In vertebra 8 the spine is low, extending but 9 mm above the level of the zygapophyses. The back border of the spine is essentially vertical, but the anterior border slants strongly forward dorsally, so that, from a width of 5 mm across the base, the dorsal margin is 10 mm in extent. There is no thickening of the dorsal margin for armor support (such as is seen in Gracilisuchus) . In the dorsal region the spine bases are stouter, but the anterior slant of the anterior margin persists. There are some poorly preserved traces of scutes above the cervical verte- brae, apparently thin and probably in a single row. As preserved, the cervical centra appear to be thin, compressed from side to side except for prominent Ncrtical ridges at either end. Even here, howe\er, there appears to be no sharp ventral longitudinal ridge, and as we proceed posteriorly the centra be- come thicker and but s^entlv rounded ventrallv. In the cervical region the sides of the centra show a longitudinal depression; farther back, with major de\"elopment of the transxerse processes, this depression is part of a more expanded excavation, bounded above by the roots of the transverse processes. The articular area for the rib capitulum lies, in the cervical region, at the base of the ridge forming the anterior margin of the centrum; it is, however, little marked in the column as preserx-ed. Presumably this articular area ascended toward or to the arch in the posterior part of the series but there is little evidence preserved in the specimen. In the column as preserved the diapophxsis is not Figure 6. Right lateral vie^v of cen'ical and anterior dorsal vertebrae and ribs, h, head of right humerus and incomplete left humerus; sccor, inner view of right scapulocoracoid. X 1. prominently developed in the cenical \ertebrae; in the dorsal region, howexer, the trans\erse processes become prominent, ex- tendina: strona:lv outward and somewhat downward and back- ward. The process is supported ventrally by ridges extending upward to its base from both anterior and posterior margins of the centrum ; above, a stout ridge connects the base of the process with the anterior zygapophysis, and a less dexeloped ridge ex- tends to the region of the postzygapophysis. Close to the posterior end of the articulated series are three further \ertebrae of appropriate size for Lewisuchus. Details are not well preser\ed, l^ut presumably these xertebrae were from 10 BREVIORA No. 390 the "lumbar" or sacral region. Several further isolated vertebrae are present in the concretion, not well preserved except for one which is clearly a posterior dorsal. Ribs are in general incompletely preserved (Fig. 6). Close to the base of the axis centrum is a short rodlike structure, expanded at one end, which is presumably an incomplete axial rib; ad- jacent to it is a slender rod, 14 mm long, which may be an atlantal rib. There are no ribs preserved associated with cervicals 3-6; ribs are present, although incomplete, with vertebrae 7-10. All are markedly two-headed ; there is no evidence of the develop- ment of accessory processes of the rib heads described in a num- ber of other thecodonts. The preser\ed portions of these ribs, measured from the tubercula, are 20, 15, 45, and 37 mm long. Although none of the four is complete, the first two appear to be slender and close to their termini where broken off; the last two are more stoutly built and seem surely to have been true dorsal ribs. A few further incomplete rib segments are present close to the articulated rib series. In another part of the nodule are several articulated dorsal ribs of a size appropriate for Lewi- suchus; one has a length of about 80 mm. The curvature of these ribs suggests a deep but narrow trunk. Close to the three vertebrae mentioned above as possible lumbars or sacrals are two structures with triangular outlines and with a dexeloped articular area at the narrow end; these may be sacral ribs. Appendicular elements. Close to, and quite surely pertaining to the articulated vertebral series are two scapulocoracoids (Fig. 7 ) , the left seen from the outer surface, the right from the inner side. Scapula and coracoid are well fused, with no apparent suture. The scapular blade is unusually tall and slender. Dis- tally it expands somewhat; the distal margin is taller posteriorly, slanting downward toward the anterior margin. The lower part of the scapula expands both anteriorly and posteriorly, with a thickened acromial ridge anteriorly and a somewhat comparable posterior ridge buttressing the upper glenoid rim. The \entral margins of the coracoids are imperfectly preserxed, but the bone seems to be primitixe in structure, with no evidence of a crocodi- loid posteroventral extension, and without evidence of any ante- rior "incision" in the plate. There is no material interpretable as being a clavicle or inter- cla\icle of this specimen. And nothing found in this concretion can be identified as pertaining to the pelvic girdle of Lewisuchus. There is a considerable amount of limb material strewn through the concretion, but most appears referable to a gompho- 1972 LEWISUCHUS ADMIXTUS 11 Figure 7. Left, left scapulocoracoid; right, incomplete left humerus in ventral view. X 1. dont or to one or more thecodonts of smaller size. Only a small amount of material appears to be of a size and nature appropri- ate to Lewisuchus. Close to the right scapulocoracoid is the head of a humerus ob\iously of slender build and a second imperfect humerus Fig. 7) ; its head seems comparable to that just mentioned. The shaft is slender; the distal end is missing, but if extrapolated from the part present, on the analogy of Hesperosuchus (Colbert, 1952, fig. 22), the length in life must have been on the order of 70 to 75 mm. I find no elements that are interpretable as radius or ulna. A femur of appropriate size (Fig. 8) measures 105 mm in length. The bone is badly crushed proximally; it ne\'ertheless shows a well-developed greater trochanter and an apparently spherical head turned in sharply from the shaft. The bone is slender, the shaft having a diameter of but 7 mm. Un- fortunately the distal end is imperfect. A slender tibia, represented mainly by an impression in the matrix, is 106 mm in length. Near the shoulder region (but also close to the vertebrae mentioned as perhaps being in the lumbo- sacral region) is a group of podial elements (Fig. 8B). Two slender bones, measuring 31 and 30 mm in length, are surely metapodials; semi-articulated with the second are three pha- langes, the terminal one clawed. Near the first of the two meta- podials is a relati\ely long phalanx, and beyond it a second series of three articulated phalanges, terminating in a clawed element. 12 BREVIORA No. 390 Figure 8. A. Right femur in dorsal view (incomplete distally) ; B, ele- ments of second and third toes of right pes; C, the same, articulated. X 1. It is most reasonable to interpret the two metapodials and the seemingly associated phalanges as toes II and III of the right pes. When articulated (Fig. 8C), it is obvious that digit II is stronger than III, and nearly as long, suggesting comparison with the proterochampsid type of foot (Romer, 1972a). DISCUSSION Because of the inadequacy of the material, I have refrained from attempting a skeletal restoration of Lewisuchus. In default of good skull material, allocation of Lewisuchus to a definite position in the order Thecodontia is difficult. The modest trend toward strengthening of the inner toes immediately suggests com- parison with Chanaresuchus (Romer, 1971; 1972a) and the proterochampsids, an assignment with which facial elongation is compatible. But strengthening of the inner toes is not confined to the proterochampsids, and the subquadrate configuration of the lateral temporal opening is sufficient to debar Lewisuchus from the Proterochampsidae, and indeed, from the suborder Proterosuchia, in which the lateral opening is relati\ely long 1972 LEVVISUCHUS ADMIXTUS 13 with the quadrate slanting backward ventrally. Lewisuchus is thus probably assignable to the Pseudosuchia. Lewisuchus retains certain primiti\e features, such as the free basal articulation between palate and braincase, and the nearly straight back margin of the lateral temporal fenestra, which lacks the V-shape here found in many advanced forms. On the other hand, the long slender tibia, about equal to the femur in length, and the long metapodials, suggest a strong advance towards a truly bipedal gait. Lewisuchus does not appear to be closely comparable to other described pseudosuchians. It differs in many obvious features from the ornithosuchids (cf. Romer, 1972b). It is possibly re- lated to LIcsperosuchus of Colbert ( 1952 ), but this form is poorly known and, furthermore, appears to difTer in certain features, such as the incipient de\'elopment of accessory rib flanges, absent in Lewisuchus. Elongation of cervical vertebrae suggests com- parison with Teleocrater, an incompletely known form from the Alanda beds, under description by Charig (1957). Lewisuchus may possibly be a form leading toward the coelurosaurs. Not improbably Lewisuchus may exentually merit being made the type of a family of its own. But for the time being it is perhaps best left as a pseudosuchian in cert ac sedis. LITERATURE CITED Charig, A. J. 1957. New Triassic archosaurs from Tanganyika including Mandasuchus and Teleocrater. Abstr. Diss. Univ. Cambridge, 1955-56: 28-29. Colbert, E. H. 1952. A pseudosuchian reptile from Arizona. Bull. Amer. Mus. Nat. Hist., 99: 566-592. Romer, A. S. 1971. The Chanares (Argentina) Triassic reptile fauna. XI. Two new long-snouted thecodonts, Chanayesuchus and Gualosui hus. Breviora, No. 379: 1-22. 1972a. The Chanares (.Argentina) Triassic reptile fauna. XII. The postcranial skeleton of the thecodont C/ianaresuc/ius. Breviora, No. 385: 1-21. 1972b. The Chanares (Argentina) Triassic reptile fauna. XIII. An early ornithosuchid pseudosuchian, GracUisuehits st'ipanici- corum, gen. et sp. nov. Breviora, No. 389: 1-24. ML'S. CO MP. ZOOL LinnARY B R E V I O R A I^Iuseom of Comparative Zoology Cambridge, Mass. August 11, 1972 Number 391 THE RELATIONSHIP OF ISLAND AREA AND ISOLATION TO COLOR POLYMORPHISM IN LIGUUS FASCIATUS (PULMONATA, BULIMULIDAE) Michael A. Rex Abstract. Liyuus fasciatus is a highly polymorphic arboreal pulnionate living on small islands called hammocks in the Florida Everglades. Color variation in this snail has in other studies been assumed to be random and of no selective significance. Multiple regression analysis showed that hammock area was a statistically significant predictor of color morph di- versity, and that measures of isolation were either insignificant or significant but subordinate to the effect of area. From this result it was inferred that color variation in Liguus was regulated by available ecological diversity (hammock area) even when interhammock migration appeared to be ex- tensive. Color polymorphism in this snail might be imposed by visual predation, larger hammocks with more heterogeneous floral composition having more morphs because they offer more possibilities for cryptic or aposematic associations to avoid predators. INTRODUCTION Conchologists have long considered color \ariation in the re- markably polymorphic Liguus fasciatus to be random and of no adaptive significance. "On the whole it is not apparent that any form of selection has been a major factor in the exolution of Florida Liguus" (Pilsbry 1946: 47). Ecological uniformity of habitats led Pilsbry ''1912, 1946) to attribute differences in color and form between monomorphic colonies to isolation alone, and to explain variation in mixed colonies by random mutation and recombination of h\pothetical Mendelian alterna- tives. However, recent studies indicate that random genetic e\ents do not, as once thought, constitute the most important mode of evolution in land snails. Crampton's (1916, 1925, 1932) con- tention that intraspecific variation in Partula was due to genetic 2 BREVIORA No. 391 drift has been contested by selectionists both on the basis of his collecting methodology (Mayr 1942; Cain and Sheppard 1950; Ford 1964) and on reexamination of his own data (Bailey 1956). Gulick (1873, 1905) was unable to correlate variation in Achatinella with ecological diversity, but Welch (1938, 1942, 1958) found changes in both color and form with akitude in achatinellids when temperature and moisture gradients were pronounced. Diver (1940) and Lamotte (1952) proposed that color variation in Cepaea nemoralis resulted from genetic drift, though it is now evident that relative frequencies of color and banding patterns in Cepaea are regulated to some extent by visual predation (review in Cain and Sheppard, 1954) and cli- matic factors (Lamotte 1959, 1966; Wolda 1967; Arnold 1968, 1969). Furthermore, Murray (1964) demonstrated that hermaphroditism, sperm storage, and multiple matings in C. nemoralis function to maximize effective population size, thereby reducing the probability of genetic drift in small colonies. Gould (1969) disclosed the adaptive significance of both color and form in his study of temporal character variation in Poecilozonites. In the present paper I show that the number of color morphs in populations of Liguus jasciatus living on small islands can be predicted by island size and isolation. It is suggested that poly- morphism in this snail is maintained by a selective balance of phenotypes. Liguus is particularly suitable for investigating efTects of environmental factors on phenotypic diversity. The snail li%^es on small islands called hammocks in the Florida Everglades. Hammocks are slightly elevated remnants of Pleistocene coral reefs that support a tropical hardwood vegetation. They are surrounded by sparse pine woods or swamp, both of which are inhospitable to Liguus. Thus, habitat size and distance between individual populations are well defined and easily measured. LIGUUS MATERIAL Liguus colonies, once abundant in Florida, have now been decimated to near extinction by land clearing, glade fires, ama- teur collectors, and introduced rats. Fortunately, a few careful investigations were carried out before the onslaught. Material used in this study came from extensive collections made by W. J. Clench and W. S. Schevill during February and March, 1931, from an assemblage of hammocks on Long Pine Key, Dade County, Florida ( Fig. 1 ) . The hammocks ranged in size 1972 COLOR POLYMORPHIS^r ^ c >;^ ^ o ^ ± n M > c n o u o P E n C o 3 S T3 ^ u E .2: J= c 3 4 BREVIORA No. 391 from 0.34 to 43.80 acres. According to Clench (personal com- munication), an endeavor was made to collect samples of all the varieties present in each hammock \isited, but not necessarily in proportion to their natural frequency distribution (non- random ) . The material, comprising several thousand shells from 48 hammocks, was sorted by Clench as to varieties and placed in the mollusk collections of the Museum of Comparative Zoology, Harvard University. I have considered the Long Pine Key material to include nine color morphs, which are described briefly below. The morphs clearly have some genetic integrity. By examining several clutches of eggs, Pilsbry ( 1 946 ) showed that separate patterns segregated out, but the exact extent to which this occurs is unknown. Pilsbry (1946) classified the morphs of L. jasciatus into sub- species, forms, and varieties. The subspecific designations repre- sented different color patterns and had no geographic connota- tion. The forms were based on the colors present in the patterns and the presence or absence of pink coloration on the early whorls and columellar callus. The varieties were usually named stages in the continuous variation present in the forms. His sys- tem for sorting out morphs is conx'cnient and I have largely adhered to it here. But the genetic criteria on which his hierarchy was based are unacceptable. For instance, the rare variety deckerti (see below) found in three widely separate localities in Florida was thought to have arisen from the form castaneo- zonatus through mutational "loss" of the factor for pink colora- tion (1946: 70). Its white apex and columellar callus could, however, be the result of a number of conceivable schemes involving matings with morphs having white apices, dominance interactions, recombination, and close linkage. The genetic mechanisms governing coloration in these snails are probably complex and I have chosen not to make any underlying assump- tions about them. For purposes of this paper it is sufficient to recognize that L. jasciatus is a highly variable species in which more or less disjunct color morphs can be discerned and that the number of morphs present in a population is some indica- tion of its genetic variance. MORPHS PRESENT IN LONG PINE KEY Names listed are those used by Pilsbry ( 1 946 ) and the pat- terns refer to Figure 2. 1. ebiirneus: shell completely white. 1972 COLOR POLYMORPHISM Figure 2. Basic color patterns for Litjuus fasciatus present on Long Pine Key, Florida (from Piishry, 19+6). See text for explanation. 2. cingulatus : faint yellow spiral zones (Pattern A) on a white background; apex and columellar callus white. 3. roseatus: yellow spiral zones (Pattern A) on a white back- ground; apex and columellar callus pink. 4. castaneozonatus: splotchy to solid brown spiral zones (Pat- tern A) on a white background; apex and columellar callus pink. 5. deckerti: brown spiral zones (Pattern A) ; apex and colu- mellar callus white. 6. luteus: uniform faint yellow coloration (Pattern B) ; apex and columellar callus white. 7. ornntus: uniform yellow coloration (Pattern B) intensifying to amber near the lip; apex and columellar callus pink. 8. testudineus: variegated and banded (Pattern C or D), with brown coloration on white to yellowish background; apex pink with pink or \vhite columellar callus. 9. marmoratus: \ariegated and banded (Pattern C or D) with brown coloration on a white background; apex and columellar callus white. Patterns C and D are endpoints of continuous variation, the dense variegation or "flames" of C gradually superimposing the bands and lightly variegated pattern in D. There appear to be no good objective criteria for discriminating various stages in this continuous variation. Thus, the distinction between morphs 9 and 10 is one of apical coloration. METHODS Computerized stepwise multiple regression analysis using pro- gram BMD02R (Dixon, 1968) was performed to determine whether the degree of polymorphism in populations from dif- 6 BREVIORA No. 391 ferent hammocks correlated with any of several independent variables measuring area or isolation. No general discussion of the estimating equations of multiple regression is presented here as they ha\e now been so extensively employed in ecological studies (cf. Ebeling et al., 1970; Hamil- ton et al., 1963; Hamilton et al., 1964; Hamilton and Rubinoff 1963, 1964, 1967; Johnson et al, 1968; Pomeroy, 1967; Vuil- leumier, 1970). The methodologv allows one to determine both whether a statistically significant prediction of the dependent variable can be made using some or all of the independent variables, and which independent variables are most important. The square of the multiple correlation coefficient R" estimates the amount of variance in the dependent variable explained by the combined effects of the independent variables. The significance of the en- tire regression was tested by the variance ratio F, which is the mean square due to regression di\'ided by residual mean square. The significance of the contribution of each independent vari- able to the total explained variance in the dependent variable (R") was tested by using the t- ratio, which equals the partial regression coefficient divided by the standard error (Snedecor and Cochran, 1967). Three versions of the regression were car- ried out: a linear model, a semi-log model using log-trans- formed independent variables, and a log-log model. The dependent variable was simply the number of morphs present on each hammock. Independent variables were restricted to those which could be measured from a large map of the collecting site prepared by W. S. Schevill (Fig. 1). Previous studies in island biogeography, in which multiple regression models were used to predict species diversity (cf. esp. Hamilton and Rubinoff 1963, 1964, 1967; Vuilleumier 1970), suggested several independent variables. The following were tested for their ability to predict the number of morphs: (1) Hammock Area. This variable is commonly used as an index of ecological diversity. Larger hammocks were expected to have more micro- habitats; i.e., more plant species, more possibilities for cryptic associations, greater foliage height diversity, etc. (2) Distance to Hammock 27. Hammock 27 (Fig. 1) was the largest island and may have been an effectixe dispersal center for the entire group of hammocks even though it was peripherally located. It also had the fortuitous advantage of being close to Paradise Key, an enormous hammock which probably exerted influence over 1972 COLOR POLYMORPHISM 7 the morph diversity of Long Pine Key hammocks, hut which was not included in this study because data on the number of color morphs present were unavailable. This variable tested efTecti\eness of isolation from a possible source area of high phenotypic variability. Hammock 27 had eight of the nine color morphs. Three additional variables determined the effects of isolation in a more localized sense. (3) Distance to the nearest large hammock, a large hammock being defined as any of hammocks 54, 23, 26, 8, 34, 27. (4) Distance to the nearest hammock, (5) Size of the nearest hammock. RESULTS The relative influence of each independent variable and an anahsis of variance are given in Table 1. In all three models hammock area was the most significant (P < .001) predictor of the number of morphs present. Area made a higher contri- bution to R~ in the semi-log model (.39) than in the linear model (.30), an indication that the data are somewhat curvi- linear with respect to area. Variables measuring isolation, with the exception of distance to hammock 27, proved to be insignificant. Distance to ham- mock 27 made a significant contribution to R~ in the linear (P < .05) and semi-log (P < .025) models. This suggested that a slight phenotypic diversity gradient extended westward from hammock 27. If hammock 27 was, or was near, the effective population center, then perhaps the population was simply ex- panding westward. There was some evidence to support this contention in the distribution of individual morphs. Two morphs, ornatus and marmoratus, did not extend farther west than hammocks 34 and 56 respectively and deckerti was con- fined to hammocks 55 and 26. They were the least frequent in occurrence and were found predominantly on large hammocks. These morphs might still have been in the process of dispersing westward. When thev were removed from the analysis, the number of morphs still correlated significantly with hammock size (.01 ^ P ^ .001). But distance measures correlated so poorlv when added to the multiple regression models that the significance of the entire regression was reduced to P < .05 in the linear and log-log \ersions. Otherwise, the morphs showed no evident propensity for east or west. Even though distance to hammock 27 made a significant contribution to R", its effect was clearly subordinate to that of area. 8 BREVIORA No. 391 c .2 re 3 > ■^ a O.I a C fn rt C O ■a 'Z .2 c CU "i -a c c B. O 1-1 < ^ 3 -g c - •5 t o tL c ■" a. o 5 4- 3 Co « * o « h o ^ c *^ « o > c -a c u D. C o o — , o O '"^ o ""J m \r\ v/1 O V A A A A V CI, en o o- u-i wn , " r^ u-1 so O .— < ^ o •* in CM •+ >o en CNJ m >+ •^^ en II u-i r^ l-H 1— » "*■ 00 o ON .-H ^.^ en -t * 00 CM o O o rg -^ M o O' ■^ o en tn 1— t \^ o (N o o o 1-H O VTN u-^ vn O V V A A A V -+ -I- CM 'I- r*-i rsf en oo oc ^H u-i rj C7\ t^ CN) O ~' \o lJ-1 en SO U-l C<0 •t CNJ II VO tn t^ so 1-H ly^ CI t^ CM .— q en CO OO oo ■ ; ON O O o oo \o en p o O o •+ u.'^ o \r\ c o o o O CM vo u^ c V V A A A V m •+• On oo -t- rj o CM CM y-* o r^ t~^ ON t^ CM so en U-) CM u-i CSJ U~t -h CM 1 en CM SO O o .— 1— ( SO so u-i ^— ( C' lO* O so O o ON en o o o o en fc"' ^ o 0 p t-l O c .1^ ^ <-i-i r3 o o o o !> C3 CM E £ E o w 4-1 aj cr. 4J > 3 O u E u u O oj V4_ K O '^ g si 4- « C 1- C 1- o !- £ c8 eg 03 tT! 4- i ft X o C < J>^' X 4-1 CM c M-t Qj O CM OJ — o u. o o Cl U-1 u oo 1 X u , W) yj L> O -Q 3 C/5 bCCM •'—CNJ 4^ X <>J ^ C: so o 1) so '■^ c CM d 1 "* O 1 =1 T3 1 X X + X ^o" o X X bC -5 r^ — CM CM -^^ X -- ■=> o O O CM O XJ '■^ E 1 o-f >-i X c-f- .. u X ^^x a; so X f'^ (N.I ^ o — btoo — °^d < -J- '~° , r^ 1 «J ■' o c en X so 1 «4M E ^X" o 1 ^'^ u MCM M-i O "^ -—CM ^ X vo d 3 X iM oo-r u eo X' 't " -H OJ -I- "^ + OO C3 " d > VJ3 -t- ,, -a s;:^!! C3 n""^ "^ _o IL b£> r! ^ _o u .. >" __- ^1 OJ ^- n 4.1 E o 1-1 « X C SEi >, "O _o - M ."ti 0; •w . O M IS C3 3 C E M .D "J.^ u .— qj O O 0, O s k4 C/5 hJ X .2 1) 'x II M t bfi 1972 COLOR POLYMORPHISM 9 It is worth noting that in three cases independent variables measuring isolation had significant simple correlation coefficients with the number of morphs : 1 ) distance to the nearest large hammock, r =^ -.407, P < .01; 2) distance to hammock 27, r = —.398, P < .01, both in the semi-log model; and 3) dis- tance to the nearest large hammock, r = -.310, P < .05 in the log-log model. These isolation variables were de-emphasized in the multiple regression analyses because they did not correlate with the dependent \ariable as significantly as area (r = .549, .538, .623, P < .001, in the linear, log-log, and semilog \'ersions respectively ) , did not explain as much of the variance in the dependent xariable as area, and were partially redundant with one another. Associations between intraspecific variability, interspecific competition, faunal dixersity, etc. and various physical or bio- logical envaronmental parameters are often obscure. The use- fulness of multiple regression lies in the ability to discern a hierarchy of importance among independent variables relating to some dependent variable from a matrix of correlations be- tween all \'ariables. In this particular study it indicated that hammock area was more important than isolation in predicting the amount of color variation in populations of L. fasciatiis. It is possible, however, that the distance measurements and transformations used may not have been the relevant ones for predictive purposes. MacArthur and Wilson (1967: 132-133) concluded on the basis of theoretical considerations that in colonization "stepping stones" should be less important to passive dispersers than to acti\'e dispersers. Snails probably disperse passix^ely and one would expect direct distance measurements to be most appro- priate in this kind of study. When hammock-hopping distances^ were tried in place of variables 2 and 3, however, the results remained \'ery nearly the same as those presented in Table 1 . R" for the linear, semi-log, and log-log model was .39, .51 and .34 respecti\'ely, and probability levels for significant variables were identical, with the exception of \'ariable 2 in the semi-log model which was significant at P < .005 instead of P < .025 'Hammock-hopping was measured by using intervening hammocks as stepping stones rather than using direct distance. Also, measurements were made through the pine wood or to the narrowest passage through glade if glade must he crossed, the assumption being that it would be difficult for an actively dispersing pulmonate to cross bodies of water. 10 BREVIORA No. 391 (a difference of less than .02 in the contribution of variable 2 to R-). Distinction in mode of dispersal that might otherwise be indi- cated by using either direct or hammock-hopping distances could be blurred by the small size of the study area as a whole. Predatory birds and high winds are likely agents of dispersal and they probably transport snails a highly variable range of distances. Simpson (1929) occasionally found live Liguus far into the open pine wood which he thought were propagules on their way to colonize new hammocks. He gives a delightfully anthropocentric account of Liguus "obeying an instinct for founding new colonies." It seems more likely that the snails he found were dropped by birds. An analysis of variance for the entire regression for each model is given in Table 1 . All three models provided significant (P < .01) results. The best predictive value (highest R~) was afforded by the semi-log model. In all models, however, at least half of the variance in the number of morphs was left un- accounted for. This "error" (E = 1 -R") was due to several factors. First, there were undoubtedly important variables that were not considered simply because the data were unavailable. For instance, more precise measurements of ecological diversity such as the number and kind of plant species occurring on each hammock may have been important. Secondly, there were errors in measuring the independent variables used in the appraisal of polymorphism. Some morphs must have been occasionally over- looked during collecting. Finally, there is the certainty that not all of the variance could ever be attributed to ecological para- meters, part of it being due to genetic mechanisms such as balanced polymorphism ( heterozygote superiority) and part of it actually being random. DISCUSSION T infer from the inability of isolation to predict the number of color morphs and the widespread distribution of the morphs in general that interhammock migration, however it occurs, is a fairly frequent event, but that its contribution to maintaining polymorphism in Liguus is strongly mediated by hammock area. The question now arises: Why is hammock area a good pre- dictor of color variation? A speculative hypothesis is the following. 1972 COLOR POLYMORPHISM 11 Boettger (1931) first suggested, without giving supporting data, that the degree of color polymorphism in Cepaea is reg- ulated by background heterogeneity through selection by visual predation. The quantitative studies of Cain and Sheppard see re\ie\v 1954) corroborated Boettger completely. More diverse backgrounds such a-s hedgerows, rough herbage, and mixed deciduous woods supported more variable populations of Cepaea than the more uniform beechwoods or short turf. Fur- ther, cryptic associations between color morphs and background color appeared to confer a selective advantage in eluding pred- ators. For instance, in a study of Cepaea in \Vytham Woods, Sheppard ( 1 95 1 ) found that brown and pink morphs were at an advantage in mid-April when the forest floor was brow^n, the more conspicuous yellow morphs being selectively predated by thrushes. In May when the forest became green the rexerse situation obtained, yellow then being at an ad\antage. A similar mechanism might regulate morph diversity in Liguus. Larger hammocks with greater foliage height diversity and a greater number of plant species might provide more pos- sibilities for cryptic associations to avoid predators. Natural predators of Liguus include the opossum (Pilsbry, 1946), the crow, and "other large birds" (Simpson, 1929). Colonies on small hammocks having limited floral heterogeneity might con- tinually have morph di\ersity depleted by predators. Eisner and \Vilson (1970) recently suggested the obverse of this hypothesis: that coloration in Liguus is aposematic I rather than cryptic ) . ^Vhen disturbed the snail withdraws into its shell and discharges large quantities of liquid. Simpson (1929) thought this fluid secretion enabled the snail to slip from the beaks of predatory birds, but Eisner and Wilson be- lieved it to be a chemical defense mechanism. The success of cryptic or aposematic coloration depends on suitable back- ground color. Since none of the wide variety of morphs will appear conspicuous in all situations, perhaps the snail employs a mixed strategy involving both kinds of coloration. Further observations are needed. The possibility of low morph di\ersity on small hammocks being due to genetic drift is not discounted, but seems unlikely for reasons presented against Pilsbry's argument below. Other hypotheses, such as regarding polymorphism as a manifestation of niche subdivision (Van Valen 1965), are concei\able, but lack substantiating evidence. 12 " BREVIORA No. 391 An explanation for the pre\iously noted cunilinearity of the data with respect to area might be that the number of cryptic associations utilized by this snail is attained at intermediate ham- mock size, and the probability of adding a new and different association with further increase in area is negligible. Pilsbr\'s argument that \'ariation in Liouus is random seems untenable for the following reason: Interhammock migration, as already mentioned, appears to be quite extensive. Since the snail is hermaphroditic and has multiple matings a single fer- tilized immigrant can introduce considerable \ariation into a colony (Pilsbry, 1912, found at least three separate morphs in one clutch of nine eggs from a single adult ) . In the early stages of colonization, larger hammocks might be expected to ha\'e populations with greater morph diversity simply because they represent larger target areas for propagules; this might be the case with ornatus, ?narmoratus, and deckerti. The great abundance of Liguus at Long Pine Key and further north in Florida suggests, however, that the snail has populated southcn Florida for a rather long period of time, possibly since the last glaciation, and that it has essentially achieved a steady state. This being the case, and if variation occurred randomly and were unaffected by selection, then there would be no reason to expect populations living on larger hammocks to be more vari- able than those li\ing on smaller ones. The highly significant positi\e correlation of hammock area and the number of morphs (P < .001) indicates, however, that morph diversity on smaller hammocks is maintained at a low level. Visual predation might be an important selective agent reducing variation on small hammocks. Visual predators are known for Liguus and a similar mechanism has been shown to operate with another polymorphic pulmonate. ACKNOWLEDGMENTS I am grateful to S. J. Gould, K. J. Boss, T. W. Schoener, A. Schoener, and E. O. \Vilson for critically reading the manu- script. A. Clark, R. Bullock, R. D. Turner, R. I. Johnson, W. J. Clench, and R. A. Howard offered helpful suggestions or read parts of the manuscript. AV. Bossert and F, Vuilleumier pro\'ided advice on statistics. C. Jones drafted Figure 1. This research received partial support from NSF Grant GB- 19922 to Harvard University (R. C. Rollins, principal investigator) and GZ-1615 to Harvard University. 1972 COLOR POLYMORPHISM 13 LITERATURE CITED Arnold, R. \V. 196S. Climatic selection in Crpaia nrmoralis (L.) in the Pyrenees. Phil. ']"rans Roy. Soc. London, B, 253: 549-593 . 1969. The effects of selection by climate on the land- snail Ccpaca nemoral'is (L.). Evolution, 23: 370-378. Baii.ey, D. \V. 1956. Reexamination of the diversity in Pariula tacn'iata. Evolution, 10: 360-366. BoETTGER, C. R. 1931. Die Entstehung von Populationen mit bestimmter Variantenzahl bei der Landschneckengattung Crpaca Held. Zeit. ind. Abst. Ver., 58: 295-316. Cain, A. J., and P. M. Sheppard. 1950. Selection in the polymorphic land snail Ccpaca nemoral'is. Heredity, 4: 275-294. , and 1954. Natural selection in Ccpaca. Genetics, 39: 89-116. Crampton, H. E. 1916. Studies on the variation, distribution and evo- lution of the genus Pariula. The species inhabiting Tahiti. Cam. Inst. Washington Publ., 228: 1-311. 1925. Studies on the variation, distribution and evo- lution of the genus Pariula. The species of the Mariana Islands, Guam and Saipan. Carn. Inst. Washington Publ., 228a: 1-116. . 1932. Studies on the variation, distribution and evo- lutiiin of the genus Pariula. The species inhabiting Moorea. Carn. Inst. Washington Publ., 410: 1-335. Diver, C. 1940. The problem of closely related species living in the same area, p. 303-328. In J. S. Huxley [ed.]. The New Systematics. Oxford: Clarendon Press, 583 pp. Dixon, W. J. [ed.]. 1968. Biomedical computer programs. University of California Publications in Automatic Computation, No. 2. Berkeley: Univ. California Press. 600 pp. Ebeling, a. W., R. M. Ibara, R. J. Lavenberg, and F. J. Rohlf. 1970. Ecological groups of deep-sea animals off southern California. Bull. Los Angeles Co. Mus. Nat. Hist, 6: 1-43. Eisner, T., and E. O. Wilson. 1970. Defensive liquid discharge in Florida tree snails (L'ujuus fasciaius). Nautilus, 84: 14-15. Ford, E. B. 1964. Ecological Genetics. London: Methuen. 335 pp. Gould, S. J. 1969. An evolutionary microcosm: Pleistocene and Recent history of the land snail P. (Poccilozoniics) in Bermuda. Bull. Mus. Comp. Zool, 138: 407-532. GuliCK, J. T. 1873. On diversity of evolution under one set of ex- ternal conditions. J. Linn. Soc. London, 11: 496-505. . 1905. Evolution, racial and habitudinal. Cam. Inst. Washington Publ., 25: 1-269. Hamilton, T. H., R. H. Barth, Jr., and I. Rubinoff. 1964. The en- vironmental control of insular variation in bird species numbers. Proc. U. S. Nat. Acad. Sci., 52: 132-140. 14 BREVIORA No. 391 , AND I. RUBINOFF. 1963. Isolation, endemism, and mul- tiplication of species in Darwin finches. Evolution, 17: 388-403. AND 1964. On models predicting abund- ance of species and endemics for the Darwin finches in the Gala- pagos Archipelago. Evolution, 18: 339-342. , AND 1967. On predicting insular vari- ation in endemism and sympatry for the Darwin finches in the Gala- pagos Archipelago. Amer. Natur., 101: 161-171. , , R. H. Barth, Jr., and G. Bush. 1963. Species abundance: natural regulation of insular variation. Science, 142: 1575-1577. Johnson, M. P., L. G. Mason, and P. H. Raven. 1968. Ecological param- eters and plant species diversity. Amer. Natur., 102: 297-306. Lamotte, M. 1952. Le role des fluctuations fortuites dans la diversite des populations naturelles de Ccpaca ncmoralts (L.). Heredity, 6: 333-343. 1959. Polymorphism of natural populations of Cepaea nemoralis. Cold Spring Harbor Symp. Quant. Biol., 24: 65-86. 1966. Les facteurs de la diversite du polymorphisme dans les populations naturelles de Cepaea nemoralis (L.) Lavori della Societa Malacologica Italiana, 3: 33-73. MacArthur, R. H., and E. O. Wilson. 1967. The theory of island bio- geography. Monographs in Population Biology No. 1. Princeton, N. J.: Princeton Univ. Press. 203 pp. Mayr, E. 1942. Systematics and the Origin of Species. New York: Columbia Univ. Press. 334 pp. Murray, J. 1964. Multiple mating and effective population size in Cepaea nemoralis. Evolution, 18: 283-291. PiLSBRY, H. a. 1912. A study of variation and zoogeography of Liguus in Florida. J. Acad. Nat. Sci. Philadelphia, 15: 429-471. 1946. Land Mollusca of North America (north of Mex- ico). Acad. Nat. Sci. Philadelphia Monograph No. 3. Vol. 2, pt. 1 pp. 1-520. POMEROY, D. E. 1967. The influence of environment on two species of iand-snaijs in South Australia. Trans. Roy. Soc. S. Australia, 91: 181-186. Sheppard, p. M. 1951. Fluctuations in the selective value of certain phenotypes in the polymorphic land snail Cepaea nemoralis (L.). Heredity, 5: 125-134. Simpson, C. T. 1929. The Florida tree snails of the genus Liguus. Proc. U. S. Nat. Mus., 73: 1-44. Snedecor, G. W., and W. G. Cochran. 1967. Statistical methods. Ames: Iowa State Univ. Press. 593 pp. Van Valen, L. 1965. Morphological variation and the width of ecologi- cal niche. Amer. Natur., 99: 377-390. VuiLLEUMiER, F. 1970. Insular biogeography in continental regions. I. The Northern Andes of South America. Amer. Natur., 104: 373-388. 1972 COLOR POLYMORPHISM 15 Welch, D. A. 1938. Distribution and variation of Achat'incUa mustelina Mighels in the Waianae Mountains, Oahu. Bull. Bernice P. Bishop Mus., 152: 1-164. . 1942. Distribution and variation of the Hawaiian tree snail Achaiinclla apfxjulva Dixon in the Koolau Range. Oahu. Smith. Misc. Coll., 103: 1-236. . 1958. Distribution and variation of the Hawaiian tree snail Achatlnclla hiilimoidcs Swainson on the windward slope of the Koolau Range, Oahu. Proc. Acad. Nat. Sci. Philadelphia. 110: 123- 211. WoLDA, H. 1967. The effect of temperature on reproduction in some morphs of the land snail Cepaea ncmoralis (L.). Evolution, 21: 117-129. Lir.nARY MAR 1 8 19B5 nARVARD nRSlTY> B R E V I O K A IMuiseum of Comparative Zoology Cambridge. Mass. August 11, 1972 Number 392 A FOSSIL PELOMEDUSID TURTLE FROM PUERTO RICO Roger Coxaxt Wood* Abstract. A fossil turtle shell, too incompletely preserved to merit a formal binomial taxonomic designation, from the Oligocene San Sebastian Formation of Puerto Rico is here described. It represents the first record of a truly endemic pelomedusid in the Caribbean region. The specimen appears to be more closely related to South American pelomedusids than to any of the North American members of the family, which are known only as fossils. Presumably the Puerto Rican pelomedusid was derived either from South American ancestors or from some progenitor that gave rise to both it and the South American forms. Fossil turtles from the Caribbean region are few and frag- mentary. Our knowledge of these has most recently been reviewed by Williams ( 1 950 ) and then further extended in a subsequent publication ^Villiams, 1952). In his earlier paper, Williams p. 8) briefly mentioned a specimen from Puerto Rico ". . . represented by numerous fragments of plastron, pel- vis, and of the buttress region of the carapace . . ." which had been donated to the American Museum of Natural History in 1924 by Seiior. Narcisco Rabell Cabrero. \Villiams noted that "the exact locality and circumstances of the find are uncertain" and that it "was unfortunately so incomplete as to afford no generic characters." Nevertheless, an accompanying map (^\'il- liams, 1950, fig. 1), in which the specimen is identified simply as an emydine, indicates that it was found in the southeastern quadrant of the island. The fost of my knowledge. The external surface of the shell (plastron as well as carapace) is essentially smooth, without the concentric striations or vermic- ulated sculpture characteristic of some pelomedusids. Rough- ened areas, posterolateral to the last neural on the \-isceral surface, indicate where the pelvis was fused to the carapace. The axial buttresses appear to ha\-e been robust, but the extent to which the inguinal buttresses may ha\e been de\'eloped cannot be ascertained, owing to the incomplete preservation of the cara- pace. As Dreserved, the midline length of the plastron is 35.7 centi- meters. I estimate that its total length must ha\e been somewhat greater than 50 centimeters. The posterior plastral lobe was BREVIORA No. 392 Figure 1. External view of what remains of the carapace of the Puerto Rican pelomedusid (AMNH 1836) ; the anterior end is toward the top of the page. The positions on the visceral surface of the left axial buttress as well as of the iliac scars of the pelvis are indicated by broken lines. Those scute sulci that can still be discerned are represented by parallel dotted lines. Missing portions of the shell are indicated by shaded areas. 1972 FOSSIL PELOMEDUSID TURTLE Figure 2. External view of what remains of the plastron of the Puerto Rican pelomedusid (AMNH 1S36); the anterior end is toward the top of the page. The positions of the pelvic scars on the visceral surface are in- dicated by broken lines. The scute sulci and missing portions of the shell are depicted as in Figure L 6 BREVIORA No. 392 evidently slightly wider than the anterior one; its width to the midline at the inguinal notch is 13.7 centimeters, whereas the corresponding width from the axial notch to the midline is approximately 12 centimeters. The sides of the posterior lobe are straight rather than curved and converge toward the rear. The anal notch is broad. The external surface of the posterior lobe is flat rather than depressed inward, which suggests (but does not conclusively demonstrate) that the specimen may have been a female. The bone forming the posterior lobe becomes gradually and uniformly thinner toward the rear. Pelvic scars are present on the visceral surface of the plastron. These are disposed in essentially the same manner as in any of the living pelomedusids. Only the left mesoplastron has been preserved; it is large and hexagonal in shape, but unlike most pelomedusids it is not more or less equidimensional. Instead, it is appreciably longer (9.9 cm) in its anteroposterior axis than in its tranverse one (8.1 cm). The pectoral-abdominal scute sulcus curves back- ward laterally to cross the front portion of the mesoplastron. Otherwise, there is nothing noteworthy about the remainder of the plastral scute sulci that can still be discerned. The pelvis is the best preserved part of the specimen. Never- theless, parts of the pubes and ischia are missing on both sides, although somewhat more remains of the right side than of the left ( Plate 1 ) . In its general appearance, the pelvis is similar to the structure typical of most living and fossil pelomedusids. There can be no question that the specimen here described represents a pelomedusid; the presence of mesoplastra together with a pelvis that is fused to the shell assures this determination. More precise placement of this Puerto Rican pelomedusid with- in the family, however, is difficult. In the New World, only two clearly definable pelomedusid genera — Bothremys (GafTney and Zangerl, 1968) and Taphrosphys (Hay, 1908) — have been reported outside the confines of South America', and these are known only as fossils. The Caribbean form does not belong to either of these genera. Bothremys differs from it in having mesoplastra that are broader than long (Schmidt, 1940: 7 and fig. 5; Zangerl, 1948: 38 and figs. 3, 15), as well as an anterior plastral lobe that is slightly broader than the posterior one at 'Two other North American fossil genera, Amhiypeza and Naladochclys, have been described on the basis of extremely fragmentary material and referred to the Pelomedusidae (Hay, 1908: 122, 125), but I regard both of these as nomina vana (Wood, MS). 1972 FOSSIL PELOMEDUSID TURTLE S u o o. • 01 Ui < -4-1 lU • • s ■^ 4^ X C 3 i> T3 u u E 4> > _o « u ex o c rs ^ o hb cS o cs ;- OJ 3 X CL, c V 0^ X c •-^ 1> o u X V > rt X u ^ <+-( o V 0) JO X '> ,J_, ^^ J= ea _M •5 *u u 4J E o H V C , es '-" U « (li • *• % u 8 BREVIORA No. 392 its maximum width (Zangerl, 1948: 38). Taphrosphys differs principally in the shape and position of the pelvic scars on the internal surface of the plastron (,Hay, 1908, fig. 118) in addition to having its anterior plastral lobe broader than the posterior one." Representatives of three pelomedusid genera are known in the South American paleontological record." Two of these — Apodichelys (Price, 1954a) and Roxochelys (Price, 1954b) — are known only as fossils from Brazil, while the third - — - Podoc- neniis — still has living species in South America as well as in Madagascar. Apodichelys is based on a single specimen of very small size, the internal mold of a shell. Its anterior plastral lobe was considerably broader than the posterior one, and for this reason it is unlikely to be related to the Caribbean form. The type and only described specimen of Roxochelys consists of a small portion from the front of a carapace and an anterior plas- tral lobe. Comparable parts of the Caribbean specimen ha\'e not been preserved, so that possible relationships between the Brazilian and Puerto Rican forms cannot yet be evaluated. For- tunately, several nearly complete shells of Roxochelys have recently been discovered, but until they are described meaning- ful comparisons cannot be undertaken. Wood and Gamero (1971) recently have summarized our knowledge of the South American fossil forms that have been referred to Podocnemis, while Siebenrock (1902) and Williams (1954) have reviewed the living species of the genus. Except for the proportions of its mesoplastron and the outlines of the last two vertebral scutes, the Caribbean specimen falls within the limits of morphological variability known to occur in Podocnemis. From the foregoing discussion, it seems reasonable to state that the Caribbean specimen is probably more closely related to South x'Xmerican pelomedusids than to those from North Ameri- ca. But difficulties arise in trying to determine the most appropriate way to express this relationship. The shape of the "This information is based on a recently discovered specimen, #18706 in the collections of Princeton University, that is much more complete than any that were available to Hay. Dr. E. S. Gaflney has studied but not yet described this specimen, and I appreciate his courtesy in permitting me to use the information mentioned here. ^Staesche (1929) has referred some fragments from the late Cretaceous or early Tertiary of Argentina to a fourth genus, Na'iadochelys, but such an attribution is highly doubtful in view of the fact that the type material from North America cannot be adequately characterized at present. 1972 FOSSIL PELOMEDUSID TURTLE 9 mesoplastron and the \ertebral scute pattern might be considered sufficiently distincti\e for the recognition of either a new species of Podocnemis or perhaps even of a new genus. Numerous fossil pelomedusid taxa have been named and described in the past on the basis of an equally small list of purportedly diagnostic features. In view of the fact that the shell is so incompletelv known, however, I prefer to refrain from giN'ing it any kind of formal taxonomic designation at either the generic or the specific level, for at least three possibilities exist: the shell may belong to either Podocneynis. Roxochelys, or a third previously unrecog- nized genus. At present I do not beliexe that currently a\ailable e\idence is adequate to discriminate among these choices. Consequently I will merely designate the Puerto Rican specimen as Pelomedusidae gen. et sp. indet. Outside of South America, only one other fossil pelomedusid of Tertiary age has previously been described in the New World. This is the specimen, consisting solely of the anterior half of the plastron, that Collins and Lynn (1936) originally described as Taphrosph\s iniocenica and that GafTney and Zangerl (1968) later (and correctly, I believe) referred to Bothremys. This specimen, now lost (GafTney and Zangerl, 1968: 209), is of Miocene age. Collins and Lynn (1936: 162) assumed that B. miocenica must have been a fresh water form because all living pelomedusids occur in fresh water habitats. They further as- sumed that their specimen had been carried out to sea before being covered by sediment. In \iew of the fact that many pelo- medusids in the past (including, I suspect, all other previously described North American fossil pelomedusids) were evidently adapted to marine conditions (Wood, MS), as well as that the bulk of fossils associated with B. yyiiocenica include strictly oceanic cheloniid turtles, cetaceans, and marine inxertebrates, it seems reasonable to postulate that this species was probably a marine form that was not transported any particularh great dis- tance before burial. Circumstantial evidence indicates that the Puerto Rican pelo- medusid w'as probably also a marine form. As already mentioned, the fauna that was reportedly discovered in association with it cannot now be located, thus eliminating a potentially important source of environmental information. Fortunately, however, one other fossil vertebrate previously has been described from the San Sebastian Formation and this species — the dugongid sirenian Caribosiren turneri (Reinhart, 1959) — may be helpful in determining the depositional environment in which the remains 10 BREVIORA No. 392 of the Puerto Rican pelomedusid were buried. Caribosiren was found in a limestone outcrop along the road between the towns of San Sebastian and Lares (Reinhart, 1959: 8); consequently, it seems probable that this species was recovered from the same general area as the locality where Senor Rabell Cabrero discov- ered his fossil turtle. The close proximity, if not coincidence, of the localities where these fossil vertebrates were found strongly suggests that they lived in the same aquatic habitat. The lime- stones from which Caribosiren was disinterred contained abundant foraminifera and molluscs and these indicate a shallow- water marine habitat (Reinhart, 1959: 16-17). Living dugongids, furthermore, are confined to marine waters (Walker, 1968: 1332). Therefore, it seems reasonable to suppose thai Caribosiren was also fully adapted to a marine existence. All these considerations lead me to believe, by inference, that the Puerto Rican pelomedusid was in all probablity a truly marine turtle. Other pelomedusids, although not fossil ones, previously have been reported from the Caribbean region. Palacky (1897: 14) cursorily mentioned the occurrence of Podocnemis durneriliana on Guadeloupe without specifying any further details. The va- lidity of this record is dubious, however, for, as Barbour ( 1934: 111) has noted: ". . . apparently material from a host of locali- ties around the Caribbean basin, at one time or another, was assembled at one of these French islands [Guadeloupe or Marti- nique] before being shipped to Paris, where the material now bears a locahty label indicating the point of shipment only." Barbour further quoted a publication dating from 1862 which stated that tortoises " '. . . are brought for sale to the islands from the coast (Venezuela) . . .' " This reference specifically singles out an area where P. durneriliana naturally occurs. Either one of these explanations could account for the reported presence of this species on Guadeloupe. In view of the fact that no one has subsequently found P. durneriliana on this island, it seems reasonable to regard Palacky's comment on its distribution as erroneous. A different pelomedusid genus does, however, actually occur on Guadeloupe. Pinchon (1967: 561) has recorded its presence there: "C'est une Tortue originaire d'Afrique tropicale, Pelusios subniger; elle fut certainement introduite par I'homme mais on ignore absolument quand et comment." According to Pinchon (personal communication), representatives of this species can be found only on the eastern, nonmountainous half of the island 1972 FOSSIL PELOMEDUSID TURTLE 11 ( Grand Terre ) , particularly in ponds in the vicinity of the town of Moule, and even here they are not particularly abundant. x\lthough no examples of Pelusios from Guadeloupe have yet been carefully examined, there are no readily apparent differ- ences serving to distinguish this population from African ones of the same species (which is, it may be worth noting, widespread in sub-Saharan Africa). It is for this reason, in addition to the fact that Pelusios does not occur in the fossil record except in Africa, that it seems reasonable to suppose that the Guadeloupe population has been introduced to the island within historic times. Of the two reported occurrences of living pelomedusids in the Caribbean region, therefore, one is erroneous and the other represents what is almost certainly a relatively recently introduced population that has achieved limited success at best. Conse- quently, the Puerto Rican specimen is the first record of a truly endemic pelomedusid in the Caribbean. On the basis of morpho- logical similarities, furthermore, it seems likely that the Puerto Rican form was derived either from South American ancestors or from some progenitor that gave rise to both it and at least some of the South American forms. ACKNOWLEDGMENTS I am indebted to Dr. E. S. GafTney of the American Museum of Natural History for his kindness in permitting me to study the fossil turtle described in this paper. Miss Charlotte Holton, also of the American Museum, was of great assistance in my efforts to locate correspondence relating to the specimen. Shelter from the hazards of the New York subway system was amicably pro- vided by Mr. and Mrs. R. H. Wadsworth during several visits to the American Museum for research on the specimen here described. Dr. Frank Whitmore, vertebrate paleontologist for the United States Geological Survey at Washington, D. C, was most helpful in directing my attention to Reinhart's publication on Caribosiren. Mr. and Mrs. Alan Patterson kindly interviewed the son of Senor Rabell Cabrero on my behalf during their visit to San Sebastian in late December, 1971; the information they were able to uncox^er proxed to be of great interest. I am in addition much obliged to Pere R. Pinchon of Fort de France, Martinique, and Dr. P. C. H. Pritchard, of the University of Florida, for information concerning the occurrence of Pelusios subniger on Guadeloupe. Prof. E. E. \\'illiams of Harvard Uni- 12 BREVIORA No. 392 versity first prompted my interest in the Puerto Rican fossil turtle and then critically read the resulting manuscript. Miss Victoria Kohler and Mr. Don Crowe assisted in the preparation of the photographic plate. Finally, I would lilce to express my gratitude to the National Geographic Society for its generous support of my research on pelomedusid turtles. LITERATURE CITED Barbour, T. 1934. Observations on Antillean tortoises and terrapins. Co- peia, (3) : 111-113. Collins, R. L., and W. G. Lynn. 1936. Fossil turtles from Maryland. Proc. Amer. Phil. Soc, 76(2): 151-173. Gaffney, E. S., and R. Zangerl. 1968. A revision of the chelonian genus Bothremys (Pleurodira: Pelomedusidae). Fieldiana, Geology, 16(7): 193-239. Hay, O. p. 1908. The fossil turtles of North America. Carnegie Inst., Publ. no. 75, 568 pp. Khudoley, K. M., and a. A. Meyerhoff. 1971. Paleogeography and geo- logical history of Greater Antilles. Geol. Soc. America, Memoir 129: 1-199. Palacky, J. 1897. La distribution geographique des cheloniens. Bull. In- ternational, Acad, des Sciences de I'Empereur Frangois Joseph I (Ceska Akademie Cisare Frantiska Josefa I, Prague), 4: 9-14. PiNCHON, R. 1967. Quelques aspects de la nature aux Antilles. Fort de France, Martinique. Printed for private circulation by the author. 254 pp. Price, L. I. 1954a. Um quelonio pleurodiro no calcario da serie Apodi, Cretaceo do estado do Rio Grande do Norte. Notas Preliminares e Estudias, Divisao de Geologia e Mineralogia, nr. 85: 1-12. . 1954b. Os quelonios da formagao Bauru, Cretaceo terrestre do Brasil meridional. Divisao de Geologia e Mineralogia (Brasil), Bol. 147: 1-34. Reinhart, R. H. 1959. A review of the Sirenia and Desmostylia. Univ. Calif. Publ. Geol. Sciences, 36(1): 1-146. Schmidt, K. P. 1940. A new turtle of the genus Podocnemis from the Cretaceous of Alabama. Geol. Ser., Field Mus. Nat. Hist., 8(1): 1-12. Schuchert, C. 1935. Historical Geology of the Antillean-Caribbean Region. New York: John Wiley and Son, Inc. xxvi and 811 pp. Siebenrock, F. 1902. Zur Systematik der Schildkroten-Gattung Podocnemis Wagl. Sitz. Kaiserl. Akad. Wiss. Wien (Math.-Naturw. Classe), 111(1): 1-14. Staesche, K. 1929. Schildkrotenreste aus der Oberen Kreide Patagoniens. Palaeontographica, 72: 103-123. Walker, E. P. 1968. Mammals of the World (2nd edition), vol. 2. Balti- more: The Johns Hopkins Press, pp. 647-1500. 1972 FOSSIL PELOMEDUSID TURTLE 13 Weyl, R. 1966. Geologic der Antillen. Beitriige zur Regionalen Geologic der Erde, vol. 4. Berlin-Nikolassee: Gebriider Borntraeger. vii and 410 pp. Williams, E. E. 1950. Trs/uJo cubnisis and the evolution of western hemisphere tortoises. Bull. Amer. Miis. Nat. Hist., 95 (art. 1): 1-36. . 1952. A new fossil tortoise from Mona Island, West Indies, and a tentative arrangement of the tortoises of the world. Bull. Amer. Mus. Nat. Hist., 99(art. 9): 541-560. 1954. A key and description of the living species of the genus Podocnemis [scnsu Boulenger) (Testudines, Pelomedusidae). Bull. Mus. Comp. Zool., 111(8): 279-295. Wood, R. C, .\nd M. L. Diaz de Gamero. 1971. Podocnemis vcnezuelen- sis, a new fossil pelomedusid (Testudines, Pleurodira) from the Plio- cene of Venezuela and a review of the history of Podocnemis in South America. Breviora, No. 376: 1-23. Zangerl, R. 1948. The vertebrate fauna of the Selma Formation of Ala- bama. Part II. The pleurodiran turtles. Fieldiana: Geol. Mem., 3(2): 19-56. in u.' •i^. v^-v-' ■ Linr.ARY MAR 1 8 1985 BREVIORA Miuiseeni of Comparative Zoology Cambridoe, Mass. August 11, 1972 Number 393 MINUTE JAMAICAN PROSOBRANGH GASTROPODS: STOASTOMA AND ITS GONGENERS Kenneth J. Boss Abstract. Several generic level taxa have been allied to the minute terrestrial helicinid Stoastoma (Mollusca; Gastropoda; Prosobranchia). The type-species of each taxon is illustrated for the first time and the status of each is discussed. The nominal congeners are four recognizable genera that are virtually restricted to Jamaica: fVilkinsoriura, Leii'is'ia, Stoastoma with its synonym Hcmicyclostoma, and Fadycnia with its synonyms Metcalfcia, Pctitia, Lindslcya, and Blaiidia. INTRODUCTION Among the terrestrial prosobranch gastropods, the family Helicinidae is exceptionally rich in species and is developed in two primary regions of the world, southeastern Asia and the West Indies. In the latter are several radiations that have been accorded subfamilial and generic rank (Keen, 1960). One genus, Stoasto7na C. B. Adams, consists of species with minute shells, usually in the size range of 1-3 mm in diameter. Of the New World helicinids, it has been almost totally neglected, be- cause of nomenclatorial problems and the tiny size of the shells. This paper reviews the generic units of Stoastoma and provides the first illustrations of their type-species. In 1849 C. B. Adams described the genus Stoastoma. Subse- quent to Adams' death in 1853, Edward Chitty, an amateur conchologist, English lawyer, and judge who lived in Jamaica, took up the task of describing land snails from this Antillean island. In 1857, after his return to England, Chitty published an overzealous paper on Stoastoma in which he recognized the group as a separate family, the Stoastomidae, and proposed seven new genera, sixty-one new species, and two new varieties from Jamaica. Although these descriptions pay considerable attention to conchological detail, none of the species was illus- 2 , BREVIORA No. 393 trated and there is some confusion as to the type-species of the subgenera. Clench and Turner (1950) provided a list of Chitty's taxa but did not figure any Stoastoma. Baker (1922, 1934) and Wenz (1938) treated some aspects of the taxonomy of Stoastoma, while Keen (1960), though listing the taxa and synonymizing some of the subgenera, did not figure any type- species. Herein are illustrated the type-species of the generic taxa described by Adams and Chitty, along with corrected type designations. Type designations have been given bv Baker (1922), Wenz (1938), Clench and Turner [\9bQi), and Keen (1960) but most of Chitty's taxa, in accordance with the International Code of Zoological Nomenclature (Article 68 [C], Recom- mendation 69 B [2] ) , ha\e their type-species by virtual tau- tonymy. Pfeiffer (1858, 1865, 1876) also provided lists of species. C. B. Adams frequently did not give any specific type locality for his species; the localities provided herein are those of the lectotypes here designated. Principle citations are also pro\ided. ACKNOWLEDGMENTS Mr. John Peake of the British Museum (Natural History) kindly arranged for the loan of specimens from the Chitty Col- lection. The manuscript was critically read by Dr. R. D. Turner and Messrs. R. I. Johnson and M. K. Jacobson. LIST OF THE TAXA Blandia Chitty 1857, PZSL, pp. 167 & 169 [type-species, by virtual tau- tonymy, Stoastoma blandianum C. B. Adams 18+9, lectotype here selected and figured (Figs. 20-22), MCZ 177243; type-locality. Pease River, Man- chester, Jamaica]; Pfeiffer, 1865: 207; 1876: 237; Baker, 1922: 57; Pilsbry, 1930: 230; Baker, 1934: 14; Wenz, 1938: 446; Clench and Turner, 1950: 9; Keen, 1960: 288. Fadyenia Chitty 1857, PZSL, pp. 167 & 168 [type-species, by virtual tau- tonymy, Stoastoma fadyrnianum C. B. Adams 1849, lectotype here selected and figured (Figs. 23-25), MCZ 177233; type-locality, hills SW of Port Henderson, Jamaica] ; Pfeiffer, 1865: 190; 1876: 237; Baker, 1922: 57; Baker, 1934: 3; Wenz, 1938: 446; Clench and Turner, 1950: 9; Keen, 1960: 288. Hemicyclostoma Chitty 1857, PZSL, p. 163 {nomfn riudum) ; Pfeiffer, 1865: 191 [type-species, by original designation, Stoastoma pisum C. B. Adams 1849, lectotype here selected and figured (Figs. 1-3), MCZ 177234], non Grav 1840. 1972 STOASTOMA 3 Lridsia Chitty 1857, PZSL, pp. 167 i- 168 [type-species, by subsequent designation (Baivcr, 1922: 57), Stoastoma philippianum C. B. A(]ains 1851, lectotype here selected and figured (Figs. 10-12), MCZ 177241; type- locality. Burnt Hill near Ackendown, Westmoreland, Jamaica]; Pfeiffer, 1865: 185; 1876: 237; Baker, 1934: 14; Wenz, 1938: 447 (incorrectly designated Stoastoma ayassiziannm C. B. Adams 1849 as tyjie-species ) ; Clench and Turner, 1950: 10 (incorrectly designated Stoastoma agassizianum C. B. Adams 1849 as type-species); Keen, 1960: 228. Lindslfya Chitty 1857, PZSL, pp. 167 & 169 [type-species, by virtual tautonymy, Stoastoma lindslcyanum C. B. Adams 1849, lectotype here selected and figured (Figs. 4—6), MCZ 177238; type-locality, Manchester, Jamaica]; PfeiflFer, 1865: 200; 1876: 237; Baker, 1922: 57; Baker, 1934: 3; Wenz, 1938: 446; Clench and Turner, 1950: 10; Keen, 1960: 288. Metcalfeia Chitty 1857, PZSL, pp. 167 & 168 [type-species, Metcalfcia mrtcalfe'iana Chitty 1857, by virtual tautonymy, lectotype here selected and figured (Figs. 7-9), BMNH 57.12.1.218; type-locality, Hanover, Jamaica]; Pfeiffer, 1865: 192; 1876: 237; Baker, 1922: 57 (incorrectly designated Stoastoma ch'ittyanum C. B. Adams 1849 as type-species); Baker, 1934: 3; Wenz, 1938: 446; Clench and Turner, 1950: 10 (designated M. mrtcalf- e'iana Chitty 1857 as type-species); Keen, 1960: 288. Pctitia Chitty 1857, PZSL, pp. 167 & 168 [type-species, by virtual tau- tonymy, Stoastoma pctitianum C. B. Adams 1851, lectotype here selected and figured (Figs. 17-19), MCZ 177237; type-locality. Pease River, Man- chester, Jamaica]; Pfeiffer, 1865: 196; 1876: 237; Baker, 1922: 57: 1934: 14; Wenz, 1938: 446; Clench and Turner, 1950: 11; Keen, 1960: 288. Stoastoma C. B. Adams 1849 (September), Monograph of Stoastoma, a new genus of new operculated land shells, pp. 1—16, Amherst, Massachu- setts; 1849 (September), Contributions to Conchology, no. 1, p. 16 [type- species, by subsequent designation, Chitty 1857, PZSL, p. 167, Stoastoma pisum Adams 1849, lectotype here selected and figured (Figs. 1-3), MCZ 177234; type-locality, Manchester, Jamaica]; Pfeiffer, 1858: 170; 1865: 184; 1876: 237; Baker, 1922: 57; Wenz, 1938: 446; Keen, 1960: 287. tVilkinsonara Chitty 1857, PZSL, pp. 167 &: 168 [type-species, by virtual tautonymy, Stoastoma vilkittsoniae C. B. Adams 1851'; lectotype here se- lected and figured (Figs. 13-16), MCZ 177251; type-locality, Vallah's Hill, Jamaica]; Pfeiffer, 1865: 186; 1876: 237; Baker, 1922: 57 as Wilkmso7iia ivilkinsoriiar; 1934: 14; U'enz, 1938: 446 (incorrectly designated Stoastoma goiihi'iantim C. B. Adams 1849 as type-species) ; Clench and Turner, 1950: 12. 'The spelling of this specific name was altered to '■■ivilkirisonarana' by Chittv, 1857: 168, to 'ii-ilkinsoniae' by Baker, 1922: 57, and to '-vHko/iso/iiar' by Clench and Turner, 1950: 12. 4 BREVIORA No. 393 DISCUSSION Pilsbry and Brown (1910) noted the occurrence of Stoastoma pisum, a terrestrial form living on stones, in the vicinity of Mandeville, Jamaica. They figured the radula and stated that it confirmed the position of the genus in the family Helicinidae. That is, the radula of S. pisum exhibits the essential features of a helicinid radula (Troschel, 1857) : a single rachidian tooth, flanked by A, B and C centrals, a lateral complex con- sisting of a comb-lateral and an accessory plate, and a marginal complex consisting of numerous teeth or uncini. This gives the formula (MC) (LC) CBARABC (LC) (MC). As illustrated by Baker (1922), its specific features include an A central with a single large outer hook (resembling Lucidella) , a B central with three heavy lobes, and a C central with four. The comb lateral has three large cusps and the inner marginals (first 14) are unicuspid, whereas the other six are bicuspid. In his extensive study on the radula of the Helicinidae, Baker (1922) retained Stoastoma, provisionally placing most of the Chitty subgenera into its synonymy and claiming that it forms a connecting link between the subfamilies Vianinae and Heli- cininae. He pointed out further (1922: 58) that, "Since Chitty's description of the many species and 'genera,' the group seems to have received little attention." He also tentatively included Lindsleva in the synonymy of Lucidella because of a similarity in radular structure. Subsequently, Baker (1934) again dealt with some of Chitty's generic taxa. He separated Stoasto?na s. s. and considered Fadyenia at the generic level, allocating Lewisia, Wilkinsonaea, Blandia, Petitia, Metcalfeia and Lindsleya either into synonymy or to unclearly defined groups. For example, Lewisia was recog- nized as a subgenus of Fadyenia and Wilkinsonaea tentatively synonymized. Likewise, Blandia was assigned a subgeneric status with Petitia as a synonym. And lastly, Fadyenia s. s. included both Metcalfeia and Lindsleya. Keen ( 1960) also considered these taxa. Stoastoma s. s., with its exact synonym Hemicyclostoma, was referred to the nominate subfamily. Fadyenia, placed in the Ceratodiscinae, was divided into three subgenera: Fadyenia s. s. with Lindsleya and Metcal- feia as synonyms, Blandia with Petitia and Wilkinsonaea as synonyms, and Lewisia. The study of the type-species of Chitty's genera indicates that there should be some rearrangement of the taxa. Based on the 1972 STOASTOMA 5 structure and morphological features of the shell, four distinct groups can be recognized and these characteristics are such as to be accorded generic rank. As in other helicinids, the internal portion of the shell is absorbed in Sloasloma, a feature long ago pointed out by Bland (1858j. The shells of most of the species described by Chitty and Adams are uniqueh tiny and rather strongly sculptured. Accentuated surface features are apparently rather rare in the Helicinidae, having been noted in some E?noda-\ike forms which are accorded generic rank by Clench and Jacobson (1971). Among all the groups, Stoastoma s. s. is distinct in the size of the shell alone, being two or three times larger than any of the forms described by Chitty. Additionally there is a spur-like flaring of the base of the aperture (Fig. 3). The shell is heavy, strong, and globose. All of Chitty"s other generic taxa differ in being smaller and more delicate. Lewisia vFigs- 10-12) is distinguished by its lack of sculpture. Its smooth discoid or lens-like shell is distincti\e. Additionally it is imperforate with the umbilicus sealed and with a spiral ridge extending to the base of the aperture. Chitty (1857) characterized Lewisia as being "Quasi double-mouthed." I imagine that he was referring specifically to a teratological specimen of Lewisia agassiziana C. B. Adams. The type-lot of this species ( BMNH 54.4.24.473-474) contains tw^o specimens, one of which has an unusual sac-like growth over the umbilical area. A similar structure also occurs in the unique (single) specimen of L. macandrewiana Chitt\, according to his de- scription. The dominant features of the type-species of Lewisia, i.e., its smooth, imperforate shell, are apparently not shared by the species that Chitty included in Lewisia, namely agassiziana C. B. Adams, woodwardiana Chitty and macandrewiana Chitty. These nominate species are sculptured; Lewisia may then be monotypic. Likewise, WUkinsonaea is most unusual ''Boss, 1972). Of a depressed sub-discoidal shape, it possesses markedly strong spiral sculpturing or carinae in the form of ridges which terminate as bulbous, digitiform extensions of the aperture. With an open umbilicus, it preserves its unique nature by possessing a \-ery unusual operculum, which is deeply conca\e, finely punctate and auriculate at both extremities. Also the last whorl is extra- ordinarily produced and detached from the body whorl. All of Chitty's other generic taxa are synonymous and can be placed in Fadyenia, since it has page priority and Baker (1934) 6 BREVIORA No. 393 utilized it preferentially. The shells vary in size from the very small subdiscoidal Stoastoma blandianum with only 3/2 whorls to the larger globose S. lindsleyanum with 5^2 whorls. How- ever, conchological features indicating a basic similarity among all the type-species include: a \ariably developed, but distinct sculpture consisting of nvmierous spiral striations; a detached portion of the last whorl; an irregularly semicircular aperture with the central margin smooth and more or less straight and the external margin frequently formed into a recurv^ed lip and variously scalloped by the terminations of the spiral carinae; a relatively deep, open umbilicus. The taxa synonymized under Fadyenia include Metcaljeia, Petitia, Lindsleya, and Blandia. In addition to the species of Stoastoma, Wilkinsonaea, Lewisia, and Fadyenia described by Chitty and C. B. Adams from Jamaica, several other West Indian forms have been ascribed to Stoastoma. Pfeiffer (1857: 51) described S. portoricense from the island of that name collected by Hjalmarson on a plantation near Manati. Crosse (1892: 46) remarked how this form links Puerto Rico and Jamaica zoogeographically. Although Dall and Simpson (1901: 447) cited its occurrence, van de^r Schalie (1948: 22-23, pi. 1, fig. 7) was the first to figure the species and provide a list of several localities. Ap- parently the species is not a marked calciphile, occurs at eleva- tions to 3000 feet, and is predominantly distributed in the northern and western parts of the island. Although nominally placed in Fadyenia by van der Schalie, the shell is distinctly more highly spired than Jamaican species. Its small size, somewhat detached outer whorl, and umbilication show some affinity with Jamaican Fadyenia. From Haiti, Weinland (1862) described Stoastoma haitianum, which was collected some distance from Jeremie in the south- west of the island; he stated that the species was related to S. philippianiwi C. B. Adams. Pfeiffer (1862), in his com- ments on Weinland's paper, remarked that this was an im- portant find, considering the habitat. He related the species to Metcaljeia, particularly S. chittyanum C. B. Adams. Crosse (1891) also noted that with the occurrence of a species of Stoastoma on Haiti, the islands of Puerto Rico and Haiti were zoogeographically related to Jamaica. A second species from Hispaniola was described as S. domingensis by Vanatta (1920: 206, pi. 6, figs. 4, 5) from one mile NW of Sanchez, Santo Domingo. Although he did not mention Weinland's species, he compared the form with S. leanum C. B. Adams from 1972 STOASTOMA 7 Jamaica, a species that Chitty (1857) had placed in Lindsleya. S. doyningensis is very similar to S. portoricense and can be referred to Fadyenia on the same basis and with the same reser- vation that portoricense is. An exceptionally tiny shell (0.7 mm in height) was described by Pilsbry as S. atomus (1930: 230, pi. 15, figs. 4, 4a, 4b). It was taken near George Town, Grand Cayman Island. Pilsbry placed the species in Blandia and stated that it was comparable to S. blandianum C. B. Adams. Indeed, a comparison of his figures with those of S. blandianum (Figs. 20-22) shows that the species are very similar, though S. atomus is decidedly smaller. Pilsbry also commented that though Stoastoma and its relatives belonged to the Hclicinidae on the basis of the radula, they were unique in their "peculiar gait." In his list of Cayman Island land mollusks. Clench (1964) also included S. atomus. Lastly Baker (1922) described a Lucidella venezuelensis from Palma Sola, Estado Falcon, \^enezuela, which he referred to Fadyenia [in litt. to M. K. Jacobson, 17 Nov. 1949). Pro- viding a figure of the shell and relating the species to S. dom- ingensis, he recounted the species more thoroughly in 1923. The three species, venezuelensis, domingensis, and portoricense are all quite similar to one another. Though tentatively placed in Fadyenia, they are distinguished from the Jamaican forms by being distinctly higher or more conic in shape, with a char- acteristically detached and somewhat depressed outer whorl. In contrast, Pilsbry's species from the Caymans is virtually con- specific with blandianum. In summary, of the generic le\el taxa attributed to the so- called Stoastomidae by Chitty (1857), four genera, which are xirtualh' restricted to Jamaica, are recognizable and retained in the Helicinidae: Wilkinsonaea, Lewisia, Stoastoma with its synonym Hemicyclostoyyia, and Fadyenia with its synonyms Metcaljeia, Petitia, Lindsleya and Blandia. 8 BREVIORA No. 393 LITERATURE CITED Adam3, C. B. 1849. Monograph of Stoastoma, a new genus of new opercu- lated shells. Privately printed. Amherst, Massachusetts. 16 pp. Baker, H. B. 1922. Notes on the radula of the Helicinidae. Proc. Acad. Nat. Sci. Philadelphia, 74: 29-67, pis. 3-7. 1923. The Mollusca collected by the University of Michi- gan-Williamson expedition in Venezuela. Occ. Papers Mus. Zool., Univ. Mich., No. 137. 58 pp., 5 pis. 1934. Jamaican land snails. Nautilus, 48: 1-14, 60-67. Bland, T. 1858. On the absorption of parts of the internal structure of their shells by the animals of Stoastoma, LucidcUa, Trochatella, Hclicina and Proserpina. Ann. Lyceum Nat. Hist. New York, 6: 75-77. Boss, K. J. 1972. Chitty Chitty Bang Bang. Bull. Amer. Malac. Union for 1971, p. 11, pi. 2. CHiTTi', E. 1857. On Stoastomidae as a family and on seven proposed new genera, sixty-one new species, and two new varieties from Jamaica. Proc. Zool. Soc. London, pp. 162—200. Clench, W. J. 1964. Land and freshwater Mollusca of the Cayman Islands, West Indies. Occ. Papers Moll., Harvard Univ., 2(31): 345- 380, pis. 61-63. , AND M. K. Jacorson. 1971. Monograph of the Cuban genera Emoda and Glyptcmoda (Mollusca: Archaeogastropoda : Heli- cinidae). Bull. Mus. Comp. Zool., 141(3): 99-130, 7 pis. , AND R. D. Turner. 1950. Edward Chitty, with a bibliog- raphy and a catalogue of his species of Jamaica land moUusks. Occ. Papers Mus. Inst. Jamaica, Kingston, No. 1. 12 pp., 1 pi. Crosse, H. 1891. Faune malacologique terrestre et fluviatile d'ile de Saint- Domingue. Jour. Conch., 39: 69—211. 1892. Faune malacologique terrestre et fluviatile d'ile de Portorico. Jour. Conch., 49: 5—71. Dall, W. H., AND C. T. Simpson. 1901. The Mollusca of Porto Rico. U. S. Fish. Comm., Bull., 20(1): 351-524. Keen, A. M. 1960. In Moore, R. C. (ed.). Treatise on Invertebrate Pale- ontology. Mollusca, Helicinidae, 1: 285-288, figs. 186-187. University of Kansas. Lawrence, Kansas. Pfeiffer, L. 1852. Monographia Pneumonopomorum Viventium. Cassell. 439 pp. 43-52. 1857. Diagnosen neuer Landschnecken. Malak. Blat., 3: 1858. Monographia Pneumonopomorum Viventium. Supple- mentum primum. Cassell. 249 pp. 1862. Bemerkungen zu den beschriebenen Arten. Malak. Blat, 9: 199-202. 1865. Monographia Pneumonopomorum Viventium. Supple- mentum secundum. Cassell. 284 pp. 1876. Monographia Pneumonopomorum Viventium. Supple- mentum tertlum. Cassell. 479 pp. 1972 STOASTOMA 9 PiLSBRY, H. A. 1930. Results of the Pinchot South Sea Expedition. 1. Land moliusks of the Caribbean Islands, Grand Cayman, Swan, Old Providence and St. Andrew. Proc. Acad. Nat. Sci. Philadelphia, 82: 221-261, pis. 15-19. ScH.ALiE, H. VAN DER. 1948. The land and fresh-water moliusks of Puerto Rico. Misc. Publ., Mus. Zool., Univ. Michigan, No. 70. 134 pp., 14 pis. Troschel, F. H. 1857. Das Gebiss der Schnecken. Berlin. Vol. 1: 80-91, figs. 57—67 (Helicinidae). Vanatta, E. G. 1920. New land shells. Proc. Acad. Nat. Sci. Philadelphia, 72: 203-206, pi. 6. Weinland, D. F. 1862. Diagnosen einiger neuen westindischen Land- schnecken. Malak. Bliit., 9: 194-199. Wenz, W. 1938. Gastropoda, hi Handbuch der Palaozoologie. Helicini- dae, 6: pt. 1, pp. 435-448, figs. 1071-1118. Berlin. 10 BREVIORA No. 393 Plate I. Figures 1-3. The type-species of Stoastoma s. s. Stoastoma pisum C. B. Adams 1849, Monograph of Stoastoma, p. 11, lectotype, here selected, MCZ 177234, dia. = 3.7 mm, ht. = 3.3 mm; dorsal, apertural, and ventral views. Figures 4—6. The type-species of Lindsleya Chitty. Stoastoma lindsleyanum C. B. Adams 1849, Monograph of Stoastoma, p. 12, lectotype, here selected, MCZ 177238, dia. = 2.6 mm, ht. ^ 2.2 mm; dorsal, apertural, and ventral views. Figures 7—9. The type-species of Mctcalfeia Chitty. Metcalfeia metcal- feiana Chitty 1857, p. 179, lectotype, here selected BMNH 57.12.1.218, dia. = 2.6 mm, ht. =^ 2.2 mm; dorsal, apertural, and ventral views. 1972 STOASTOMA 11 12 BREVIORA No. 393 Plate II. Figures 10-12. The type-species of Lcivisin Chitty. Stoastoma ph'tlip- pianiim C. B. Adams 18 51, Contributions to Conchology, 9: 158, lectotype, here selected, MCZ 177241, dia. = 2.2 mm, ht. ^ 1.1 mm; dorsal, apertural, and ventral views. Figures 13-16. The type-species of Jf'ilkiiisoriara Chitty. Stoastoma vilk'insoniac C. B. Adams 1851, Contributions to Conchology, 8: 148, lecto- type, here selected, MCZ 177251 (lost after photographs made), about 1.8 mm in diameter and 0.9 mm in height; dorsal, two apertural (showing protrusion of digitate aperture), and ventral views. Figures 17-19. The type-species of Pctitia Chitty. Stoastoma pctitianum C. B. Adams 1851, Contributions to Conchology, 8: 151, lectotype, here selected, MCZ Ml 111 , dia. =: 1.7 mm, ht. =: 1.1 mm; dorsal, apertural, and ventral views. Figures 20-22. The type-species of Blandia Chitty. Stoastoma blandianum C. B. Adams 1849, Monograph of Stoastoma, p. 6, lectotype here selected, MCZ 177243, dia. = 1.3 mm, ht. = 0.7 mm; dorsal, apertural, and ventral views. Figures 23-25. The type-species of Fadyenia Chitty. Stoastoma fadycnla- num C. B. Adams 1849, Monograph of Stoastoma, p. 7, lectotype, here selected, MCZ 177233, dia. = 2.1 mm, ht. = 1.7 mm; dorsal, apertural, and ventral vaews. All specimens were coated with magnesium before being photographed. 1972 5TOASTO>rA 13 MAR 1 8 198^ B R E V I O R A.* Muiseom of Comparative Zoology Camrridge. Mass. August 11, 1972 Number 394 THE CHANARES (ARGENTINA) TRIASSIC REPTILE FAUNA. XV. FURTHER REMAINS OF THE THECODONTS LAGERPETON AND LAGOSUCHUS Alfred Sherwuud Romer Abstract. Further fragmentary postcranial remains in the Chanares material attributable to the pseudosuchians Lagcrpcton chanarcnsis and Lagosuchiis talampayensis are described and figured. A sepcimen including a peculiar pelvis, hind leg, and vertebrae is described as Lagosuchus liUocnsis sp. nov. In addition to material definitely attributable to the seven thecodonts previously described in this series of papers on the Chanares fauna, various nodules collected in 1964-65 include numerous fragmentary remains of thecodonts, mainh- of small size. I ha\e gone o\-er this series of specimens and attempted to identify such materials. Some apparently represent immature specimens of Gracllhuchus and Chanaresuchus; some remain problematical; still others seem certainly to pertain to the small thecodonts that I hax'e described as Lagerpeton chayia- rensis and Lagosuchus talampayensis (Romer, 1971). Lagerpeton chanarensis. I have pre\iously described the type specimen, consisting of the slender hind leg of a small pseudo- suchian in which the elongate pes differs from the typical pseudosuchian pattern in that digit 4 is longer than digit 3, and digit 1 is unusuallv short; I also mentioned the presence in the Instituto Lillo collections of an almost identical specimen. I may note that in the type the lengths of femur, tibia, and metapodials 1-4 are, in order: 77, 92, 8, 24, 45, and 48 mm. Further materials apparently belonging to Lagerpeton are present in se\eral Chaiiares nodules. In one nodule \vhich, in addition to much of a gomphodont skeleton, included the t\ pe of Lewisuchus admix tus f Romer, 1972) and a few bones of Lago- suchus, there are materials that appear to pertain to Lagerpeton BREVIORA No. 394 Figure 1. Laacrpcxon chanarcnsis. Material in nodule with the Lcwi- suchus type. A, Left femur in dorsal view, head restored from right femur; B, the same in ventral view; C, fibula; D, partial fibula; E, partial tibia; F, right ilium; G, dorsal vertebrae; H, presumed sacrals, in ventral view. XI. (MCZ 4121), but represent an individual somewhat smaller than the two pre\'iously described ( Fig. 1 ) . Lying close together and presumably associated, they include two femora, with lengths of 58 and 62 mm, a fibula 84 mm long, proximal parts of a tibia and of a second fibula, a right ilium and part of the left, and a number of vertebrae. The limb elements compare well in structure with those of the type. The femur has a strongly developed sigmoid cur\'ature, the head is well set off from the shaft, and the longitudinal ridge representing the fourth tro- chanter is well developed. The fibula is expanded distallv for articulation with the tarsals. The acetabular cavity of the ilium, deeply incised, extends unusually far posterodorsally, and the in- ner surface of the bone is deeply grooxed for the sacral ribs. There 1972 FURTHER THECOnOXT REMAIN'S Figure 2. Left scapulocoracoid of Laijosuchus lalampnyoisls. X +/3- are several short series of \ertebrae, rather imperfecth' preserved. I figure se\eral elements of a series of 1 2 dorsals, with mean central lengths of 7.5 mm. Near the right ilium are three \-erte- brae seen in \-entral \ie\v, with central lengths of 6.5 mm; these bear short stout ribs, and are apparenth sacrals. In the nodule, but somewhat separated, are remains of two scapulocoracoids, the better one of which is shown in Figure 2. This measures 42 mm in height. The scapular blade is relati\ely tall and narrow at the base. There is a modestly developed acromial ridge. Scapula and coracoid are firmly united, as preser\-ed. The coracoid is deep dorso\entrally and narrow anteroposteriorly. The glenoid appears to face more directly backward than in most pseudosuchians. MCZ 3691 is a slab mainlv showing the confused remains of two partially articulated gomphodont skeletons. Present, how- e\-er, are two pairs of slender tibiae and fibulae; none are com- pletely preser\-ed and exoosed, but one tibia has, as far as \-isible, a length of 80 mm. These elements quite surely belong to Lagerpeton. Lagosuchus talajnpayensis. The type remains of this form (Fig. 3) were contained on a slab (La Plata Museum 64-XI- 14-11) that also includes the holotype of Gracilisuchus stipani- cicorum fRomer, 1972). Present on the slab are incomplete remains of both hind legs and an incomplete humerus with articulated radius and ulna. Lengths of femur, tibia, and 4 BREVIORA No. 394 A <. Hloeiii- fontein, 1 : 141-148. 1959. A new small thecodont from the red beds of the Stormberg series. Palaeont. Africa, 6: 109—115. Broili, F., and J. Schroder. 1934. Beobachtungen an Wirbeltieren der Karrooformation. \'. Ueber Chasmatosaurus van hocpcni Haughton. Sitzungsber. Bayer. Akad. Wiss., 1934, No. 3: 225-264. Broo.vi, R. 1903. On a new reptile ( Protcrosuchus fcrgusi) from the Karoo beds of Tarkastad, South Africa. Ann. S. Afr. Mus., 4: 159- 163. 1904. On a new crocodilian genus (Notochampsa) from the Upper Stormberg beds of South Africa. Geol. Mag., ser. 5, 1 : 582- 584. 1906. On the remains of Erylhrosuclnts afrlcanus Broom. Ann. S. Afr. Mus., 5: 187-196. 1913. On the South African pseudosuchian Euparkcria and allied genera. Proc. Zool. Soc. London, 1913: 613—633. 1927. On Splicnnsuchus and the origin of the crocodiles. Proc. Zool. Soc. London, 1927: 359-370. 1932. On some South African pseudosuchians. Ann. Natal Mus., 7: 5S-S9. 1946. A new primitive proterosuchid reptile. Ann. Trans- vaal Mus., 20: 343-346. Camp, C. L. 1930. A study of the phytosaurs, with description of new material from North America. Mem. LTniv. California, 10: 1-174. Casamiquel.a, R. M. 1960. Noticia preliminar sobre dos nuevos estagono- lepoideos Argentines. Ameghiniana, 2: 3-9. 1961. Dos nuev^os estagonolepoideos Argentines (de Ischigualasto, San Juan). Rev. Asoc. Geol. Argentina, 16: 143-203. 1967. Materiales adicionales y reinterpretation de Actosauroidcs scagliai. Rev. Mus. La Plata, New Ser., Paleont., 5: 173-196. Charig, a. J. 1957. New Triassic archosaurs from Tanganyika including Mandasuchus and Tcleocratcr. Abstr. Diss. Univ. Cambridge, 1955- 56: 28-29. 1966. Stance and gait in the archosaur reptiles. Brit. Assoc. Adv. Sci., 22(103): 537. 1972. The evolution of the archosaur pelvis and hind- limb: an explanation in functional terms. In Joysey. K. A., and T. S. Kemp, (Eds.), Studies in Vertebrate Evolution (essays pre- sented to Dr. F. R. Parrington, F.R.S.). Edinburgh: Oliver and Boyd, pp. 121-155. J. Attridge, and a. W. Crompton. 1965. On the origin of the sauropods and the classification of the Saurischia. Proc. Linn. Soc. London, 176: 197-221. Colbert, E. H. 1947. Studies of the phytosaurs Machacroprosopus and Rutiodon. Bull. Amer. Mus. Nat. Hist., 88: 53-96. 20 BREvioRA No. 395 1952. A pseudosuchian reptile from Arizona. Bull. Amer. Mus. Nat. Hist, 99: 561-592. , AND C. C. MooK. 1951. The ancestral crocodilian Pro- tosiichus. Bull. Amer. Mus. Nat. Hist, 97: 143-182. CRUICKSH.4NK, A. R. I. 1972. The proterosuchian thecodonts. In Joysey, K. A., and T. S. Kemp (Eds.), Studies in Vertebrate Evolution (es- says presented to Dr. F. R. Parrington, F. R. S.). Edinburgh: Oliver and Boyd, pp. 89-119. Ewer, R. F. 1965. The anatomy of the thecodont reptile Euparkeria capensis Broom. Phil. Trans. Roy. Soc. London, B 248: 379-435. Fraas, O. 1867. Dyoplax arenacciis, ein neuer Stuttgarter Keuper-Sau- rier. Jahresh. Ver. Naturk. Wiirttemberg, 23: 109-112. 1877. Aetosaurus fcrratus Fr. Die gepanzerte Vogel-Echse aus dem Stubensandstein bei Stuttgart. Jahresh. Ver. Naturk. Wiirt- temberg, 33: 1-22. Gregory, J. T. 1953. Typoiliorax and Dcsmalosuchus. Postilla, Yale Peabody Mus., No. 16: 1-27. 1962. The genera of phytosaurs. Amer. Jour. Sci., 260: 652-690. , AND F. Westphal. 1969. Remarks on the phytosaur gen- era of the European Trias. Jour. Paleont., 43 : 1296—1298. GUTH, C. 1963. Au sujet de restes de reptiles de Madagascar. C. R. Acad. Sci. Paris, 256: 2661-2663. Harland, W. B. ct al. . 1967. The Fossil Record. A Symposium with Documentation. Geol. Soc. London. 827 pp. Haughton, S. H. 1915. A new type of thecodont from the Stormberg beds {Sphc7iosuchus acutus, gen. et sp. nov.). Ann S. Afr. Mus., 12: 98-106. 1924a. On a new type of thecodont from the Mid- dle Beaufort beds. Ann. Transvaal Mus., 11: 93-97. -_ 1924b. The fauna and stratigraphy of the Stormberg Series. Ann. S. Afr. Mus., 12: 323-497. Hoffman, A. C. 1965. On the discovery of a new thecodont from the Middle Beaufort beds. Navors. Nas. Mus. Bloemfontein, 2: 33-40. HoFFSTETTER, R. 195 5. Thecodontia. In Traite de Paleontologie, 5: 665- 694. HuENE, F. V. 1911. Ueber Erythrosuclnis, Vertreter der neuen Reptil- Ordnung Pelycosimia. Geol. Palaeont., Abhandl., New Ser., 10: 1- 60. 1914a. Beitrage zur Geschichte der Archosaurier. Geol. Palaeont Abhandl., New Ser., 13: 1-53. 1914b. Coelurosaurier-Reste aus dem unteren Muschel- kalk. Centralbl. Min. Geol. Pal., 1914: 670-672. 1914c. Neue Beschreibung von Ctenosatirus aus dem Gottinger Buntsandstein. Centralbl. Min. Geol. Pal., 1914: 496-499. 1920a. Bemerkungen zur Systematik und Stammesge- 1972 THECODONT CLASSIFICATION 21 schichte einiger Reptilieii. Z. Indukt. Ahstamm.-\"ererbungsl., 22: 209- 212. 1920b. Ein Parasuchier aus dem obeit-n Muschelkalk von Bayreiith. Senckenbergiana, 2: 143-145. 1921. Neue Pseudosuchier und Coelurosaurier aus detii Wiirttembergischen Keuper. Acta Zool., 2: 329-403. 1925. Die Bedeutung der Sp/irnosuc/ius CJruiipe fiir den Ursprung der Krokodile. Z. Indukt. Ah>tainm.-Vererbungsl., 38: 307-320. 1936. The constitution of the Thecodontia. Amer. Jour. Sci., Ser. 5. 32: 207-217. 1938. Ein grosser Stagonolepide aus der jiingeren Trias Ostafrikas. Neues Jahrb. Min. Geol. Pal., Beilage-Bd., 80: 264-278. 1939. Ein kleiner Pseudosuchier und ein Saurischier aus den ostafrikanischen .Mandaschichten. Neues Jahrb. Min. Geol. Pal., Abt. B, 81: 61-69. 1940. Eine Reptilfauna aus der altesten Trias Nordruss- lands. Neues Jahrb. Min. Geol. Pal., Abt. B, 84: 1-23. 1942. Die Fauna der Panchet-Schichten in Bengalen. Zentralbl. Min. Geol. Pal., Abt. B, 1942: 354-360. _. 1944. Die fossilen Reptilien des Siidamerikanischen Gondwanalandes. Ergebnisse der Sauriergrabungen in Siidbrasilien, 1928/1929. Munchen: C. H. Beck'sche. 332 pp. 1956. Palaontologie und Phylogenie der niederen Tetra- poden. Jena: G. Fischer. 716 pp. 1958. Ein Pseudosuchier aus dem Buntsandstein des Schwarzwalds. Jahresh. Ver. Vaterland. Naturk. Wiirttemberg, 113: 111-112. 1959. Saurians in China and their relations. Vert. Pal- asiatica, 3 : 119-123. 1960. Ein grosser Pseudosuchier aus der Orenburger Trias. Palaeontographica, Abt. A, 114: 105-111. 1962. Die Pseudosuchier als Wurzelgruppe der meisten Landsaurier der Jura- und Kreidezeit. Neues Jahrb. Geol. Min. Pal., Monatsh., 1962: 1-6. Hughes, B. 1963. The earliest archosaurian reptiles. S. Afr. Jour. Sci., 59: 221-241. J.\EKEL, O. 1910. Ueber einen neuen Belodonten aus dem Buntsandstein von Bernburg. Sitzungsber. Ges. Naturf. Freunde, Berlin, 19I(i: 197— 229. Janeksch, W. 1949. Ein neues Reptil aus dem Keuper von Halberstadt. Neues Jahrb. Min. Geol. Pal., Monatsh., Abt. B, 1949: 225-242. Jepsen', G. L. 194S. A Triassic armored reptile from New Jersey. New Jersey Dept. Conserv. Misc. Geol. Pap., 1948: 5-20. KiTCHING, J. W., J. \V. COLLINSON, D. H. El.LIOT, .A\D E. H. CoLBERT. 1972. Lystrosauriij Zone (Triassic) fauna from Antarctica. Science, 175: 524-527. 22 BREVIORA No. 395 Krebs, B. 1965. Ticinosuc/ius fcrox nov. gen. nov. sp. Ein neuer Pseudo- suchier aus der Trias des Monte San Giorgio. Schweiz. Palaont. Abh., 81: 1-140. 1966. Ziir Deutung der C/iirot/itrium-FahTten. Natur und Museum, 96: 389-396. . 1969. Ctrnosaurisnis kornetil (v. Huene), die Pscudosuchia und die Buntsandstein-Reptilien. Eclog. Geol. Helvet., 62: 697-714. McGregor, J. H. 1906. The Phytosauria, with especial reference to Mys- triosuchus and Rhytidodon. Mem. Amer. Mus. Nat. Hist., 9: 29-101. Nash, D. 1968. A crocodile from the Upper Triassic of Lesotho. J. Zool. London, 156: 163-179. Newton, E. T. 1894. Reptiles from the Elgin sandstone.-Description of two new genera. Phil. Trans. Roy. Soc. London, B 185: 573-607. Ortl.am, D. 1967. Fossile Boden als Leithorizonte fiir die Gliederung des Hoheren Buntsandsteins im nordlichen Schwarzwald und sud- lichen Odenwald. Geol. Jb., 84: 485-589. Otschev, V. G. 1958. (New data on Russian pseudosuchians.) Doklady Akad. Nauk USSR, 123: 749-751. Peabody, F. E. 1952. Prtrolacosaurus kanscnsis Lane, a Pennsylvanian reptile from Kansas: Vertebrata, Article 1. Univ. Kansas Pal. Con- trib., 10: 1-41. Price, L. L 1946. Sobre um novo pseudosuquio do Triassico superior do Rio Grande do. Sul. Bol. Serv. Geol. Min. Brasil, 120: 7-38. Reig, O. a. 1959. Primeros datos descriptivos sobre nuevos reptiles arcosaurios del Triasico de Ischigualasto (San Juan, Argentina). Rev. Asoc. Geol. Argentina, 13: 257-270. . 1961. Acerca de la posicion sistematica de la familia Rauisuchidae y del genero Saurosuchus (Reptilia, Thecodontia). Publ. Mus. Munic. Cien. Nat. Trad. Mar del Plata, 1: 73-114. 1963. La presencia de dinosaurios saurisquios en los "es- tratos de Ischigualasto" (Mesotriasico superior) de las provincias de San Juan y La Rioja (Republica Argentina). Ameghiniana, 3: 3-20. 1967. Archosaurian reptiles: a new hypothesis on their origins. Science, 157: 565—568. 1970. The Proterosuchia and the early evolution of the archosaurs; an essay about the origin of a major taxon. Bull. Mus. Comp. Zool., 139(5) : 229-292. RoMER, A. S. 1971a. The Chanares (Argentina) Triassic reptile fauna. VIIL A fragmentary skull of a large thecodont, Luperosuclius jractus. Breviora, Mus. Comp. Zool., No. 373: 1-8. 1971b. The Chanares (Argentina) Triassic reptile fauna. X. Two new but incompletely known long-limbed pseudosuchians. Breviora, Mus. Comp. Zool., No. 378: 1-10. 1971c. The Chanares (Argentina) Triassic reptile fauna. XL Two new long-snouted thecodonts, Chanaresuchus and Gualo- siichiis. Breviora, Mus. Comp. Zool., No. 379: 1-22. 1972 THECODONT CLASSIFICATION 23 _. 1971d. Unorthodoxies in reptilian phylogeny. Evolution, 25: 103-112. _. 1972a. The Chanares (Argentina) Triassic reptile fauna XII. The postcranial skeleton of the thecodont CJiaiKursiK hits. liieviora, Mus. Comp. Zool., No. 385: 1-21. .. 1972b. The Chanares (Argentina) Triassic reptile fauna. XIII. An early ornithosuchid pseudosuchian, Gracilisuchus stipanici- corum gen. et sp. nov. Breviora, Mus. Comp. Zool., No. 3S9: 1—24. 1972c. The Chafiares (Argentina) Triassic reptile fauna in XIV. Lcivisuc/ius admixtus, gen. et sp. nov., a further thecodont fro the Chanares beds. Breviora, Mus. Comp. Zool., No. 390: 1-13. ROZHDESTVENSKII, A. K. 1964. ( Thecodontia ) . In Osnovy Paleontologii, A. Orlov, ed. [Volume on Amphibians, Reptiles and Birds], pp. 497- 506. RUSCONI, C. 1947a. Reptil Triasico de Uspallata. Bol. Paleont. Buenos Aires, No. 22: 1-2. 1947b. Primeros hallazgos de coprolitos de reptil en el Triasico de El Challao, Mendoza. Cien. Invest., 1947, No. Id: 521- 523. 1951. Laberintodontes Triasicos y Permicos de Mendoza. Rev. Mus. Hist. Nat. Mendoza, 5: 33-158. Sawin, H. J. 1947. The pseudosuchian reptile Typothnrax mradri. jour. Paleont., 21: 201-238. Sharov, A. G. 1970. [A strange reptile from the Lower Triassic of Fergania.] Paleont. Zhurnal, Akad. Nauk, 1970(1): 127-13S. See also Nauka i Zhizn, 7: 28-32, 1971. 1971. [New flying reptiles from the Mesozoic of Kazakh- stan and Kirghizia.] Trudy Paleont. Inst., Akad. Nauk, 130: 104— 112. Sill, W. C. 1967. Proterochampsa barrionuevoi and the early evolu- tion of the Crocodilia. Bull. Mus. Comp. Zool., 135(8): 415-446. Simmons, D. J. 1965. The non-therapsid reptiles of the Lufeng basin, Yunnan, China. Fieldiana: Geol., 15: 1-93. SoERCEL, W. 1925. Die Fahrten der Chirotheria. Fine palaobiologische Studie. Jena: G. Fischer, vii + 92 pp. Tatarinov, L. p. 1960. [Discovery of pseudosuchians in the l^pper Per- mian of the USSR]. Pal. Zhurnal, 1960(4) : 74-80. 1961. [Materials on the pseudosuchians of the USSR]. Pal. Zhurnal, 1961: 117-132. Van Hoepen, E. C. N. 1915. Contributions to the knowledge of the rep- tiles of the Karroo formation. 4. A new pseudosuchian from the Orange Free State. Ann. Transvaal Mus., 5: 83-87. Walker, A. D. 1961. Triassic reptiles from the Elgin area: Stui/nnolrpis, Dasygnathus and their allies. Phil. Trans. Roy. Soc. London, B 244: 103-204. 1964. Triassic reptiles from the Elgin area: Ornillio- 24 BREVIORA No. 395 sticliui and the origin of carnosaurs. Phil. Trans. Roy. Soc. London, B 248: 53-134. . 1966. Elachistosuchus, a Triassic rhynchocephaiian from Germany. Nature (London), 211: 583-585. 1968. Protosuclius, Protrrocliampsa, and the origin of phytosaurs and crocodiles. Geol. Mag., 105: 1-14. . 1970. A revision of the Jurassic reptile Hallopus 'Vic- tor (Marsh), with remarks on the classification of crocodiles. Phil. Trans. Roy. Soc. London. B 257: 323-372. Welles, S. P. 1947. Vertebrates from the Upper Moenkopi foriratlon of northern Arizona. Univ. Calif. Publ., Bull. Dept. Geol. Sci., 27(7): 241-294. Westphal, F. 1963. Phytosaurier-Gattungen und -Arten aus dem siid- westdeutschen Keuper (Reptilia, Thecodontia). Neues Jahrb. Geol. Palaont. Abh., 118: 159-176. 1970. Phytosaurier-Hautplatten aus der Trias von Mad- agaskar — ein Beitrag zur Gondwana-Palaogeographie. Neues Jahrb. Geol. Palaont. Mh., 1970, H. 10: 632-638. Woodward, A. S. 1907. On a new dinosaurian reptile (SclcromocIAus taylori gen. et sp. nov.) from the Trias of Lossiemouth, Elgin. Quart. Jour. Geol. Soc. London, 63: 140-144. Young, C. C. 1936. On a new Chasuiatosauitis from Sinkiang. Bull. Geol. Soc. China, 15: 291-311. 1944. On a supposed new pseudosuchian from Upper Triassic saurischian bearing red beds of Lufeng, Yunnan, China. Amer. Mus. Novitates, No. 1264: 1-4. 1951. The Lufeng saurischian fauna in China. Palaeont. Sinica, New Ser. C, No. 13: 19-96. 1958. On the occurrence of Chasmatosaurus from Wuh- siang, Shansi. Vert. Palasiatica, 2: 259-262. 1963. Additional remains of Chasmatosaurus yuani Young from Sinkiang, China. Vert. Palasiatica, 7: 220-222. 1964. The pseudosuchians in China. Palaecnt. Sinica, New Ser. C, No. 19: 106-205. MAf? 1 8 1985 'HARVARD B R E V I 0"W''% ^loseom of Conipsiratave Zoology Cambridge, Mass. November 6, 1972 Number 396 THE GHANARES (ARGENTINA) TRIASSIG REPTILE FAUNA. XVII. THE GHANARES GOMPHODONTS Alfred Sherwood Romer Abstract. Much of the gomphodont material in the Chanares collec- tions appears to pertain to two closely related but distinct species, I\Ias- setOfftiai/ziis pascual'i and Massctognathus tcruggii. An especially large skull is described as Massctognathus major, sp. nov. Two skulls represent a distinct form, Mcgagompliodon oligodcns, gen. et sp. nov., differing mainly in its smaller but more numerous cheek teeth. As noted previously, a large proportion of the reptilian remain.s from the Chanares beds pertain to the peculiar gomphodont side branch of the Cynodontia, abundant in all Middle Trias.sic terrestrial faunas. /Although much of the 1964-65 collection from these beds was, because of political complications, long delayed in shipment, the one box that reached the laboratory promptly contained a number of gomphodont skulls, on the basis of which I described two species as Massetognathus pas- cuali and Massetognathus teruggii (Romer, 1967). The skull materials then available sorted out clearly into two size groups, differing in skull length by about 40 percent. This figure was too great to be due to .sex differences, and since there were in this sample of the collection no intermediates in size between the two groups, erection of two species seemed fully justified. Today, with the full collection axailable, a much larger array of gomphodont skull material lies before me, and a re-study of the situation is called for. In many instances in vertebrate pale- ontology, specific diagnoses have been made on .supposed size differences when only a few specimens were known, only to be pro^•ed invalid when more abundant material became avail- able. Might that not be the case here? Two large specimens pro\e to be of a distinct type (descril^ed below) but mo.st ap- pear to pertain to Massetognathus and show a wide \'ariation in 2 BREVIORA No. 396 size. They are here tabulated according to skull length meas- ured to the condyles. Owing to imperfections in the material or incomplete preparation, this measurement was axailable onlv in a fraction of the cranial specimens, but in a number of other cases this basal length could be calculated fairly accurately from other measurements. Table I. Massetognathus cranial specimens grouped by BASAL length, IN MM. 61-70 mm 1 71-80 mm 11 81- 90 mm 16 91-100 mm 13 101-110 mm 8 111-120 mm 6 121-130 mm 4 131-140 mm 5 141-150 mm 2 151-160 mm 1 161-170 mm 4 200-210 mm 1 In any population of living reptiles or any adequate sample of a fossil form, the size distribution is a characteristic one; the great proportion of the specimens represent young adults, but in addition there are present a few forms of somewhat greater size, presumably older individuals in which further growth had occurred. To some degree our distribution is of the type that suggests a single species, with the greater part of the specimens concentrated in the size range attributed in my earlier paper to M. pascuali, the holotype skull of which measured 87 mm in length. But in two regards the collection does not agree with the assumption that we are dealing with a single species. ( 1 ) In a typical one-species population only a very few "elderly" in- dividuals are present far above the "young adult" size; here nearly half of the specimens extend onward toward sizes far exceeding that of the young adults. (2) Even excluding the single extra-large skull tabulated, these larger specimens run upward to a skull size about double that of "young adults" of M. pascuali — a situation quite out of the range of possibility of size increase in any known reptile population. It seems cer- tain that in these larger specimens we are dealing with repre- sentatives of a second species, M. teruggii, less abundant than 1972 CHANARES GOMPHODONTS 3 M. pascuali, but of larger size — the holotype skull of M. teruggii measurino^ 125 mm in basal length. Confirmation of this conclusion is given b\ a consideration of the "molar" dentition. In many cases the lower jaws arc firmly occluded with the upper, so that (in default of difficult prep- aration) the tooth surfaces are not clearly seen. Howc\cr, the "molars" are \isible in surface view in a number of specimens of both the smaller, M. pascuali, and of the larger, M. teruggii, types. In the M. pascuali specimens the t\pical "molars" have an anteroposterior width which a\"erages close to 3 mm; those of M. teruggii average 3.75 mm. Further, Mr. John Hillman, who has studied gomphodont dentitions intensi\-ely, pointed out to me that in the teruggii specimens the "molars" are also proportionately broader mediolaterally than in the smaller M. pascuali specimens. It is highly improbable that this increase in indi\-idual molar size could occur during the lifetime of an indixidual. There is almost no exidence of any \ertical replacement of these very deep-rooted teeth in the adult. There is e\-idence here, as in other gomphodonts, of a trend for suppression of one or two of the smaller anterior "molars" during the hfetime of an individ- ual, and for the addition of one or more teeth at the posterior end of the series. But there is no indication of de\elopment of the complex type of tooth replacement found in the manatee, by which a whole series of "molars" might be replaced by larger successors pushing forward from the back end of the series. If such replacement were to occur, we would expect the new teeth added at the back of the series to increase in size. For the most part the "molars" in the back part of the series are larger than those anteriorly placed. But Mr. Hillman, who has made careful measurements of the Massetognathus dentitions, tells me that in a number of instances the last one or two teeth in the series are smaller, rather than larger, than those anterior to them. Apart from the two species of Massetognathus discussed above, the Chafiares gomphodont series certainly includes other vari- ants. As discussed below, two specimens appear to represent a form generically distinct from Massetognathus. And in addition, the specimens assigned to that genus are quite surely not all as- signable to M. pascuali and M. teruggii. As mentioned above, the M. teruggii type has a skull length of 125 mm. \\'e would reasonably expect a number of older indi\iduals to exceed this figure to a modest degree, up to about 160 mm or so — -that is, BREVIORA No. 396 Figure 1. The holotype skull of Massrtoc/iiat/ius major in dorsal view. X 2/5. to a skull length of as much as 30 percent in excess of a "young adult." But the finding of four specimens in the 160-170 mm bracket is disturbing, and a skull of over 200 mm in length — two-thirds again as large as the type- — ^ gives us an impossible situation. We have here, quite certainly, a third, large species. Massetognathus major sp. no v. Holotype: La Plata Museum, No. 64-XI-14-15, (field no. 55 ) . From the Chafiares formation, about 4 km southeast of the mouth of the Rio Chaiiares, La Rioja Province, Argentina. Diagyiosis. Generally comparable to other species of Mas- setognathus, but orbits extend relatively far forward, antorbital region narrower than in other species; cheek tooth rows less divergent posteriorly; size large, the holotype with a basilar skull length of 205 mm. The species is based primarily on a single skull (Figs. 1, 2) far larger than any other assigned to this genus. The specimen 1972 CHANARES GOMPHODONTS Figure 2. The holotype skull of Massrtngnaihus major in ventral view. X 2/5. was found exposed ^\•ith the palatal surfaces upward, and in a somewhat weathered condition, so that the cheek teeth do not show the crown pattern well, and the posterior part of the skull is imperfectly preserved. In most regards the skull agrees well with the previously described species of Alassetognathus. Dis- tinctive, howe\er, is the relati\'e narrowness of the snout and a consequent!)- lesser de\elopment of the broad shelf which, in ventral view, extends far out on either side of the cheek tooth series. The series of cheek teeth are but little cur^•ed, and diverge but little posteriorly. The orbits, instead of being essentially sub- circular in outline, extend forward in triangular fashion, with the apex of the triangle lying at the entrance to the lacrimal duct. In relation, presumably, to large size, the sagittal crest is well developed, the ridges bounding the temporal openings fus- ing medially a short distance back of the postorbital bar, with complete obliteration of the parietal foramen. Thirteen cheek teeth are present on either maxilla; the most anterior are rela- BREVIORA No. 396 Figure 3. The skull of Megagomphodon oligodcns in dorsal view. This figure and figs. 4—6 are composites, based on the holotype and MCZ 4138. X 2/5. lively smaller in size, as compared with those farther back, than is the case in either M. pascuali or M. teruggii. Megagomphodon oligodens gen. et sp, nov. Holotype: La Plata Museum No. 64-XI-14-16, (field no. 65 ) . Chanares formation. La Rioja Province, about 6 km ENE of the mouth of the Rio Chaiiares. Diagnosis. A relatively large traversodontid gomphodont, with a basal skull length on the order of 180 mm. Cheek teeth relatively small, especially anteriorly, and about 17-18 in num- ber. Skull relatively slender, the width across the orbital region being but about two thirds the total skull length. In the collection two skulls, the holotype and MCZ 4138, rep- resent a gomphodont type clearly distinct from Massetognathus. Incom.plete jaws, but no postcranial materials, are associated with both. Neither is too well preserved; the holotype has fairly well preserved cheek teeth, but is imperfect posteriorly; the posterior part of the braincase is preserv^ed in MCZ 4138, but teeth are represented only by their roots. My description is based on a combination of features present in one skull or the other, and my illustrations (Figs. 3-5) are hkewise composite. On neither skull are the sutures well shown, and I have in consequence omitted most of them in mv figures. 1972 CHANARES GOMPHODONTS Figure 4. The skull of Megagnmphodon oligoJitis in lateral view. X 2/5. Figure 5. The skull of Megagomphodon oligodcns in ventral view. X 2/5. Figure 6. The dentary of 'Megagomphodon oligodcns in lateral view. X 2/5. 8 BREVIORA No. 396 c^^^ Figure 7. Left upper and lower cheek teeth of Mcgagomphodon ol'igo- dens. X 3/2. The skull is somewht more slender than in Aiassetognathus. The "muzzle" is notably narrow, with a constriction back of the canine region, and expansion in width does not take place until well back toward the subcircular orbits. In correlation with large size, the sagittal crest is well dexeloped, with the two crests becoming closely apposed not far back of the le\el of the post- orbital bar. Dentaries are preserved in both specimens (Fig. 6). The "angular" region is well developed, and in MCZ 4138 has a backwardly pointed tip. As in all traversodonts the ascending ramus is highly developed, and extends far back dorsally. In most specimens of Chaiiares gomphodonts in which the bone is well preserved, the posterior end of the ramus is rounded; in both specimens of the present form this process is sharp-tipped posteriorly. On the inner surface of the dentary there is a longi- tudinal recess, typical of advanced cynodonts, for the reception of the supporting bar formed by surangular + angular + pre- articular. The most distinctive feature of this genus is the nature of the cheek teeth (Figs. 5, 7). These are numerous, and in the holo- type, where the dentition is nearly completely preser\ed, there appear to be 17 or 18 "molars" in both upper and lower jaws. This is, of course, a definitely higher count than in Masseto- gnathus. On the other hand, the individual teeth are definitely 1972 CHANARES GOMPHODOXTS 9 smaller than in that genus. The length of the entire row of upper cheek teeth is about a third of the skull length in this form and in Massetognathus as well; but since the number of teeth in Megagoynphodon is greater, the anteroposterior dimen- sions of indi\idual teeth is relati\ely small; the a\erage antero- posterior length of an indi\idual tooth in Massetognathus is about 3 percent of the skull length, in Megagoynphodon only about 2 percent. The Megagoynphodon teeth are also relatively small in transverse measurement; the broadest teeth in this genus measure only about 4 percent of skull length, whereas this width in Massetognathus is approximately 5 percent. Except for the reduced tooth size, Megagoyyiphodon is ob- \-iouslv not distantly related to Massetognathus and this genus mav perhaps ha\"e been derived from such a form as M. yyiajor. Collection and preparation of this material was made possible bv successive grants from the National Science Foundation. REFERENCE CITED RoMER, A. S. 1967. The Chanares (Argentina) Triassic reptile fauna. III. Two new gomphodonts, Massetognathus pascuali and .1/. tcruggVi. Breviora, Mus. Comp. Zool., Xo. 264: 1—25. ML'S. CCMP. ZOOL LIRRARY MAR 1 8 1965 B R E V I Q.Ja,A l^Ioseiim of Comparative Zooilogy Cambridge, Mass. November 6, 1972 Number 397 XYLOREDO, A NEW TEREDINID-LIKE ABYSSAL WOOD-BORER (MOLLUSCA, PHOLADIDAE, XYLOPHAGAINAE) Ruth D. Turner Abstract. Xylorrdo, a new genus of Xylophagainae characterized by having a long burrow with a calcareous lining, is described. Three new species belonging to this genus are also described, Xylorrdo nooi and X. ingolfia from the Atlantic and A', naccli from the Pacific. The systematic position of the genus, its superficial resemblance to the Teredinidae, and the factors controlling the distribution of the species are discussed. INTRODUCTION The Xylophagainae are benthic, mainly deep-sea wood-borers. Their distribution is dependent on the presence of woody plant material and their dispersal on the free-swimming lar\al stage. From available e\idence it appears that wood in the deep sea is scarce and its distribution patchy. Bruun (1959) reported on the remnants of plant material collected by the R/\' GALA- THEA during a circumna\igation of the globe. J. Knudsen (1961) reported on the Xylophaga collected on that cruise and also discussed the presence of plant debris in the deep sea, sum- marizing the results of earlier expeditions. An analysis of the data in these reports shows that 56 of the 180 successful bottom stations made by the GALATHEA in depths between 400 and 7290 meters yielded plant material. Only 22 of these hauls had woody material of sufficient size (e.g., over 10 grams) to sup- port borers, but species of Xylophaga were taken at only 1 1 stations. On the basis of material now available, the ranges of many of the species in this subfamily appear to be limited. The known ranges may be a reflection of the limited number of dredging stations but may also be because 1 ) the dispersal efficiency of the larvae is poor J. Knudsen, 1961) and 2) wood in the deep sea is not only scarce but is found in "islands" of plant debris, 2 BREVIORA No. 397 usually in deep trenches near land, particularly in the regions of large rivers (Bruun, 1959). In order to better understand the distribution of these borers, we need to know more about their life history, particularly the length of time the larv^ae spend in the plankton and their behavior at the time of settlement. Most Xylophaga are obtained by dredging, though a few species occurring in high latitudes can be found in shallow water, some even just subtidallv in old \vrecks or piling (Tomlin, 1920; Dons, 1929a, 1929b, 1940). In 1961 H. Turner attached a series of wood panels to the mooring line of a buoy set out by the Woods Hole Oceanographic Institution at 39°30'N; 69°40'W (on the Gay Head, Martha\s Vineyard-Bermuda transect) in 3000 meters. When the panels were retrieved in October 1961, the bottom one was found to have been heavily attacked by Xylophaga (approximately 100 per square centi- meter). This was the first time that "new" wood had been submerged to such a depth sufficiently rapidly so that there was no possibility of the larvae settling on the way down. Unfor- tunately, the specimens were too young for positive identification and no additional panels have been exposed at this site. About the same time the United States Navy began testing the per- formance of various materials, including wood, in the deep sea (DePalma, 1969; Muraoka, 1964-1967). These tests at depths of 1000 to 2000 meters have added greatly to our knowledge of the Xylophagainae and resulted in the discovery of a new genus, represented by three new species. Species in this new genus {Xyloredo, p. 3) are remarkably close to the teredinids in superficial appearance, as they produce long, often tortuous tunnels which have a calcareous lining; the valves, as in all Xylophagainae, are teredinid-like. The lack of pallets for clos- ing the burrow and of apophyses for the attachment of the foot muscle, and the presence of a mesoplax, definitely place them in the Pholadidae, subfamily Xylophagainae. Though mentioned in the hterature (R. Turner, 1966a, 1969), this new genus was not named or described. Specimens of Xyloredo nooi (p. 5), the first new species to be received, were taken from tests put out by the U. S. Na\'al Oceanographic Office in the Tongue of the Ocean, Bahama Islands, in 1737 meters (DePalma, 1969). At this site, panels set in contact with the bottom were completely riddled, while those 15 meters ofT the bottom were entirely free of borers. Wood panels sub- merged by the U. S. Naval Civil Engineering Laboratory south of San Miguel Island off Port Hueneme, California, in 2072.6 1972 XYLOREDO 3 meters were the source of the second new species, Xyloredo naceli (p. 9). At this site, panels 15 centimeters off the bot- tom were much more se\erely attacked b\' two species of xylo- phagaids than those only one meter up on the rack (Muraoka, 1966b). Tipper (1968), working with Xylophaga washingtona Bartsch off the Oregon coast, found that the number of borer penetrations dropped most markedly within the first 6 centi- meters abo\e the bottom. These findings suggest that the larxae probably remain close to the bottom throughout their free- swimming stage and that, except in areas where there are strong bottom currents, thev probablv are not widelv dispersed (R. Turner, 1966b, 1968). Xyloredo ifigolfia (p. 7), the third new species belonging to this genus, came from a piece of wood dredged in 1896 by the INGOLF Expedition south of Eyrabakki, Iceland, from 1 783 meters. This species appears to be more closely related to the California species than to the one from the Tongue of the Ocean. As with many species of molluscan wood-borers, it is diffi- cult to determine which characters of the species of Xyloredo will pro\'e most stable and useful for taxonomic purposes until large series from several localities, representing all growth stages, are a\ailable for study. As intimated above, wood-borers from great depths are difficult to obtain, and such series may not be a\'ailable for many years. In fact, the publication of the taxa included in this report has been delayed for five years while ef- forts were made to obtain additional material. Hopefully, knowledge of the existence of these remarkable Xylophagainae will stimulate personnel on dredging cruises and deep submers- ible dives to collect and preserve woody plant material. In addition to borers, dredged wood is a rich source of many spe- cies of benthic invertebrates. Xyloredo^, new genus Type species. Xyloredo nooi Turner, new species. Description. X^ahes t\pical of subfamih Xylophagainae and virtually indistinguishable from valves of Xylophaga; apophyses lacking; chondrophore, internal ligament, umbonal-ventral sul- cus and ridge, and ventral condyle well developed. Mesoplax 'A C(;mpouiid word, combining Xylu- from XylophcKja and -redo from Teredo, indicating the teredinid-like appearance of these Xylophagainae. 4 BREVIORA No. 397 small, flat, triangular and only lightly calcified. Burrow, de- pending on size and age of specimen, 5 to 30 times length of valves. Posterior two-thirds of burrow with calcareous lining marked with distinct rings and covered by thin, outer periostra- cal layer which extends anteriorly as border (Plate 1, fig. 2). Periostracal covering of vaKes extending posteriorly as sheath that is continuous with heavy periostracal band on anterior end of tube. Mantle fold attached to anterior end of tube, point of addition of both periostracum and calcium. Burrow lining in very young specimens either entirely periostracal or lightly calcified. Combined incurrent and excurrent canals extend length of tube and attach to calcareous lining at anterior end of siphons. Siphons short, separate and apparently margined with few cirri. Visceral mass similar to Xylophaga (Purchon, 1941; Turner and Johnson, 1971), contained completely between valves. Ctenidium of sinsle demibranch nearly coNcring sides of vis- ceral mass. Mouth broad and slitlike; labial palps inconspic- uous. Esophagus short, wide, and flattened; stomach large, subglobular, with large crystalline style sac opening into it at posterior end. Style sac bent forward so that lower end pro- trudes slightly through \isceral mass postero\entral to foot. Caecum large, U-shaped, and located mainly on the right side. Remarks. The general description of the anatomy of Xylo- redo given above was derived from a study of the remains of specimens of all three species. Unfortunately none was suffi- ciently well preserxed for detailed dis.section or sectioning. Xyloredo differs from Xylophaga by having : 1 ) a long, tere- dinid-like burrow with a calcareous lining, 2) a thin periostracal sheath extending from the valves to the heavier periostracal border of the tube, and 3) extended incurrent and excurrent canals. The siphons of Xyloredo are short and separate, and when the animal is remo\'ed from its burrow it resembles a shipworm without its pallets. Unlike the Teredinidae, Xyloredo has short, broad ctenidia, and all of the visceral mass is contained between the valves; a mesoplax is present but pallets and apophyses are lacking. In addition, the calcareous tube of Xyloredo is thin, regularly ringed, and covered externally with a shiny periostracum. The superficial resemblance of Xyloredo to the Teredinidae is a result of convergence. The Xylophagainae are probably most closely related to the Jouannetinae, a subfamily of rock- 1972 XYLOREDO 5 boring pholads, which also lack apophyses and have the pedal retractor muscle inserted in the typical bi\'al\'e position anterior to the posterior adductor muscle. The disco\"ery of this teredinid-like genus of Pholadidae neces- sitates a reexamination of fossil tcredinids, especially those with ringed tubes. If \al\es are not present, or if the inner surface of the valves cannot be examined to ascertain whether apophy- ses are present, it may be impossible to distinguish fossil Xyloredo from the Teredinidae. It is, of course, possible that all fossil tubes with regular rings are Xyloredo or some closely related fossil genus. Many teredinid tubes are marked with rings, but not in the regular, ex'enly spaced fashion found in Xyloredo. Fossil deposits produced in the deep sea are \irtually un- known, so perhaps there is no need to be concerned about fossil Xyloredo. Howexer, like Keopilina, it might well be that at one time their ancestors li\ed in shallower waters. It is possible that, as a result of competition with the more efficient Teredinidae, only the deep-water species sur\'ive. Range. Species of Xyloredo are known in the Atlantic Ocean from off Iceland and from the Tongue of the Ocean, Bahama Islands. In the eastern Pacific they are known only from ofT San Miguel Island, near Port Hueneme, California. Xyloredo may well be world-wide in distribution in depths of over 1500 meters, but further dredging and testing are needed to proxe this. Xyloredo nooi^, new species Plates 1 and 2 Type locality. From test panels submerged in the Tongue of the Ocean ('25°54'N; 77°49'W), off the north end of Andros Island, Bahama Islands, in 1737 meters, from 4 April 1962 to 17 February 1965 i DePalma, 1969). Types. Holotype, Museum of Comparative Zoology 279631. Paratypes from the same and other panels exposed at the same locality. Museum of Comparati\e Zoology 279632, 279633, 279634, 279635, and the Universitetets Zoologiske Museum, Copenhagen. Description. Shell globose, valves reaching 10 mm in length and 10.5 mm in height, thin, fragile: umbos prominent. Perio- 'An acronym based on the initials of the Navy Oceanographic Office, Washington, D.C., which was responsible for the tests in the Tongue of the Ocean, Bahama Islands, from which the specimens were obtained. 6 BREVIORA No. 397 stracum relati\ely thick, golden brown, glistening and covering entire valve. Beaked portion of anterior slope sculptured with numerous, finely denticulated ridges (24 ridges on holotype). Posterior portion of anterior slope narrow; ridges closely spaced, coarsely denticulated, usually thickened posteriorly and extend- ing only about one-half distance to shallow umbonal-ventral sulcus. Disc sculptured with fine growth lines only. Posterior slope high, reflected near dorsal margin and sculptured with fine growth lines. Umbonal reflection thin, wide and adhering closely to surface of valves posteriorly, free anteriorly. Inner surface of valves smooth and glistening. Umbonal-ven- tral ridge narrow-, high, slightly to strongh segmented and not greatly enlarged at ventral condyle. Chondrophore and inter- nal ligament prominent (Plate 2, fig. 5). Disc separated from posterior slope by pronounced groove extending from umbo to posterior ventral margin. Posterior adductor muscle scar large, elliptical, and divided; upper part (on reflected portion of pos- terior slope) with irregular impressions, lower part with chevron- shaped impressions. Anterior adductor muscle scar covering umbonal reflection. Siphonal retractor muscle scars not im- pressed. Pedal retractor muscle scar elongate, located about midway on anterior margin of posterior adductor muscle scar. Mesoplax small, consisting of two flat, broad, subtriangular plates lying on dorsal surface of anterior adductor muscle and composed largeh- of periostracum. Burrow long, teredo-like, with thin, calcareous lining for about three-fourths its length. Tubular lining marked with dis- tinct rings and covered with thin, yellow-brown periostracum which extends as border anteriorly ( Plate 1 , fig. 2 ) . Portion of animal between valves and tube covered by smooth, golden- brown, periostracal sheath, continuous anteriorly with perio- stracum of \-al\es and posteriorly with periostracal border of tube. Tubes of young specimens (up to 10 mm long) may be entirely periostracal. Burrow opening on surface of wood with white, slighth raised, often divided cone about 1 mm in diam- eter. Siphons short and separate. Protoconch unknown. Valve measureynents. Length Height 9.5 mm 10.0 mm Holotype 9.5 9.8 Paratype 7.2 7.8 // 6.7 6.9 // 5.0 5.1 // 1972 XYLOREDO 7 Remarks. This species is closely related to both Xyloredo naceli and A', ingolfia (see also Remarks under ingolfia). It differs from them in ha\-ing- a much thinner burrow linins:, a high, reflected posterior slope on the \al\es, and a proportion- ately smaller, more highlv placed and di\'ided posterior adduc- tor muscle scar. In addition, the periostracal sheath extending between the \al\cs and the calcareous tube is smooth. On the basis of the material at hand, nooi appears to be a much larger species. Size in borers, howe\er, is often not a reliable taxo- nomic character, and an understanding of the size range re- quires the examination of large series from \ aricd substrata. Stenomorphic adults often result from o\er-crowding or from penetration of an unusually hard substance. The panels from Avhich specimens of nooi were remo\ed had been in the w^ater for o\-er 2/2 years and were so badly rid- dled that the white, calcareous tubes showed through the thin surface. The general appearance of the wood was the same as that of a panel riddled with teredinids. It was not until the borers were exposed and examined closely that their pholad af- finities were recognized. Many of the specimens were dead and their burrows filled with mud, with only the calcareous tube and \'ah'es remaining. Unfortunately, the panels were not preser\ed immediately upon remo\al from the water, so the borers were in poor con- dition. Therefore, little anatomical work could be done, though it was possible to determine that the visceral mass was contained between the \ahes and that there was a large, U-shaped, wood- storing caecum. The bottom at the test site was a grav mud; the temperature was 4°C and the salinity 35 7oo (DePalma, 1969). Range and specimens exaynined. Known onh' from the t)pe locality. Xyloredo ingolfia\ new species Plates 3-5 Type locality. From wood dredged by the INGOLF Expe- dition at station 67, south of Evrabakki, Iceland (61°30'N; 22°30'W), in 975 fathoms [1783 meters]. Types. Holotype, Museum of Comparati\-e Zoology 279636. Paratypes from the same locality. Museum of Comparati\e ^Named for the Danish IN'GOLF Expedition, which collected the wood from which the specimens were obtained. 8 BREVIORA No. 397 Zoology 279637, and the Universitetets Zoologiske Museum, Copenhagen. Description. Shell globose, val\-es reaching 2.5 mm in length and 2.0 mm in height, thin, fragile, with prominent umbos; thin, glistening, almost colorless periostracum covering disc and posterior slope. Beaked portion of anterior slope wide, extend- ing more than one-half distance to ventral margin; sculptured with close-set and very finely denticulated ridges. Posterior por- tion of anterior slope about two-thirds width of beak, sculptured with close-set ridges which extend to very slightly impressed umbonal-ventral sulcus. Disc sculptured with well-marked growth lines. Posterior slope small, low and not clearly demar- cated on outer surface of valve. Umbonal reflection thick, nar- row, short and free except at posterior end. Inner surface of N'alves smooth, slightly shiny to chalky (per- haps owing to long preservation ) . Umbonal-xentral ridge wide, flattened, often varying in width, irregularly segmented and not enlarged at \entral condyle. Chondrophore and internal liga- ment large. Disc not clearly separated from posterior slope. Posterior adductor muscle scar large, slightly raised, elliptical, extending nearly to ventral margin, with irregular, transxerse impressions. Anterior adductor muscle scar covering umbonal reflection. Siphonal retractor muscle scars not impressed. Pedal retractor muscle scar small, elongate to oval and located just an- terior to posterior adductor muscle scar. Mesoplax of two very small, narrow, subrectangular, flat, calcified plates lying on dorsal surface of anterior adductor muscle. Burrow 10 to 15 times length of valves; calcareous tubular lining three-fourths length of burrow. Tube relatively heavy, marked with uniform, close-set, raised rings, and covered with light tan periostracum which extends anteriorly as border. Por- tion of animal between vahes and tube co\'ered by thin, irregu- larly ridged periostracal sheath. Burrow opening often with small, white cone about 0.5 mm in diameter. Siphons short; incurrent siphon slightly longer than excurrent. Protoconch large, medium golden-brown and sculptured with fine, concentric ridges. Valve yneasurcments. Len gth Hei ght 2.5 mm 2.3 mm Holotype 2.5 2.0 Paratype 2.1 2.0 // 2.0 1.9 ft 1.5 1.4 It 1.5 1.2 " 1972 XYLOREDO 9 Remarks. This species is most closely related to Xyloredo nacdi from the eastern Pacific. It difTers in havins^ a less well- de\'eloped posterior slope, a shallow, indistinct umbonal-\'entral groo\e, a flattened umbonal-\entraI ridge, and in having the valves longer than high. In addition, the periostracal sheath posterior to the valves is not pustulose as in naceli, and the cal- careous tube is much heavier and has raised rings. It also differs from naceli in having a much larger protoconch, which is sculp- tured with fine, concentric ridges. It differs from A', nooi in ha\'ing the valves longer than high, in haxing a low, rounded posterior slope, in lacking the distinct groove on the inner surface separating the disc from the poste- rior slope, and in ha\ing the plates of the mesoplax subrectangu- lar and well calcified. The wood dredged b) the INGOLF Expedition had been hea\'ily attacked by large teredinids before sinking to 1783 meters. These were all dead, only the empty tubes remaining as evidence of their presence. Filling every available space be- tween the teredinid burrows were pipe-organ-like tubes of Xylo- redo going both across and with the grain of the wood. Many of these tubes were 40 mm long and 1.2 mm in diameter. Jen- sen (1912: 56), in discussing the distribution of Idas [= Idasola] argenteus Jeffreys, states: "The specimens of the IN- GOLF Expedition were taken south of Iceland (61°30'N.L.; 22°30'W.L.) at 975 fathoms, under similar conditions to the last; the trawl brought up two large pieces of pine-wood, which had been pierced through and through by Teredo; in some of the Teredo tunnels were in addition mud and worm-tubes, fur- ther worms and small bivalves, namely Idas argenteus.'^ This is the same piece of wood, remnants of which were preserved in the Copenhagen Museum, from which specimens of Xyloredo ingolfia were taken. The small, straight tubes were probably thoue;ht to be vouns; teredinids. The bottom at station 67 was a brown-gray transition clay of a fine, clayey-sandy consistency (Boeggild, 1900); the tem- perature was 3°C and the salinitv 35.18 "/oo (M. Knudscn, 1899). Range and specimens examined. Known only from the type locality. Xyloredo naceli\ new species Plate 6 ^^n acronym based on the initials of the Naval Civil Engineering Lab- oratory, Port Hueneme, California, which was responsible for the tests from which the specimens were obtained. 10 BREVIORA No. 397 Type locality. From the U. S. Naval Civil Engineerina; Lab- oratory Submersible Test Unit at test site I [STU 1-4], about 30 miles south of San Miguel Island, off Port Hueneme, Cali- fornia (33°46'N; 120°46'W), in 6800 feet [2072.6 meters], sub- merged from June 1964 to July 1965. Types. Holotype, Museum of Comparati\e Zoology 279638. Paratvpes from the same localit\', Museum of Comparative Zoology 279639. Description. Shell globose, valves reaching 1.5 mm in length and 1.5 mm in height, thin, fragile; umbos prominent. Perio- stracum thin, pale yellow, covering disc and posterior slope. Beaked portion of anterior slope with 8-12 pronounced, widely spaced, finely denticulated ridges. Posterior portion of anterior slope narrow; ridges extending to uml:)onal-\entral sulcus and more coarsely denticulated. Umbonal-\'entral sulcus narrow, slightly depressed. Disc and posterior slope sculptured with fine growth lines only. Umbonal reflection narrow, free except at posterior end. Inner surface of valves smooth and glistening. Umbonal- ventral ridge narrow and indistinctly segmented. Chondrophore and internal ligament well developed. Disc separated from posterior slope by low ridge. Posterior adductor muscle scar elliptical, extending from dorsal margin nearly to ventral mar- gin and with indistinct, irregular, transx-erse markings. Pedal and siphonal retractor muscle scars impressed. Mesoplax of two small, thin, triangular plates composed almost entirely of perio- stracum. Burrow six times length of valves; calcareous tubular lining two-thirds length of Ijurrow, thin, marked with distinct rings, and covered with glistening, nearly colorless periostracum, which extends anteriorly as border. Portion of animal between \alves and anterior end of calcareous tube with finely pustulose, light tan periostracal sheath. Siphons short, of about equal length and apparently lacking cirri. Protoconch \'erv small, smooth (Plate 6, fig. 5). Valve measurements. Lenejth 1 . 1 mm 1.5 1.2 1.0 Remarks. This species, of which only eight small specimens were found, appears to be most closely related to Xyloredo in- Height 1.2 mm Holotype 1.5 Paratype 1.3 n 1.0 n 1972 XYI.OREDO 1 1 golfia from off Iceland. It diffeis by ha\"ing a imu h smaller protoconch (hence the Ian,ae in the pedixeliger stai^je must be about t\s()-thirds the size of those of ingolfia) . It also differs in that the umbonal-Nentral sulcus is impressed and the height of the \al\es is the same as or greater than the length. The l)ottom at the test site \vas a green mud; the tempera- ture was 2.1 °C. the salinity 34.52 °/o„, and the dissohed oxygen 1.26 ml L Muraoka, 1966b). This species has not been taken at test site II of the Naval Cixil Engineering Labora- tory, located north of San Miguel Island (34°06'N; 12b°42'W), where the depth is 2340 feet [713.23 meters^, the bottom tem- perature 7.2^C, the salinity 34.37 '^/n.i, and the dissohed oxygen 0.42 ml/L .Muraoka, 1965). Specimens of Xyiophaga washingtona Bartsch were found at both sites. It would appear that temperature and /or dissoh'ed oxygen may be factor (s) controlling the distribution of Xyloredo naceli. How'e\-er, so little is known about the biology of the Xylophagainae that no definite statements can be made at this time. Range and specimens examined. Known onl\ from the t\ pe locality. ACKXO\\LEDGMENTS The author is grateful to John DePalma, U. S. Navy Ocean- ographic Office, Washington, D.C.; James Muraoka, U. S. Na\al Clixil Engineering Laboratory, Port Hueneme, California: and Dr. Jorgen Knudsen, L'ni\-ersitetets Zoologiske Museum, Copenhagen, Denmark, for the test panels and dredged wood from which the specimens were obtained. Thanks are also extended to Prof. Kenneth J. Boss of the Mollusk Department. Museum of Comparati\e Zoology, and Nancy Kno\\lton. my assistant, who critical!) read the manuscript. The aid granted by the Office of Naval Research through Contract no. N00014-67- A-0298-0027 with Har\ard L^niversity is particularly appre- ciated. REFERENCES BoEGGiLD, O. B. 1900. The deposits of the sea-bottom. The Danish IX- GOLF-Expedition, Copenhagen, Vol. 1(2): 1-S9, plates 1-7, 5 text- fi.gs. Bruun', a. F. 1959. Cjeneral introduction to the reports and list of deep- sea stations. GALATHEA Report, Copenhagen, \'oi. 1 : 7-48, plates 1 — (■, text-figs. 2—11 [section 5, Remnants of plants found in the deep- sea, pp. 1 5—17]. 12 BREVIORA No. 397 DePalma, J. R. 1969. A study of deep ocean fouling, Straits of Florida and Tongue of the Ocean, 1961 to 1968. Informal Report IR No. 69- 22, U. S. Naval Oceanographic Office, Washington, D.C., pp. 1-26, text-figs. 1—6. Dons. C. 1929a. Zoologiske notiser. IV. Xylophaga pracstans, ny for Norges fauna. Norske Vid. Selskab Forhandlinger, 1(57): 169-172, text-figs. 1-7. 1929b. Zoologiske notiser. V. Xylophafia dorsalls i Norge. Norske Vid. Selskab Forhandlinger, 1(65): 196-199, text-figs. 1-6. 1940. Marine boreorganismer. III. Vekst og voksemate hos Xylopluuja dursalis. Norske Vid. Selskab Forhandlinger, 13(18): 76-78, text-figs. 1-2. Jensen, A. S. 1912. Lamellibranchiata (Part I). The Danish INGOLF- Expedition, Vol. 2(5) : 1-119, plates 1-4, text-figs. 1-5. Knudsen, J. 1961. The bathyal and abyssal Xylophaga (Pholadidae, Bivalvia). GALATHEA Report, Vol. 5: 163-209, text-figs. 1-41. Knudsen, M. 1899. -Hydrography. The Danish INGOLF-Expedition, Vol. 1(1): 23-161, plates 2-35, 2 text figs. MURAOKA, J. S. 1964. Deep-ocean biodeterioration of materials — Part I. Four months at 5,640 feet. Technical Report R 329, U. S. Naval Civil Engineering Laboratory, Port Hueneme, California, pp 1-35, text- figs. 1-24. 1965. Deep-ocean biodeterioration of materials — Part II. Six months at 2,340 feet. Technical Report R 393, U. S. Naval Civil Engineering Laboratory, pp. 1-42, text-figs. 1-5. 1966a. Deep-ocean biodeterioration of materials — Part III. Three years at 5,300 feet. Technical Report R 428, U. S. Naval Civil Engineering Laboratory, pp. 1—47, text-figs. 1—36. 1966b. Deep-ocean biodeterioration of materials — Part IV. One year at 6,800 feet. Technical Report R 456, U. S. Naval Civil Engineering Laboratory, pp. 1-45, text-figs. 1—31. 1966c. Deep-ocean biodeterioration of materials — Part V. Two years at 5,640 feet. Technical Report R 495, U. S. Naval Civil Engineering Laboratory, pp. 1—46, text-figs. 1—36. 1967. Deep-ocean biodeterioration of materials — Part VI. One year at 2,370 feet. Technical Report R 525, U. S. Naval Civ'il Engineering Laboratory, pp. 1—57, text-figs. 1—43. PuRCHON, R. D. 1941. On the biology and relationships of the lamelli- branch Xylopluuja dorsalis (Turton). Journal of the Marine Biologi- cal Association of the United Kingdom, 25: 1-39, text-figs. 1-16. Tipper, R. 1968. Ecological aspects of two wood-boring molluscs from the continental terrace off Oregon. Department of Oceanography, School of Science, Oregon State University, pp. 1-137, text-figs. 1-50. ToMLiN, J. R. L. B. 1920. Notes on Xylophaga pracstans, Smith. Pro- ceedings of the Malacological Society of London, 14(2-3): 73. Turner, H. J. 1961. Deep teredo [Xylophaga]. Oceanus, 8(2): 11. 1972 XYLOREUO 13 Turner, R. O. 1966a. Implications of recent researcli in the Teredinidae. In Hoiz und Organisnien (Cj. Beci'). Figure 2. Enlarged section of burrow showing calcareous lining with anterior periostracal border (10 X ). Figure 3. Surface of wood showing entrances of burrows (5 X ). 1972 XYLOREDO 17 1 mm Plate 4 XylorrJo Ingnlfid Turner Fiom INGOLF-Expeditlon, staticn 67, south of Eyrabakki, Iceland. Figures 1—2. (Holotype.) Lateral and dorsal views of entire animal showing the periostracal sheath, the extended anal and siphonal canals, and the short siphons. Figure 3. Lateral view of very young specimen show- ing large prodisoconch. Figure 4. Enlarged dorsal view of anterior end of entire specimen showing umbonal reflection, anterior adductor muscle, and small plates of the mesoplax in place. 18 BREVIORA No. 397 Plate 5 Xyloredo ingolfia Turner From INGOLF-Expedition, station 67, south of Eyrabakki, Iceland. Figure 1. Outer view of left valve showing w'ide anterior slope and low, rounded posterior slope. Figure 2. Inner view of left valve showing large prodisoconch, strong umbonal-ventral ridge, reduced posterior slope, and lightly impressed muscle scars. Figure 3. Inner view of right valve. Figure 4. Inner view of left valve. Figure 5. Outer view of left valve with periostracal sheath attached. 1972 XYLOREDO 19 Plate 6 Xylorrdo naceli Turner From U. S. Naval Civil Engineering Laboratory test site I, south of San Miguel Island, Santa Barbara Islands, California. Figure 1. Lateral view of specimen partially dissected from the wood showing the calcareous tube \vith the anterior periostracal margin and the papillose periostracal sheath covering the animal between the tube and the valves. Figure 2. Anterolateral view of entire specimen showing foot, mesoplax, and siphons. Figure 3. Lateral view of holotype. Figure 4. Outer view of left valve. Figure 5. Inner view of left valve showing muscle scars, chondrophore, umbonal-ventral ridge, and small prodiso- conch. MAR i 8 7985 B R E V I O t-* IMuseinn of Comparative Zoology Cambridge, Mass. March 6, 1973 Number 398 PSEUDOBEACONIA, A PERLEIDIFORM FISH FROM THE TRIASSIC SANTA CLARA FORMATION, ARGENTINA Peter Hutchinson^ AbstR-ACT. Two species of the genus Psevdobeaconia, P. braeac- dnii and P. elegans, fx-om the Santa Clara Formation, Argentina, are described. The genus is an advanced member of tlie family Colobo- dontidae (subclass Chondrostei, order Perleidiformes), and is closely- related to Meyidocinia from the same locality. It is suggested that the success of the Perleidiformes is due, in part, to the evolution of a crushing dentition. The detailed structure of the unpaired fin rays indicates that control of the fin web area was possible in Pseudo- beaconia, and probably in all Perleidiformes. The identification of Psendobcacoiua does not help the determination of the age of the Santa Clara Formation. Abstracto. Se describen dos especies del genero Pseudobeaconia, P. bracacclnii y /'. eleyans, de la Formacion de Santa Clara, Argen- tina. El genero es un miembro avanzado de la familia Colobodontidae (subclase Chondrostei, orden Perleidiformes), y esta intimamente relacionado con Mendocbiia de la misma localidad. Se sugiere que el exito de los Perleidiformes se debe, en parte, a la evolucion de una denticion molar. La estructura detallada de los rayos de las aletas impares indica que el control del area de la membrana de las aletas era posible en P.'^rudnbcnconia, y probablemente en todos los Perleidi- formes. La identificacion de Pseudobeaconia no facilita la deteniii- nacion de la edad de la Formacion de Santa Clara. INTRODUCTION In 1944 three genera of fossil fish were described from the Santa Clara Formation, a series of shales, mudstones, and sand- stones that outcrop between the pro\inces of San Juan and Mendoza, western Argentina ( Bordas, 1 944 ) . Two of these genera were new, Mendocinia and Pseudobeaconia, while the 'Gothic House, Drayton, Berks, England. 2 BREVIORA No. 398 third was doubtfully identified as Cleithrolepis. More recently Schaeffer (1955) has redescribed Mendocinia and has shown that it is a member of the Perleidiformes. Pseudobeaconia re- mained relatively unknown, although it appeared in most class- ifications as a member of the Redfieldiiformes {e.g., Gardiner, 1967a) because Bordas had described similarities between it and a redfieldiiform genus called Beaconia ( = Brookvalia, Hutchinson, 1972) from the Triassic Hawkesbury Series of New South Wales ( Wade, 1 935 ) . The material described here was collected by an expedition from the Museo de La Plata and the Museum of Comparative Zoology, Harvard University, in 1964, and is derived from two localities in the Santa Clara Formation. The first is near the north end of the formation, near the Puesto of Santa Clara, at the east margin of the Precordillera in southern San Juan Prov- ince. This is the same locality that produced the fish described by Bordas in 1 944. The second locality is north of the Quebrada de la Montana, a short distance east of the boundary' between the Santa Clara beds and the Villavicencio group (Romer, 1966). There is little doubt that the specimens described here rep- resent the two species of Pseudobeaconia described by Bordas in 1944. The reasons for this identification are given in the dis- cussion below. P. bracaccinii occurs in both the locaUties men- tioned above, while P. elegans occurs only in the former. Pseudobeaconia is preserved in a matrix of dark mudstone or, more commonly, shale. The shales from the second locality usually contain bands of a light calcareous deposit. There may be up to ten such bands in a sample 10 mm thick. As well as fish, fragments of plant material are preserved in the shales. Preservation is extremely good and the fish have either been prepared with a fine needle or sometimes, when they occur in calcareous bands, dissolved in acetic acid to reveal natural moulds. All the specimens had been crushed to a thickness of, at most, 2 mm. The Santa Clara Formation is difficult to date because it is limited to a block isolated by faulting during the Tertiary. Bor- das (1944) is of the opinion that the Santa Clara Formation is contemporaneous with the Hawkesbury Sandstone of Brook vale. New South Wales. This conclusion is based on the supposed close relationship between Pseudobeaconia and Beaconia, a con- clusion that is unfounded according to evidence presented here. Harrington (unpublished MS) notes that, near the locality de- 1972 TRIASSIC FISH FROM ARGENTINA 3 scribed by Bordas, there are plant remains that include Glossop- teris, Walkonia. Pecopteris, Gondwanidium, Cordaites, and Di- cranophyllu7n, which are Permian in age. But, as noted above, the Santa Clara Formation is isolated from other sediments in the area, and there is no evidence that the flora listed by Har- rington occurred within its boundaries. Schaeffer ( 1 955 : 3 ) has estimated the age of the beds of the Cacheuta basin (and by inference the Santa Clara Formation) to be Norian. This estimate is based on a similarity between floras of the Cacheuta and Upper Triassic sediments of the Northern Hemisphere. Lack of similar comparati\e data from the Middle Triassic means that this estimate can only be regarded as an indication of the youngest possible age of the Cacheuta beds. A possible estimate of the age of the Santa Clara Formation depends on correlations made recently bv Bonaparte and runs as follows. The Santa Clara Formation forms one of several outcrops that lie within the Cacheuta basin of deposition (Stipanicic, 1967: 3). Another of these outcrops contains the Potrcrillos Formation from ^vhich a cynodont lower jaw, Col- bertosaurus, has been recovered (Minoprio, 1954). Colbertosau- rus is similar to Pascualgnathus from the Puesto Viejo Forma- tion (Bonaparte, 1966a: 4-5). Thus the Potrerillos and Puesto Viejo formations can be considered to be very similar in age. The age of the Puesto Viejo formation is considered to be Scy- thian because it also contains a dicynodont Kannemeyeria (Bon- aparte, 1966b: 266) and a cynodont Cynognathus (Bonaparte, 1969), both typical genera of the Lower Triassic South- African fauna. It should be emphasized that this estimate of Lower Triassic age for the Santa Clara Formation is tentatiNC and open to two objections. Firstly, it is not certain that all the outcrops within the Cacheuta basin were once continuous. Secondly, the re- semblance between Colbertosaurus and Pascualgnathus is not completely convincing, and has limited significance, according to Bonaparte (1966a: 5). The identification of Alendocinia and Pseudobeaconia from Santa Clara adds nothing of significance to the debate on the possible age of the formation. Both genera are more advanced than Perleidus from the Lower Triassic of Madagascar, Green- land, and Spitsbergen but, because they are not directly de- scended from Perleidus, this does not necessarily indicate that the Santa Clara Formation is younger than Lower Triassic. An- other member of the same perleidid radiation, Meidiichthys, 4 BREVioRA No. 398 from the Cynognathus zone of South Africa, demonstrates that advanced features such as an anteriorly inclined suspensorium evolved extremely rapidly, and the presence of such characters in forms from Santa Clara does not exclude the possibility that the Santa Clara Formation is Lower Triassic in age. DESCRIPTION Order Perleidiformes Family COLOBODONTIDAE Stensio 1916 Pseiidobeaconia Bordas 1944 Pseudoheaconia Bordas, 1944: 454. Type species. Pseudobeaconia hracaccinii Bordas 1944 Diagnosis (emended). Skull short anteroposteriorly. Skull roof evenly curved to snout tip, upper jaw steeply inclined. Skull roof composed of two extrascapulars, a parietal, and a frontal. Lateral walls of skull roof composed of a dermopterotic and an accessory dermopterotic. Ventral end of the suspensorium in- clined anteriorly so that the jaw articulation lies vertically be- low the posterior orbital edge. Opercular series composed of opercular, subopercular, and three or four branchiostegal rays. Opercular anterior in position so that its anterior edge lies in advance of the level of the posterior edge of the parietal. Sub- opercular larger than opercular. Triangular dermohyal and sub- orbital elements. Dermohyal small, extending ventrally along half the anterior edge of the opercular. Dermosphenotic short anteroposteriorly, in contact dorsally with both the posterior supraorbital and the dermopterotic. Dorsal orbital edge bor- dered by two supraorbital elements. Maxilla not greatly ex- panded posteriorly. Snout composed of paired nasals, antor- bitals and (probably paired) toothed premaxillae, and a medial rostral and postrostral. Nasal bordering anterior orbital edge. Rostral large, bordering the anterior edge of the nostril. Nos- tril elliptical. Marginal teeth on maxilla and dentary conical and pointed, palatal teeth low and rounded. Body fusiform, attaining a total length of up to 120 mm. Dorsoventral length of scales in flank region up to three times their anteroposterior length. Scales in the transverse row im- mediately behind the pectoral girdle twice as long anteropos- teriorly as succeeding body scales. Scale ornament of between two and six ridges running concentrically along posterior and ventral borders. Scales in lateral line series with an indentation in their posterior borders. No hinge line at base of caudal lobe 1972 ■ TRIASSIC FISH FROM ARGENTINA 5 of body. Ridge scales present anterior to unpaired fins and on the dorsal side of the body midway between the skull and dor- sal fin. Dorsal and anal fins posterior in position. Caudal fin hemiheterocercal. All fin rays jointed and usually bifurcated. Basal fulcra present anterior to all fins. Paired fin rays appar- ently lacking fringing fulcra. Proximal lepidotrichia of all fins at least three times as long as distal lepidotrichia. Proximal lepidotrichia of unpaired fins equal in number to endoskeletal supports, and expanded proximally, with insertions for depres- sor and erector muscles. Pseudoheaconia bracaccinii Bordas 1944 Figures 1-5 ■ Pseudoheaconia bracaccinii Bordas, 1944 : 454-455, pi. 1. Diagnosis (emended). Parietal just under half the length of the frontal. Opercular plate roughly rectangular, its anteropos- terior length being under half its dorsoventral length. Ventral edge of opercular deeply convex. Posterior edge of maxilla lying anterior to the level of the posterior orbital edge, x\nterior supra- orbital extending anteriorly as far as the anterior end of frontal. Suprascapulars separated at midline. Dorsoventral length of supracleithrum twice that of cleithrum. Body fusiform. Body length, from snout tip to the posterior tip of the caudal lobe of the body, three times the greatest depth of the body. Scale formula : 25 13 21 35 Length of the longest pectoral fin rays 1.5 times the length of the longest pelvic fin rays. Some paired fin rays bifurcated distally. Dorsal fin rays not bifurcated. Dorsal fin slightly in ad\'ance of anal fin. Dorsal lobe of caudal fin sHghtly longer than ventral lobe. Ilolotypc. Museo Argentino de Ciencias Naturales at Ber- nardino Rivadavia, Catalogue of Paleontology (Vertebrates) No. 14.868 (Bordas, 1944: pi. 1). Additional material. Thirty-two specimens in the collection of the Museum of Comparative Zoology, Cambridge, Massa- chusetts. Skull. The dermal bones of the skull are restored in Figure 1. The skull roof slopes evenly to the snout, and is composed of extrascapulars (which are divided [MCZ 12891]), parietals. BREVIORA No. 398 ex pmx clav Figure 1. Pseudobeaconia bracaccinii Bordas. Restoration of skull and pectoral girdle, ornament omitted. X 5 approx. na pmx pspi Fig-ure 2. Pseudobeaconia bracaccinii Bordas. MCZ 12883. Left side of skull and pectoral girdle. 1972 TRIASSIC FISH FROM ARGENTINA 7 and frontals. The parietal is square; its anteroposterior length is just under iialf the length of the frontal. The frontal is deep and excavated above the orbit so that its widest part lies above the posterior edge of the orbit. The lateral part of the skull roof is composed of two ele- ments, a dermopterotic and an accessory dermopterotic, which are separated from one another by a vertical suture that runs from a point midway along the \entral margin of the parietal (Fig. 2). In MCZ 12883, the dermopterotic appears to be shallow, but this is because the preopercular has been displaced and overlaps its ventral part. The full depth of the dermoptero- tic is seen in MCZ 12894. The dermopterotic meets the pos- terodorsal border of the dermosphenotic, and may also have point contact with the posterior supraorbital element. A post- spiracular lies posterior to the accessory dermopterotic and meets the \entral border of the extrascapular series. In lateral view the anterior part of the postspiracular is overlapped by the opercular, but in MCZ 12893 it is seen to extend anteriorly along the entire dorsal border of the opercular. The opercular is circular, and the subopercular roughly ob- long but with a curved posteroventral border. In most speci- ments the ventral part of the subopercular is crushed against the pectoral girdle, but its outline is clearly seen in MCZ 12898, where it has been slightly displaced. Its dorsoventral length is slightly longer than that of the opercular. There are only three or four rather broad branchiostegal rays. The preopercular is high, and meets the \entral edge of the dermopterotic. Its ventral part is inclined forwards so that the lower jaw articula- tion lies almost directly belovv the posterior edge of the orbit. The preopercular has a short infraorbital process. In MCZ 12883 (Fig. 2) an element lies between the posterior edge of the maxilla and the anteroventral corner of the subopercular. This is not a fragment of the preopercular, for in MCZ 12900 there is a notch on the median surface of the \entral end of the preopercular which indicates the position of a separate element (Fig. 1). It is likely, therefore, that a quadratojugal is present. There is no sign of any sensory canal on the quadratojugal. The dorsal end of the preopercular is constricted between triangular dermohyal (MCZ 12893, 12891) and suborbital elements (MCZ 12883). In MCZ 12893, which includes a natural mould of the median surface of the cheek, there is a small triangular accessory opercular at the antcrodorsal corner of the subopercular. This element is not visible in lateral view, as it is completely overlain by the subopercular. 8 BREVIORA No. 398 The dorsal orbital edge is bordered by two supraorbitals. There is some variation in the relative lengths of these elements; in MCZ 12898 and probably also in MCZ 12883 (Fig. 2) the posterior element is slightly longer than the anterior element, whereas in MCZ 12891 (Fig. 3A), the posterior element is slightly shorter than the anterior one. The posterior orbital edge is bordered by a narrow dermosphenotic, which has an ex- panded dorsal end, and the ventral orbital edge by two in- fraorbitals. The anterior infraorbital is expanded anteriorly, while the posterior infraorbital is expanded midway along its length. In MCZ 12883 (Fig. 2), the maxilla has been displaced so that its posterior part is overlapped by the preopercular. The preopercular is so thin in this region, however, that the outline of the maxilla is clearly seen as a ridge. The upper jaw mar- gin is steeply inclined, but lies at right angles to the anterior edge of the subopercular. The maxillary teeth are pointed (MCZ 12894, 12895, 12898, and 12901),' and appear to be present onh' along the anterior half of the maxillar)- border. There are about 20 teeth in the upper jaw row. The snout is best preserved in MCZ 12893, 12891 (Fig. 3A), and 12894 (Fig. 3B). The nasal borders the anterior or- bital edge and is in contact dorsallv with both the frontal and rr(l) 2nnm A B Figure 3. Pseudobeaconia bracaccinii Bordas. A. MCZ 12891» Snout regon, right side unless otherwise indicated. B. MCZ 12894. Snout region, left side. 1972 TRIAS SIC FISH FROM ARGENTINA 9 the anterior supraorbital. The \entral part of the snout is corri- posed of paired antorbitals, which border the antero\entral corner of the orbit, and premaxillaries, which bear up to six teeth (MCZ 12894, Fig. 3B). It is not known for certain whether the premaxillaries are paired or fused. The anterior part of the snout is composed of medial rostral and postrostral elements. The rostral bears the ethmoid commissure and ex- tends dorsally so that it borders the anterior margin of the nos- tril (Fig. 3), while the postrostral is a small bone which just borders the anterodorsal comer of the nostril. The suture be- tween the rostral and postrostral elements is indistinct in MCZ 12891, but clearly seen in MCZ 12894 (Fig. SB). The lower jaw is best preserxed in MCZ 12893. Most of the dermal bones of the skull are smooth. Tubercles are present, however, on the anterior ends of the frontal and anterior supraorbital, on the nasal, and on the infraorbital part of the maxilla. Low ridges are seen on the opercular and sub- opercular elements and near the edges of the branchiostegal ravs ( Fig. 4 ) . The sensory canals of the head are enclosed in massive tubes which open to the surface through well-defined pores (MCZ 12883, Fig. 2). In adchtion, there is a vertical pit line on the ventral part of the preopercular, while short, curved, middle and posterior pit lines are present on the parietal. All the skulls of P. bracaccinii are crushed and few details of the internal anatomy are displayed. In MCZ 12897, however, part of the palate is visible and about ten low and rounded palatal tusks can be seen. In MCZ 12897 two branchial arch elements are seen, lying against the anterior edge of the cleith- rum; they are flattened and their posterior edges bear numerous toothlike projections that are less than 0.1 mm long. Pectoral girdle. The suprascapulars are D-shaped and do not meet at the midline as they are separated by a pair of scales that meet the posterior edges of the dorsal extrascapular ele- ments ^MCZ 12883, Fig. 2; MCZ 12891). The infraorbital sensory canal crosses the antcro\'entral corner of the suprascapu- lar (MCZ 12893). The supracleithrum, which extends ven- trally to a point close to the ventral edge of the opercular, has an excavated dorsal end. This excavation does not appear to fit directly onto the ventral edge of the suprascapular, for in several specimens it terminates at a level nearer the midline than does the suprascapular edge (MCZ 12893, 12891, and 12894). It is possible, therefore, that the dorsal end of the pectoral girdle is buttressed, in part at least, against the pos- terolateral wall of the braincase. The donsoventral length of the 10 BREVIORA No. 398 73 0) to V u _o a; c o c £ c en to o m i e c e 5£ o o Oh . X o Si n s btl "= X 1972 TRIASSIC FISH FROM ARGENTINA 11 cleithrum is twice as long as that of the supracleithrum. The cleithrum has a broad \cntral plate, and its anterior border bears an ornament of sinuous rugae. Small cla\icles are pres- ent (MCZ 12893, 12899, 12901).' Paired fins. The paired fins are restored in Figure 4. In both, the rays are jointed so that the proximal lepidotrichia are up to six times as long as the distal ones, while the longest rays are bifurcated at their extreme distal ends ( MCZ 1 2893, 12906). Fringing fulcra were not observed and are probal)ly absent, but in both fins a stout basal fulcral scale participates in the leading edge. The pectoral fin is composed of 1 3 rays, which have between 4 and 12 joints. In well-preserved specimens (for example MCZ 12883, Fig. 2), the bases of the anterior rays are inserted in a more dorsal position than are those of the poste- rior rays. The pcKic fin lies two-thirds of the way between the pectoral and anal fins. It is composed of seven rays, which ha\e between four and eight joints. Unpaired fins. The unpaired fins are restored in Figure 4. In all, the rays are jointed, so that the proximal lepidotrichia are from three to eight times as long as the distal ones. The rays of the anal and caudal fins are bifurcated, and in all the unpaired fins the leading rays bear fringing fulcra. The dorsal fin is composed of 16 rays, which have between three and nine joints, and is preceded by three basal fulcra. The anal fin is composed of 14 rays, which have between 9 and 22 joints, and is preceded by a single basal fulcral scale. The dorsal and anal fins are similar in shape, but the former is larger and is slightly more anterior in position than the anal fin. The caudal fin is hemiheterocercal and the rays of its dorsal lobe are slightly longer than those of its ventral lobe. It is composed of 21 rays, which have between 15 and 24 joints. The four rays, which have their origins on the dorsal side of the caudal lobe of the body, do not bifurcate. There are about six basal fulcral scales on the dorsal side of the caudal lobe of the body; their exact number is difficult to determine because anteriorly they become very similar in form to the ridge scales, which extend posteriorly from the posterior edge of the dorsal fin. In MCZ 12892 and 12905 some details of the articulation of the anal fin rays are seen, and in MCZ 12912 (Fig. 5) similar details are seen in the dorsal fin. In both fins the number of radials equals the number of rays. In the dorsal fin the prox- imal lepidotrichia have expanded proximal ends but, whereas the posterior lepidotrichia are iilmost straight, the anterior ele- 12 BREVIORA No. 398 w o o c s c3 _o c3 o ■*-> CO o K o CO 00 o X 1972 TRIASSIC FISH FROM ARGENTINA 17 bracaccinii, a feature to be expected in the squamation of a more fusiform body. DISCUSSION The Systematic Position of the Genus Pseudobeaconia Before discussing the systematic position of Pseudobeaconia, it is necessary to state why the material described in this paper is considered to represent the genus described by Bordas in 1944. Pseudobeaconia is usually assigned to the order Redfieldiiformes (Gardiner, 1967b), whereas the specimens described here are clearly members of the Perleidiformes, as will be sho\vn below. The original diagnosis of Pseudobeaconia (Bordas, 1944: 454) is based on poorly preser\ed material and is therefore brief. The most important diagnostic features of the genus noted by him are: the absence of branchiostegal rays; the presence of a reduced tabular and postparietal ; a dorsal fin divided into two parts, an anterior part near the head and a posterior part lying in the posterior part of the body; and finally, scales longer dor- soxentralh- than anteroposteriorly. These features are seen in the material described here, or can reasonably be considered as misinterpretations based on poorer material of the same taxa investigated in the present paper. The tabular and postparietal of Bordas correspond to the two extrascapular elements, and the anterior dorsal fin to the row of ridge scales that lie between the head and the dorsal fin. The supposed absence of branch- iostegal rays is probably due to the poor preser\ation of Bordas' material. The similarities discussed so far would not allow cer- tain identification of the present material with Pseudobeaconia. However, Bordas recognized two species of the genus: P. bracaccinii, \vhich was characterized by a quadrangular opercu- lar, a subopercular slightly longer dorso\entrally than antero- posteriorly, and a dorsal fin that was slightly anterior to the level of the anal fin; and P. elegans, which was characterized by a more delicate body shape, an opercular that was less quad- rangular than in P. bracaccinii, and a dorsal fin that was more clearly anterior to the le\el of the anal fin (Bordas, 1944: 455, 456). These difTerences correspond exactly to the differences between the two species recognized in the present material {cf. Figs. 4 and 8 ) , strongly suggesting that we are dealing with the same taxa. The material described here can therefore be as- cribed to the genus Pseudobeaconia with a high degree of con- fidence. This conclusion is strongly supported b)- the fact that 18 BREVIORA No. 398 Bordas* material and most of that described here have been collected from the same locality. Returning to the systematic position of Pseudobeaconia, it is clear that the genus is not a redfieldiiform. Members of the Redfieldiiformes are characterized by a number of features, such as a dermosphenotic that borders the orbital edge, an R-shaped preopercular, an antorbital that borders the nostril edge, and an excess of rays over endoskeletal supports in the unpaired fins (Hutchinson, 1972), none of which are seen in Pseudobeaconia. On the contrary, consideration of the morphology of Pseudo- beaconia leaves no doubt that it is a member of the Colobodon- tidae ( = Perleididae ) , a perleidif orm family well represented in both marine and freshwater Triassic deposits. W^ith t\vo ex- ceptions, Pseudobeaconia has all the diagnostic characters of this family {cj. SchaefTer, 1955: 19). The exceptions are the presence of a suspensorium that slopes forwards, and a rostral that extends dorsally to take part in the nostril border. The first of these characters represents an advanced state of a trend well known in the family, while the second must be regarded as a specialization of the genus. When compared to other members of the Colobodontidae, Pseudobeaconia appears to be similar to Mendocinia (Fig. 9B), which occurs in the same locality (Bordas, 1944; SchaefTer, 1 955 ) , but differs from that genus in having the following characteristics: subdixided extrascapulars; lateral part of skull roof composed of dermopterotic and accessory dermopterotic elements; ventral end of suspensorium inclined anteriorly; op- ercular anterior in position relatixe to skull roof elements; der- mohyal small; two supraorbital elements; maxilla not greatly expanded posteriorly; rostral large and bordering anterior edge of nostril; scales with ornament of ridges; scales of lateral line series indented; unpaired fins with fringing fulcra. Pseudobeaconia is also extremely similar to Praesemionoius from the Lower Triassic of Germany (Jorg, 1969). There are howexer, two characters mentioned in the diagnosis of Prae- semionoius that do not appear in Pseudobeaconia. They are an interopercular and a narrow preopercular that separates the opercular series from the suborbital bones (Jorg, 1969: 89). The presence of tl;iese characters led to the classification of Praesemionoius in the order Parasemionotiformes. Jorg has recently revised this opinion (personal communication), and now bclie\es that there is no interopercular, and that the preop- ercular extends over that part of the cheek prexiously described 1972 TRIASSIC FISH FROM ARGENTINA 19 as being composed of both preopercular and suI)orbital bones. There Ts Uttle doubt that Prncsemionolus is a perlcidiform, and that it should be referred to the famih' Colobodontidae. In conclusion, therefore, Pseudobeaconia is in no way re- lated to the redfieldiiforni Beaconia. It represents the termina- tion (as far as is known at present) of a line of colobodontid perleidiforms. Although its immediate ancestors are not known, they are likely to have been those that also ga\e rise to Men- docinia and Praesemionotus. The Evolution of the Skull in the Perleidiformes The evolution of the holostean feeding mechanism from the palaeoniscoid type is well known. The functional adxantages of the former have been discussed by Schaeffer and Rosen (1961) and by Gardiner (1967b), and reviewed recently by Olson (1971).^ As Schaeffer and Rosen have noted (1961: 203), the modifications that occurred during the palaeoniscoid- holostean transition "permitted a considerable adaptive radiation in the feeding mechanism at the holostean level." However, during the Triassic period, two chondrostean orders, the Red- fieldiiformes and the Perleidiformes, underwent a considerable adaptive radiation of their own, yet neither group achiexed the holostean condition. The Redfieldiiformes owed their success in part to the evolution of a suctorial habit of feeding (Schaef- fer, 1967; Hutchinson, 1972), and it remains to examine the Perleidiformes to see if there are equally obvious reasons for their success. The skulls of four perlcidiform genera are compared in Fig- ure 9. They are not directly related to one another, but they are members of a radiation that must ha\e had a beginning sometime during Upper Permian times. In an earlier radiation, haplolepid genera show parallel evolution of an almost ver- tical suspensorium from an oblique one (VVestoll, 1944) ; in the perleidiforms the skull structure shows even more variation. Three main trends in their evolution can be traced : 1 . The e\olution of an antero\entrally inclined suspensorium from one that is posteroventrally inclined. 2. The shortening of the lower jaw from a length equal to 68 percent of the anteroposterior length of the skull (Perleidus madagascariensis) to one only 55 percent of the length of the skull (Pseudobeaconia bracac- cinii). 3. The ele\ation of the anterior end of the upper jaw. Associated with these changes are alterations in the shape of 20 BREVIORA No. 398 the maxillary-preopercular complex, and a shift of the level of the jaw articulation from a point well behind the orbit to a point below the orbit. The advantage of a suspensorium that is anteroventrally in- clined has been demonstrated by Schaeffer and Rosen; it al- lows greater posterior and lateral expansion of the orobranchial chamber (1961: 191). However, evolution of such a suspen- sorium does not necessarily involve significant shortening of the lower jaw (See, for example, the haplolepids [Westoll, 1944: fig. 44] ) , and the latter phenomenon in perleidids cannot there- fore be regarded simply as a corollary of the former. The advantage of a short lower jaw over a longer one is Figaire 9. Restorations of the skulls and pectoral girdles of rep- resentatives of four perleidifoi-m genera to show variation in shape of the preopercular, maxilla, and dentary. Ornament and sensory canals omitted. Not to scale. A. Perleidus madagascariensis. Lower Triassic, Madagascar. From Lehman, 1952, fig. 85. Mendocinia brevis. Lower Triassic, Argentina. From Schaef- fer, 1955, fig. 4. Cleithrolejndina exto'ni. Lower Triassic, Oi-ange Free State. From Hutchinson, 1972, fig. 41. D. Psenxiobeaconia bracaccinii. Low^er Triassic, Argentina. B. C. 1972 TRIASSIC FISH FROM ARGENTINA 21 that it can exert a greater force when adducted, assuming other factors such as the mass of the adductor muscle are constant. It may be concluded, therefore, that changes in the perlcidid skull architecture are related, not only to the e\'olution of more effi- cient abduction of the orobranchial chamber, but also to the need to exert a powerful force on adduction of the lower jaw. It is not difficult to discover why this last requirement was of great adaptive significance to the perleidids. The group is char- acterized by the presence of low rounded teeth on the palate (Perleidus, ?Colobodus, and Pseudobeaconia) and on the lower jaw (Perleidus, Colobodus, and Aieridensa). These teeth sug- gest that an important element of the perleidid diet consisted of food that required crushing, such as small molluscs and crus- taceans. The third trend in perleidid evolution, the gradual elevation of the anterior end of the upper jaw, is probably related to the evolution of shorter jaws. The disad\antage of short jaws is that the gape is reduced, but this appears to have been com- pensated for by elevation of the anterior end of the maxilla. This brief analysis suggests that an important reason for perleidid success is that the evolution of short jaws and crushing teeth enabled them to feed, in part, on prey not accessible to other forms. This interpretation is not complete, however, for it does not explain the success of the Cleithrolepididae, which have rather slender lower jaws and few teeth ( Fig 9C ) . Mem- bers of this family, although early offshoots from perleidid stock (Hutchinson, 1972), have an advanced type of suspensorium and laterally flattened bodies, and must have occupied a some- what different niche to that of other perleidids. The Structure of THE Unpaired Fins in the Perleidiformes The detailed structure of the unpaired fin rays in Pseudobea- conia shows that control of the fin web area was possible. The unpaired fins of modern selachians are similar to those of chondrosteans in that each radial is associated with a lars^e number of ceratotrichia (lepidotrichia in chondrosteans). In selachians, lateral movement of the fin web is effected by con- traction of radial muscles that originate on the radials and that have broad insertions on either side of the proximal ends of the ceratotrichia. The area of the fin web cannot, however, be altered., and the fins cause considerable drag even when not 22 BREvioRA No. 398 being used to aid stability. The unpaired fins of Acipenser have essentially the same structure and restricted range of mo\ements as have those of selachians (Alexander, 1967: 38), and the same is almost certainly true of the fossil chondrosteans in which the number of rays exceeds the number of radials. In the unpaired fins of teleosts a discrete musculature, which originates on the radials, enables erection, depression, and lat- eral movement of each ray. Such a system is only possible be- cause the numbers of rays and radials are equal. Westoll (1944: 83) has suggested that control of the fin web area was possible in haplolepids because in the Haplolepi- didae the unpaired fin rays are small in number, stout, and bev- eled along their anterior edges to facilitate overlapping when the web is collapsed. In Pseudobeaconia there is even more evi- dence which suggests that in that genus, and possibly in all 'sub- holosteans' in which the unpaired fin rays and radials are equal in number, the fins were capable of almost as much mobility as is seen in teleosts. The dorsal fin of P. bracaccinii has been described above ( Fig. 5 ) . The rays and radials are equal in number, a condi- tion which suggests that a discrete musculature between the two was present. This is confirmed by the presence of an anterior and a posterior depression at the proximal end of each ray. The anterior depression is here interpreted as the point of in- sertion of an erector muscle, and the posterior depression as that of a depressor muscle. Anteriorly, the erector muscle insertions become progressively larger and the depressor insertions pro- gressively smaller. Because of their position and greater length, the anterior rays are more affected by water forces that tend to collapse the fin. As a result, more effort is required to erect, and less to depress, these anterior rays, and these requirements are reflected by the variation in size of the erector and depres- sor muscle insertions. There is no sign of any depression \vhich would indicate the presence of an inclinator muscle, but inclina- tion of the fin is theoretically possible if the erector and de- pressor muscles of one side are contracted simultaneously. Such a mechanism does in fact occur in modern Siluriformes (Alex- ander, 1967: 40), but cannot be confirmed in Pseudobeaconia because the state of preservation of the joint between each lepi- dotrichium and radial does not allow an accurate assessment of the degrees of freedom of the ray. The grooves that run along the anterior and posterior edges of the rays in P. bracaccinii (Fig. 5) probably helped to sup- port the soft fin web tissues. 1972 TRIASSIC FISH FROM ARGENTINA 23 ACKNOWLEDGMENTS I am grateful to Dr. C. Barry Cox of the Department of Zoology, University of London, King's College, for the loan of equipment which made this work possible and for reading and criticizing this paper; also to Maite Lores, who translated many of the references and typed the manuscript. My best thanks, howe\er, are due to Professor Alfred S. Romer, who kindly made available the material described in this paper. REFERENCES Alexander, R. McN. 1967. Functional Design In Fishes. London: Hutchinson University Library. 160 pp. BONAP.^RTE, J. F. 1966a. Chronolog-ical survey of the tetrapod- bearing Triassic of Argentina. Breviora, No. 251 : 1-13. . 1966b. Una nueva "fauna" Triasica de Argen- tina (Therapsida: Cynodontia: Dicynodontia). Consideraciones filogeneticas y paleobiogeograficas. Ameghiniana, 4: 243-296. 1969. Cynognathus minor n. sp. (Therapsida: Cynodontia). Nueva evidencia de la vinculacion faunistica Afro- Sudamericana a principios del Triasico. In Gondwanaland Stratigraphy. LU.G.S. Symposium, Buenos Aires, 1967. Paris, UNESCO, pp. 273-281. BORDAS, A. F. 1944. Feces triasicos de la quebrada de Santa Clara (Mendoza y San Juan). Physis, 19: 454-460. Gardiner, B. G. 1967a. Further notes on palaeoniscoid fishes with a classification of the Chondrostei. Bull. Brit. Mus. (Nat. Hist.), Geology, 14: 143-206. 1967b. The significance of the preoperculum in actinopterygian evolution. J. Linn. Soc. (Zoology), 47: 197-207. Harrington. H. J. Descripcion geologica de la Hoja 22c — Ram- blon Provincias de Mendoza y San Juan. Unpublished MS: 1-62. Hutchinson, P. 1972. A revision of the redfieldiiform and per- leidiform fishes from the Triassic of Bekker's Kraal (South Africa) and Broolcvale (New South Wales). Bull. Brit, Mus, (Nat. Hist.), Geology, 22: 197-318. JoRG, E. 1969. Eine Fischfauna aus dem Oberen Buntsandstein (Unter-Trias) von Karlsruhe-Durlach (Nordbaden). Beitr. Naturk. Forsch. SudwDtl., 28: 87-102. Lehman, J. -P, 1952. fitude complementaire des poissons de I'Eo- trias de Madagascar. K. Svenska Vetensk. Akad. Handl., 2: 1-201, MiNOPRio, J. L. 1954. Theriodonte en el Triasico de Mendoza. An. Soc. Cient, Argentina, 157: 31-37. Olson, E. C. 1971. Vertebrate Palaeozoology. New York: Wiley- Interscience, 899 pp. 24 BREVIORA No. 398 ROMER, A. S. 1966. The Chanares (Argentina) Triassic reptile fauna. I. Introduction. Breviora, No. 247: 1-14. SCHAEFFER, B. 1955. Mendochiia, a subholostean fish from the Triassic of Argentina. Am. Mus. Novitates, No. 1737: 1-23. . 1967. Late Triassic fishes frojn the Western United States. Bull. Am. Mus. Nat. Hist, 135: 285-342. . , AND Rosen, D. E. 1961. Major adaptive levels in the evolution of the actinopterygian feeding mechanism. Am, Zool., 1: 187-204. Stipanicic, p. N. 1967. Triasico. 1st Internat, Symposium Gond- wana Stratigraphy Palaeontology. Mar del Plata, 1967. 9 pp. Wade, R. T. 1935. The Triassic Fishes of Brookvale, New South Wales. London: British Museum (Nat. Hist.). 110 pp. Westoll, T. S. 1944. The Haplolepidae, a new family of late Carboniferous bony fishes. A study in taxonomy and evolution. Bull. Am. Mus. Nat. Hist., 83: 1-122. ABREVIATIONS aop accessory opercular adp accessory dermopterotic an angular ant antorbital b branchial arch element br branchiostegal ray el cleithrum clav clavicle d dermohyal de insertion of depressor muscle den dentary dp dermopterotic ds dennosphenotic ecomm ethmoid commissure er insertion of erector muscle ex extrascapular f fulcral scale fr frontal g groove inf infraorbital (1) left side mx maxilla na nasal no nostril border op opercular pa parietal par parasphenoid pmx premaxilla po postorbital pop preopercular pro postrostral pspi postspiracular q.j quadratojugal r radial ro rostral sb suborbital so suprascapular scl supracleithrum so supraorbital sop subopercular B R E V I O R A/ Museom of Comparative Zoology Cambridge. Mass March 6, 1973 Number 399 THE E\ OLUTION OF MAMMALIAN FROM REPTILIAN DENTITIONS J. ^V. OSBORN^ AND A. W. CrOMPTON^ Abstract: A motlel to explain the evolution of a typical mammalian tooth replacement pattern is proposed. Osborn (1971) has described the early development of the teeth and replacement pattern in a typical modem reptile (Lacerta vivipara). The re- placement of the postcanine teeth in an early Triassic mammal-like reptile, a cynodont, Thrinaxodo'n liorhinus, is basically the same as that of Lacerta vivipara except that tJie number of replacements closely approaches the mammalian condition. The rate of replace- ment has slowed down considerably and towards the back of the jaw only three teeth develop at each locus. Three different types of postcanine teeth can be recognized, an A tjV^ with a simple crown pattern towards the front of the row, an M type in the midsection with a more complex pattern, and a P tyi^e with the most complex pattern towards the back of the row. A P type is replaced by an M type and the latter by an A tjTpe. The pro^essive reduction in the complexity of the teeth produced at each locus in Thrinaxodon and in mammals (deciduous molar as compared with the premolar replacing it) is ascribed to the aging of the dental lamina. The typical mammalian pattern can be derived from that of Thrinaxo- don by further reducing the rate of replacement. INTRODUCTION Two theories ha\'e been proposed to explain the e\okition of mammalian dentitions, in which teeth are only replaced once (diphyodonty), from reptilian dentitions in which teeth are replaced throughout life (polyphyodonty), Bolk (1922) con- cluded that reptilian dentitions consist of two sets of alternat- 'Anatomy Department, Guy's Hospital Medical School, London SEl, 9RT, England 'Museum of Comparative Zoology, Harvard University, Cambridge, Massachusetts BREVIORA No. 399 Ai eje|e|ejejej.eieAeAe|e • •••••••oo ooooooooooo oooooooooo All eeeeeeeeeee Bi o o e °-°»:» Bi eeeeeeeeeee Ci Cii '/ '^ 'j C Dm/ Dm^ Dnrij Dm^ M/ M^ M^ h h h c Pm^ Pnrij Pm^ FigTire 1. Ai represents Bolk's concept of the sets of alternat- ing teeth present in a reptile. All members of a set develop at the same time. Aii shows what he considered to be the homologies be- tween polyphyodont reptilian and diphyodont mammalian dentitions. Bi (reptile) and Bii (mammal) illustrate Edmund's (1960, 1962) concepts. He suggested that teeth develop in the sequence showTi by the arrows. (Time is represented by a vertical axis.) Ci shows the actual sequence of tooth development in a primitive reptilian dentition (note alternation of teeth). Cii (arrows) shows the actual sequence in mammalian postcanine dentitions (Osborn, 1970). In all diagrams anterior is to the left and the mouth above. 1972 MAMMALIAN DENTITIONS 3 ing teeth, each set being replaced alternately throughout life (Fig. lAi). He suggested that parts of a mammalian dentition is equi\'alent to one set that has become pushed together and that the replacing teeth are equivalent to the other alternat- ing set (Fig. lAii). Edmund (1960, 1962) concluded that reptilian teeth are initiated in sequence from the front to the back of the jaw in rows which he called Zahnreihen (Fig. IBi). He suggested that the two mammalian dentitions are equiva- lent to two reptiHan Zahnreihen (Fig. IBii) . It has recently been demonstrated (Osborn, 1971) that in an embryo reptile teeth are initiated in alternating sequences from the back to the front of the jaw (Fig. ICi), while in the postcanine dentitions of embryo mammals (Osborn, 1970), the first (most anterior) deciduous molar is always the last of the deciduous molars to develop and only the permanent molars develop in sequence from the front to the back of the jaw (Fig. ICii). We conclude that neither Bolk's nor Edmund's theory is acceptable because neither takes into account the sequences in which teeth actually de\elop. One other feature clearly distinguishes between the dentitions of mammals and primitive reptiles: in mammahan dentitions, the postcanine teeth generally have difTerent shapes (hetero- donty) whose complexity reaches a peak somewhere within the permanent molar series; primitive reptilian teeth are generally all peg-shaped and alike (homodonty). We are concerned in this paper with the phylogenetic process by which the heterodont, diphyodont dentitions of mammals e\olved from those of the homodont, polyphyodont dentitions of primitive reptiles. In the absence of any recent animal which has a dentition e\en remotely intermediate between these two extremes, a solution to these problems must either be entirelv speculative (Bolk, 1922; Edmund, 1960, 1962) or based on e\'idence from the fossil record. The gap between mammals and reptiles was bridged by the mammal-like reptiles of the Triassic. For our study we have chosen Thrinaxodon, whose tooth morphology and pattern of tooth replacement have al- ready been investigated (Crompton, 1963; Parrington, 1936), but not in sufficient detail to answer the problems which we have posed. POST-CANINE TEETH OF THRINAXODON Preliminary Arrangement of Material For the present in\estigation ten extremely well-preser\ed, as yet undescribed, Thrinaxodon jaws have been studied, to- 4 BREVIORA No. 399 gether with four of those described by Crompton ( 1 963 ) , two of which were available to Parrington ( 1936) . These have been labelled A to N. The investigation was confined to the lower postcanine teeth. In order to arrange these specimens in a growth series, photo- graphs, reproduced at the same magnification, were taken of the lingual sides of the jaws. Tracings were made of the jaw outlines of all the available material. These tracings were com- pared by superimposing one on another. It was necessary to speculate on the growth processes of the mandible to make ac- curate comparisons between the positions of the postcanine den- titions in jaws of different sizes. These speculations will be described later. On the basis of this study we arranged the specimens in a tentative growth sequence. We speculated that the majority of growth in length of the jaw took place at the posterior end of the dentary where room was being made for the addition of postcanine teeth (Crompton, 1963) and that there was only a hmited subperiosteal deposi- tion of bone at the anterior end of the jaw (Fig. 2, cf. Fig. 7). From this we now concluded that successive replacement incisors remained in approximately the same anteroposterior position in relation to a hypothetical fixed marker, or that they erupted in only a slightly more anterior position than their predecessors. Successive canines might have become more posteriorly located Figure 2. A diagTam in which growth processes of the lower jaw of Thrinaxodon are postulated. Uninternapted line = younger animal; interrupted line = older animal. The jaw has increased in length by the deposition of bone at the borders of the dentary. The anterior border of the coronoid process has been resorbed, making room for the addition of anotlier postcanine in the older animal. There has been very little deposition of bone at the anterior end of the jaw. The most anterior of the post^-anines in the younger ani- mal has been lost in the older ani^nal. 1972 MAMMALIAN DENTITIONS because we obserxed that the tip of a replacing canine was gen- erally located posterior to the functional canine. However, we finally decided that as a canine erupted it moved into a more anterior position than would ha\ e been predicted from the posi- tion of its erupting tip ( Fig. 7 ) . The tentative growth sequence established in this way was modified following a study of the shapes of the teeth and the positions of replacing teeth. For this purpose the sections of the prints containing the postcanine teeth were cut from full prints of the jaws. These small prints were rearranged until the best correspondence between adjacent dentitions was achieved, bear- ing in mind the original tentative arrangement based on jaw sizes. Tooth Shape in Thrinaxodon It appeared that the preliminary arrangement was in many cases inaccurate because frequently there was poor correspond- ence between what initially appeared to have been equivalent teeth in jaws which had been judged as being of comparable age ( based on the similarity of their sizes ) . By comparing tooth shapes and by making estimates of the ages of teeth based on Crompton's (1963) observations, a definitive growth sequence was established. Before presenting this we will briefly describe the shapes of the teeth. We finally established that Thrinaxodon possessed only three distinctive postcanine tooth types. It was because variations of each tooth type were present that Crompton (1963) did not recognise the division into three tooth types. W^e will refer to these as A = anterior, M = middle, and P =r. posterior. Be- tween two and four A types (Figs. 3, 4) are always present at LeftE Right B LeftE LeftE LeftE LeftB Right B position4 positions positions position? positions position? positions M types P types A types Figure 3, The three tooth types defined in this study. P = pos- terior, M = middle, A =r anterior. More detailed descriptions of the postcanines in Thrinaxodon are given by Crompton (1963). 6 BREvioRA No. 399 the front of the postcanine row. The crown is dominated by the main cusp and flanked by very small anterior and posterior accessory cusps. Lingual cingulum cusps are poorly developed or absent. Frequently the main cusp is markedly tilted poste- riori)-. The most anterior A type is often extremely small (Fig. 4). Either two or three M types (Figs. 3, 4) are always present in the middle of the postcanine row. The main cusp dominates the tooth but accessory cusps are usually situated much higher up its flanks than in the A type, reducing its apparent height when compared with the equivalent cusp on the A types. Cingu- lum cusps are generally well developed, producing a frill around the lingual surface of the tooth. The postcanine dentition ter- minates with either two or three P types (Figs. 3, 4) (dentition E contained only one ) . Included with these in the older animals may be found a "hybrid" MP type situated at the back of the dentition {e.g., H, I, J, M in Fig. 4). These MP types will be considered with the P types because they are present at the back of the jaw. Their significance will be discussed later. In the P types the main cusp is very much reduced in height by large anterior and posterior accessory cusps situated high up on its flanks. Well-developed anterior and posterior cingulum cusps are usually placed at the ends of the tooth in such a way as considerably to increase its anteroposterior proportions when compared with A and M types. A well-de\eloped array of lingual cingulum cusps gives the tooth what approximates to an occlusal surface although it did not occlude with the upper teeth. The occasional MP type has, as might be expected from the terminology, a shape intermediate between M and P types. From the above definitions it can be visualised that the sizes of the anterior and posterior accessory cusps 1 ) increase from the front to the back of the dentition and 2) are directly re- lated to the prominence of the Ungual frill of cusps. In all the present material and in all well-preserved material which we have subsequently examined the postcanine dentition begins with be- tween two and four A types which can be clearly distinguished from the succeeding two or three M types. These M types can be clearly distinguished from the P types which terminate the dentition. Figure 4. Fifteen postcanine quadrants described in this study arrang-ed in the suggested age sequence. The prints of the left jaws were reversed to simplify comparisons between dentitions. A and P types are shaded. Developing teeth are shown in the appropriate positions. 1972 MAMMALIAN DENTITIONS 1 i2 3 1 4 5 I 6 I 7 I 8 I 9 « BREVIORA No. 399 The Definitive Arrangement of Specimens We finally decided on the sequence of dentitions shown in Figure 4. This can be compared with Figure 5 in which each postcanine has been allotted a tooth form A, M, or P and Fig- ure 6, which is a simplified version of Figure 5. Mandibular sizes are shown in Figure 7. 1 2 3 4 5 6 7 8 9 10 11 12 A B C D E F G H I (Right) J K L (Left) M N a a [a] rri m P. -^ ? ?A ? a rriA m PA p ^ -^ -^^ a 7 7a ? 7. 7 A a aA ^ a X X X X X X 7 a^ a m^ m p^ p a^ a^m m M m. m .fiA m Pa E a a. rriA m P P a -^m. rn p. p x^ a 7 _ a a^ ^ X a aA a mA ■? 7 ? 7 m? X X m^ m p^ p m^ m^p A YA ~ m^p m a nrij^ m m X B o JDAJILm n. _B_ aA aA a mA XA Pa P m p^ m ■^p • 1972 MAMMALIAN DENTITIONS 9 In support of this sequence the tooth types present in each dentition in positions 4 to 9 should be studied from Figures 4 and 5, Position 4 : The M type of dentition A is replaced in denti- tion B b\ an A type. This A type is replaced twice b>- further A types. Only a root is present in dentitions M and N. Position 5 : The M type of dentitions A and B is replaced in C by an A type, which is replaced once by another A type. Position 6: The P type of dentition A is replaced in B by an M type. This tooth can be traced through to G being re- placed in I by an A type, which can be recognised in the remain- ing dentitions. Position 7: The P type of dentitions A and B is replaced in C by an M type, which can be recognised through to denti- tion L. On one side of dentition M and probably of N it has been replaced by an A t)pe tooth. Position 8: The P type in dentitions A to D is replaced in E by an M t\pe. This M t\pe can be recognised through to the last dentition. The small size of the P type in dentitions A, B, and C will be referred to later. Position 9: In C a P type has erupted in the equivalent alveolus present in B. This P type can be recognised up to den- tition K. In dentitions L, M, and N it has been replaced by an M type tooth. Figure 5. The dentitions of all the specimens studied are rep- resented in the form of an ag-e sequence. Specimen identifications are to the left and tooth positions above. Where the teeth on left and right sides of a dentition differed the right side is represented above a line and the left side below the line. X represents a badly mutilated or lost tooth in an intact region of the jaw. The query marks represent teeth that were probably present in regions from which the jaw was missing. The triangles represent developing t«eth, the ovals are the crypts which contained them. The origin of the material is as follows: A =r TM 80; B = TM ?; C = TM SOB; D = BMNH R3731 = Crompton's specimen C := Pai-ring-ton's spec- imen E; E = NM C354; F = SAM; G = NM C318; H = AMNH; I = SAM K377 = Crompton's specimen D; J = TM 180; K = BMNH R511 =: Crompton's specimeii F; L = TM 81; M = SAM ?; N = BMNH R511A = Crompton's specimen H = Parrington's specimen I. TM = Transvaal Museum; BMNH = British Museum (Natural History); SAM = South African Museum; NM ^ Na- tional Museum (Bloemfontein) ; AMNH = American Museum of Natural History. 10 BREVioRA No. 399 In each of the above sequences it is important to recognise that correct aging of teeth has been observed. Consider position 7. The young P type in dentition A is old in B. The replacing M type in C is young (there is no replacing crypt). It is not until dentition H that a crypt is observed beneath this now aging M type tooth. On one side of dentition M (Fig. 5) and probably of N (Fig. 4) it is replaced by an erupting A type. Tooth Replacement in Thrinaxodon Before discussing the present data we must point out that it is by no means certain that all the animals studied belonged to a homogeneous population. The discrepancies between denti- tions, particularly the smallness of the P types in position 8 of dentition A, position 9 of C, D, and E, position 10 of G, and position 1 1 of I and J could represent terminal stages in the growth series of animals of different sex or species. Nevertheless we will treat the data as if the animals belonged to a homoge- neous population. If the growth series presented here is correct then the pattern of tooth replacement in the heterodont postcanine dentition of Thrinaxodon conforms precisely with that described by Ed- mund (1960) for the majority of the homodont dentitions of nonmammalian vertebrates. This is shown in Figure 8B in which the numbering of the tooth positions corresponds to that in Figures 4, 5 and 6. By adding two further positions on the left (Po and P-i) we calculate that it is possible to fit Cromp- ton's (1963) very small specimen A, in which only the maxil- lary teeth are visible, into the series. In most reptiles the actual number of tooth replacements at each tooth position is determined first by the longevity of the animal and second by the rate at which replacement teeth are initiated. Compared with other reptiles the rate of replacement of postcanine teeth in Thrinaxodon appears to have been so re- duced in proportion to its longevity that posterior to position 6 at most only three teeth occupy each position. The number of replacement teeth in the anterior postcanine positions are limited in the following way. The replacing canines developed very close to the roots of the most anterior A types (Hopson, 1964). Hopson (ibid.) suggested that the presence of the developing canine prevented the development of replacements for these A types. The result was that even in the anterior postcanine posi- tions, at most fi\'e teeth (position 4) erupted in any tooth posi- 1972 MAMMALIAN DENTITIONS 11 1 2 3 4 5 6 7 8 9 10 11 12 A Right a a^ ^ ^a '^ Pa P fP B Left C Right *1 D Right E Left F Left*2 G Right H Left I Right J Left*3 K Right L Left M Left*4 N Right *5 a a, m^ m p^ p • a a^ a a m^ m p^ p X a a^ a m^ m p^ p X aj^ a m^ m [mj Pa ^ X a a m^ m [m] p^ ^ X a amj^m[m]p p X a^ a [aj m^ m p^ mp m^?m Pa P [p X x^ a a a^ a^ [aj ^A f^ Pa P EE X a aA a rriA m Pa p ^ Pa P a a a n\ m m X a^ a X a. m^ m p^ mp aj Xa m Pa P *1 2-4 left *2 Position 10 right *3 Position 9 right *4 Position 11 right *5 Positbn 10 left Figure 6. The information in Fig. 4 is simplified in this dia- gram. Where a tooth is missing or unrecognisable on one side of the jaw the corresponding tooth on the intact side of the dentition is rep- resented. In the few cases where the left and right dentitions of an animal are different the tooth form that more closely fits the predicted sequence is shown. A box indicates that the tooth was not yet fully erupted. A triangle represents an erupting tooth, an oval represents an alveolus. 12 BREVIORA No. 399 Figure 7. Tracings of the jaws and postcanine dentitions from one side of all specimens except I, M, and N for which suitable photogi-aphs were not available. The tracings have been arranged in accordance with the growth processes suggested in Fig. 1. tion. Thus, although Thrinaxodon was strictly polyphyodont (because the dental lamina was active in budding new teeth through the life of the animal) the number of replacement teeth was severely limited when compared with most modern reptiles which have been studied. The following argument suggests that Thrinaxodon may have been evolving towards a condition in which the activity of the dental lamina terminated during the life of the animal. This is the characteristic which distinguishes polyphyodont from other dentitions. In the anterior postcanine tooth positions the last replacement teeth were so small (position 3 in dentitions D to K; position 4 in dentitions L to N) that they can hardly have had any functional value. Thus, it can be argued that, despite its continued presence, the dental lamina in these positions no 1972 MAMMALIAN DENTITIONS 13 longer contributed significantly towards the maintenance of an efficient dentition. This is likely to be the first stage in the evo- lution of non-polyphyodont {e.g., diphyodont) from polyphyo- dont dentitions. The same type of process appears to have taken place at the back of the dentition. Instead of bulky P type teeth, xery re- duced hybrid MP types were erupting at the back of dentitions H, I, and J: once more, the dental lamina seems to be showing a reduced activity in the adult animal. In conclusion, although Thrinaxodon was undoubtedly polyphyodont, the adult ani- mal shows a tendency toward producing small teeth of lim- ited value at the front and back of the dentition. It is as if the dental lamina was becoming senescent during vigorous adult life, a feature which is not typical of a polyphyodont dentition. As described in the next section, this possibility may have impor- tant implications for determining the shapes of successional teeth. Tooth Morphology in Thrinaxodon It will be observed that in each tooth family there is a pro- gressive simplification of cusp detail starting with the complex P type and ending with the very simple A types. This was ac- counted for by Crompton (1963) in the following way. Basing his argument on the concept of embryological fields ( Butler, 1939), he proposed that several morphogenetic fields moved anteroposteriorly along the jaws of Thrinaxodon. In terms of the present observations he suggested that initially a "P type" field traxelled slowly backwards through the jaw. Under its influence the first tooth in each tooth position developed into a P type. Following closely behind this was an "M type" field which induced the second tooth in each tooth position to de- velop into a M type. In turn this was followed by two or three A type fields. However, there is a more simple and biologically probable explanation for the present observations. We speculate that in each tooth position there was a gradual decrease in the morphogenetic capacity of the dental lamina and that this was an aging phenomenon. A newly differentiated "segment" of the dental lamina was capable of inducing the development of a complex P type tooth. This same "segment" of dental lamina now aged in such a way that the second tooth it produced was the less complex M type tooth. If time allowed, further aging resulted in its producing successively less complex A types. This 14 BREVIORA No. 399 A I 3 5 2 4 6 I 3 5 2 4 6 I 3 5 2 4 6 7 9 \ 8 10 \ 7 9 II ^ 8 10 12 7 9 II 13 8 10 12 14 7 8 9 10 B c Hatchling Theoretical Dentition A a./ 0/ Aj M5 Py Op A 4 M 5 Pg Qj A5 M^ Pg O4 ^6 ^8 P/0 M. " Dentition M Figure 8. "A" shows the sequence in which teeth develop in a primitive reptile (Osbom, 1971). Teeth equivalent to those at the right of the interrupted line may have evolved into postcanines in ThHnaxodon (B). A mammalian dentition may be represented by tlie teeth enclosed within the box in B. Equivalent teeth are given mamjnalian nomenclatures in C. 1972 MAMMALIAN DENTITIONS 15 same sequence of aging was reproduced sequentially in each segment of dental lamina as it extended backwards through the jaw into each new tooth position. In other words, heterodonty in the postcanine dentition of Thrinaxodon was achieved by a genetic process linked to the phenomena of aging rather than to the diametrically opposed phenomena of embryology (typified amongst other things by morphogenetic fields). Such a hypo- thesis could have considerable e\olutionary significance, for it will be noted that if the postulated senescence of the dental lamina were to be accelerated, a stage would be reached at which the dental lamina was no longer capable of initiating a new tooth. Such a dentition ceases to be polyphyodont and suggests a possible origin for diphyodonty. It is now necessary to allot a functional significance to the waning of the potential of the dental lamina to induce the de- velopment of cusps in the replacement teeth of Thrinaxodon. Newly hatched reptiles require a functional dentition. Thus even in the youngest animals strongly cusped teeth (relative to the size of the animal) are required at the back of the jaw where maximum power is developed. The teeth in positions 3 and 4 were probably at the back of the postcanine dentition in the new-born Thrinaxodon (Fig. 8B). Thus P types would initially have been present here and throughout life new P types would have erupted in sequence at the back of the jaw. But a neonatal Thrinaxodon jaw could only accommodate a relatively short and fragile P type tooth in position 4 in comparison to the size of the P type tooth which could be accommodated in an adult jaw. This same restricted space was a\"ailable for the teeth which subsequently occupied position 4 in older animals. It is not difficult to visualise that it was an advantage to fill this space with a proportionately shorter but more robust M type tooth than a proportionately longer and more fragile P type. This same space was sufficient to accommodate an e\en more robust A type tooth in the adult animal. Thus, by reference to Figure 4, it can be seen that in position 6 the A type of denti- tion L has roughly the same length as the M type of dentition E and the P type of dentition A. However, the final A types in positions 3, 4, and 5 were very diminuti\e teeth which had little functional significance. The most significant feature of the abo\e hypothesis in rela- tion to the ontogeny of postcanine teeth in Thrinaxodon is the suggestion that the successive tooth types in each tooth position were developed in relation to a form-determining stimulus which 16 BREVIORA No. 399 was repeated in every section of the dental lamina (or adjacent ectomesenchyme ) . No sequential stimulus travelling between adjacent segments of the dental lamina, such as that inherent in the embryological concept of a morphogenetic field, is required. The same difference in interpretation separates the hypotheses of Edmund (1960, 1962) and Osborn (1970, 1971) which have been put forward to explain the manner in which wave replace- ment of teeth develops in lower vertebrates. Edmund (1960, 1962) suggests that the timing of tooth initiation from the free margin of the dental lamina is co-ordinated by an impulse that travels backwards through the length of the jaw {cf. a morpho- genetic gradient travelling backwards along the length of the jaw) . By modifying Edmund's hypothesis Osborn (1970, 1971 ) explains wave replacement in terms of a control which is localised to each tooth position and which is repeated in every tooth posi- tion ( cf. the P, M, and A types repeated in each tooth position ) . The absence of any evidence of a caudad impulse or a morpho- genetic gradient (in postcanine dentitions) suggest that, in terms of biological possibilities, the more simple explanations offered here are more probable. THE EVOLUTION OF MAMMALIAN DENTITIONS It is now possible to speculate on the evolution of mamma- lian postcanine dentitions from polyphyodont dentitions (Fig. 8 ) . We speculate that dental ontogeny in a primiti\e early rep- tile was hke that described for a modern reptile (Osborn, 1971 ) which retains the archetypal pattern of tooth replacement ob- served in fossil reptiles (Fig. 8A). In this figure a canine might have evolved at position 6; therefore we are only concerned with positions 7 and beyond (the postcanine teeth). We speculate that in Thrinaxodon (Fig. 8B) postcanine development might have begun at about position 9. Anterior to this, just as in the primitive reptile, teeth could have developed alternately from back to front although there is no evidence for this. The tooth at position 8 developed later than that at positions 9 and 7, and owing to aging of the dental lamina, may therefore have been an M type rather than a P type. Behind position 9, teeth de- veloped in sequence from front to back, just as in the primitive reptile. The first teeth at each tooth position were the complex P types. It can be seen that the sequence of tooth development in Thrinaxodon may have been similar to that in a primitive reptile. 1972 MAMMALIAN DENTITIONS 17 We now speculate that a primithe mammalian postcanine dentition with four deciduous molars, four premolars, and three molars was equi\alent to the teeth within the box in Fig- ure 8B. Equi\alcnt teeth ha\"e been gixen mammalian nomen- clatures in Figure 8C. The sequence of de\elopment of decid- uous molars and premolars is from back to front. This sequence matches data which show that Dm 4 is usually the first and Dm 1 is always the last to develop (Osborn, 1970). Furthermore these teeth decrease in complexity from back to front just as in the box shown in Figure 8B. The Dm 1 in mammals appears ne\er to be replaced. This also is a trend which can be seen in the equivalent tooth position in Thrinaxodon. The most molari- form teeth in mammals develop in sequence from front to back; again this agrees with the model in Figure SB. Finally, the pre- molars are generally less complex than the deciduous molars and this too matches our model ( Fig. SB) . SUMMARY In a primiti\e reptilian embryo the first tooth in each jaw quadrant develops at about the ninth tooth position from the front of the jaw. In front of this position teeth dexelop in the odd-numbered positions from the back to the front (9, 7, 5, 3, 1 ) to be followed by a similar sequence in the even-numbered positions (8, 6, 4, 2). Behind this position teeth develop in sequence from front to back (9, 10, 11, 12, etc.). Teeth con- tinue to be replaced throughout life. We ha\e confirmed that the pattern of tooth replacement in the postcanine dentition of the mammal-like reptile, Thrinaxo- don, was similar to that seen in primiti\e reptiles. We specu- late that the sequence of tooth dexelopment in embryos was also similar to that in a primiti\e reptile. However, the rate of tooth replacement was so much slower that at most fi\e, and usually fewer, replacement teeth were developed at each tooth position. It has been shown (Osborn, 1971) that in primitive mam- mals either the fourth or third deciduous molar is the first to develop and the 1st deciduous molar is the last to develop (i.e., development is from back to front as in primitive reptiles and Thrinaxodon) . Behind the fourth deciduous molar, teeth de- velop in sequence from front to back (again as in the reptiles). Thus, by further slowing down the rate of tooth replacement in Thrinaxodon, we arrive at the peculiar sequence of tooth initi- ation which has been obser\ed in primitive mammals. 18 BREVIORA No. 399 In Thrinaxodon the first tooth to develop in each postcanine tooth position was the most complex, to be follwed by replacing teeth which were necessarily less complex. We speculate that this sequence was related to aging of the dental lamina, the epithelial ingrowth from which all teeth develop, and further- more, that the same is true for mammals. This theory explains why both the deciduous molars and premolars become progres- sively less complex from back to front; it is the sequence in which they develop. Premolars are less complex than the teeth they replace (the deciduous molars) because they develop later. Permanent molars are complex because they are the first teeth to develop at their respective tooth positions. REFERENCES BoLK, L. 1922. Odontological essays. 5. On the relation between reptilian and nianimalian dentition. J. Anat., 57: .55-75. Butler, P. M. 1939. Studies of the mammalian dentition. Dif- ferentiation of the i:)Ost-canine dentition. Proc. Zool. Soc. Lon- don (B), 109: 1-36. Crompton, a. W. 1963. Tooth replacement in the cynodont Thnn- axodon liorhinus Seeley. Ann, South African Mus., Cape To%\ai, 46: 479-521. Edmund, A. G. 1960. Tooth replacement in the lower vertebrates. Roy. Ontario Mas., Life Sci. Contrib., 52: 1-190. 1962. Sequence and rate of tooth replacement in the Crocodilia. Roy. Ontario Mus., Life Sci. Contrib., 56: 1-42. HOPSON, J. A. 1964. Tooth replacement in cynodonts, dicynodont and therocephalian reptiles. Proc. Zool. Soc. London, 142: 625- 654. OsBORN, J. W. 1970. New approach to Zahnreihen. Nature (Lon- don), 225: 343-346. 1971. The ontogeny of tooth succession in Lacerfn vivipara Jacquin (1787). Proc. R. Soc. London, B, 179: 261-289. Parrington, F, R. 1936. On tooth replacement in the theriodont reptiles. Phil. Trans. Roy. Soc. London (B), 226: 121-142. ''"^'-'c:. CCMP. 200L BREVIORA IMoseom of Comparative Zoology Cambridge, Mass. March 6, 1973 Number 400 THE CAROLINA SALT MARSH SNAKE: A DISTINCT FORM OF A^^ TRIX SIPEDON Roger Conant and James D. Lazell, Jr. Abstract. Hybridization between two species and intergradation between races of one of them has long obscured the nomenclatural status of the water snakes of the Natrix sipedon-fasciata group in eastern North Carolina. A study of recently acquired material and the recognition that hybridization is occurring in several localities where fresh- and brackish-water habitats meet make possible a clari- fication of the confusing engelsi situation. Because the name engelsi was inadvertently applied to a snake of the wrong species, a tax- onomic adjustment is made and the name Natrix sipedon william- engelsi is erected for the salt marsh snake that occurs on several of the islands of North Carolina's Outer Banks and along the mainland perimeter of Pamlico Sound and associated estuaries. INTRODUCTION The confusing status of the water snakes of the Natrix sipedon complex from the Outer Banks of North Carolina and the ad- jacent mainland has long intrigued both of us. Conant (1963: 29-32) summarized much of the data available at that time under the heading of "brackish-water populations." Lazell, as part of studies on the biota of the Banks, has recently accumu- lated considerable fresh material that helps to throw light on the situation. We ha\'e deemed it advisable to pool our knowledge and resolve the "engelsi" problem, which we review in some de- tail. Three recognizable forms of the Natrix sipedon-fasciata group occur in eastern North Carolina. Natrix sipedon and Natrix fasciata are distinct species, based on their wide sympatry in the Mississippi Valley, along the Gulf Coast, and in northeastern North Carolina (Conant, ibid.). In the Carolinas fasciata is confined almost exclusi\'ely to freshwater habitats on the Coastal 2 BREVIORA No. 400 Plain ; sipedon occurs in the mountains and Piedmont, but it also invades the Coastal Plain north of Albemarle Sound, occurs on the peninsula between Albemarle and Pamlico Sounds, enters and is quite tolerant of brackish-water conditions bordering Pam- Uco and Core Sounds, and also occurs on the islands of the Outer Banks. In much of this general area it is differentiated at the subspecific level, thus providing the third form of the complex from eastern North Carolina. That pecuhar specimens of Matrix sipedon occur on the Outer Banks is a view scarcely novel with us. Barbour ( 1 943 ) described Matrix sipedon engelsi on the basis of a single individual taken from Mullet Pond on the Shackleford Banks. Mullet Pond, which was formerly a bay on the north shore of the island (Engels, 1952: 704), now has no connection with salt water and is fresh; it supported a dense stand of cattails (Typha) and there was very little open water when a field party of the American So- ciety of Ichthyologists and Herpetologists visited it on 5 Septem- ber 1964. The snake Barbour used for his type (MCZ 46688) unfortunately is identifiable as Matrix jasciata. We have exam- ined this specimen both independently and together, and we agree that from all diagnostic morphological characters it is a jasciata. The name engelsi, therefore, should be relegated to the synonymy of jasciata, a view previously suggested by Conant (1961: 19). The history of the description of engelsi is of considerable in- terest, and we quote from a letter that Dr. William L. Engels sent to the senior author under the date of 27 February 1962: "Dr. Barbour spent a few days in Chapel Hill in May 1942 .... Dr. Coker brought him around to me. I was just finishing the manuscript for my report on the Ocracoke fauna, and I showed him the king snake I had from there which was certainly a pecu- har looking, chocolate-colored individual. I also showed him the only other hve snake I had, the rat snake from Shackleford Banks. He immediately got quite excited, came back the next day and stared at those snakes some more, and wanted me to sit down immediately to describe them as new. I was about to go into the Air Force, did not have any comparative material, and at any rate was totally lacking in experience in systematic herpetology. The end of it was that I sent him the snakes and went myself into service. He wrote me proposing that either my name should be given to one of them or that my name should be put on the descriptions as joint author. I agreed to the lat- ter, and that ended our collaboration. I dehberately did not 1973 THE CAROLINA SALT MARSH SNAKE 3 show him the water snakes — I thought that was a Httle project I could keep for myself, for after the war (they did strike me as being unusually dark). But Dr. Coker did not know this. He did know that Dr. Barbour had become quite excited about 'the new herpeto fauna' which I 'had discovered' — the quotes are Barbour's words. So, having come into possession of a Matrix from the Mullet Pond on Shackelford, while I was in Egypt, he sent it to Barbour. That became the type of .V. s. engelsi." The practice of describing new taxa from single specimens, although always potentially risky, was widespread 30 years ago. Consideration should also be given to the fact that Barbour was an alpha taxonomist; he seldom examined large series of speci- mens, and he was accustomed to describing new species from islands of the West Indies. It is understandable that he could engender strong enthusiasm about unusual insular specimens. His type of engelsi, a \ery dark snake in which black pigment ob- scures man\ of the pattern details, is now known to be a rep- resentati\e of a deme inhabiting Mullet Pond that shows strong exidence of hybridization between sipedon and jasciata. As is noted below, some of the snakes of this population are morpho- logically indistinguishable from sipedon, others could be identified as jasciata, and others combine characters of both. This phe- nomenon is not unique to the Shackleford Banks. It also occurs at other places where fresh- and saltwater habitats meet and in a few localities within the Fall Line zone of both the Carolinas and Georgia (Conant, 1963). In an effort to end the confusion that has thwarted a real understanding of the Matrix sipedon-jasciata complex of the re- gion for three decades, we take pleasure in describing: NATRIX SIPEDON WILLIAMENGELSI n. ssp. Holotype. MCZ 129298, a young adult male, collected by Paul Elias, 14 June 1971, along Island Greek on Ocracoke Island, Hyde Co., North Carolina. Paratypes. (All from North Carolina. ) CORE BANKS, Car- teret Co.: North Carolina State Museum (NCSM) 11 796, MCZ 131993, Cape Lookout. HATTERAS ISLAND, Dare Co.: United States National Museum (USNM) 72573; MCZ 130278-79, Hatteras \^illage. OCRACOKE ISLAND, Hyde Co.: MCZ 129290, Horse Pen Creek; MCZ 129299, 129302, Island Creek; MCZ 129242, between Island Creek and Pamlico Sound; MCZ 131990-91, Ocracoke ViUage; MCZ 131988, 4 BREvioRA No. 400 Quork Hammock; MCZ 129288, 129300, Quorks Point Creek; MCZ 129249, Old Hammock Creek; MCZ 129250, Pony Pas- ture Creek; NCSM 11798, 1 mi. W of The Knoll; NCSM 11797, Try Yard Creek. PORTSMOUTH ISLAND, Carteret Co.: MCZ 129251-74, 129289, Portsmouth Village. MAIN- LAND: Carteret Co.: NCSM 8003, 1 1 mi. S of Merrimon; MCZ 131992, SW part of Cedar Island. Hyde Co.: American Museum of Natural Histor>' (AMNH) 88076, Rose Bay Creek. Pamlico Co.: AMNH 88075, Oyster Creek near Lowland. Diagnosis. A strongly melanistic race of Matrix sipedon in which: (1) the dorsum in adults is essentially black with the interspaces between the crossbands and blotches so dark that pattern details often can be discerned only if the specimen is immersed in hquid; (2) the venter posterior to midbody is pre- dominantly black; (3) there are no reddish- or brown-centered ventral crescents or half -moons posterior to the 50th ventral; and (4) the light scales between the dark crossbands average one and one-half (maximum three) on the neck at the level of the second dorsal scale row. Description of the type. MCZ 129298 is a young adult male 725 mm in total length; the tail is complete and measures 168 mm. There are 138 ventrals, 72 subcaudals, and a maximum of 23 dorsal scale rows. The anal is divided and both hemipenes are everted. The general coloration is black with dull gray in- terspaces between the black bands and blotches. The interspaces vary, at the level of the second dorsal row, from one to two scales in width; most are about a scale and one-half. The dark dor- sal bands are continuous, from ventral edge to ventral edge across the anterior part of the body; they are replaced posteriorly by alternating lateral bars and middorsal blotches, as in the nomi- nate race. The entire snake is so dark that the pattern is most clearly observed under liquid. The ventrals are virtually uniformly black at the posterior end of the body, but are lightly spotted with white; the white spots are larger and much more prominent on the anterior third of the body. There are no reddish- or brown-centered ventral crescents posterior to ventral 48. Variation. We have examined 50 specimens of this race (26 males and 24 females). Of these, the senior author examined three alive, the junior author more than 20 alive or freshly killed on the road. The very dark coloration of williamengelsi is little affected by preservation, and there is far less variation than in the nominate race. All specimens are black or dark brownish 1973 THE CAROLINA SALT MARSH SNAKE 5 black with \er\ narrow, lighter, grayer interspaces between the daric bands. The bands are so wide relati\c to the interspaces that most indi\iduals appear uniformly black in the field. One specimen ( MCZ 131990) from Ocracoke \'illage was aberrantly red. The areas normally white or gray, ventrally and laterally, were rose or rustv. There were, howe\ er, no red-centered cres- cents posterior to \entral 44. Many specimens show an indication of the pale vertical bar that crosses the angle of the jaws as illustrated in Conant ( 1963, fig. 6E). This is frequent in Natrix sipedon from southern coastal localities, and usually quite different from the pale longi- tudinal stripe of Natrix jasciata (Conant, 1963, fig. 6F) . Most specimens have red- or brown-centered ventral crescen- tic markings on the anterior \enter; the most posterior of these \ary in position from ventral six to \entral 48. Ven- young speci- mens, such as MCZ 129252-74, from Portsmouth, Portsmouth Island, have no red- or brown-centered \entral crescents at all. Scale characters are of little use in distinguishing Natrix jas- ciata from Natrix sipedon sipedon (Conant, 1963: 11-15), and they are of no value in separating williamengelsi from the nominate race of sipedon. Overall variation in williamengelsi is as follows: ventrals 129-143, mean 138 (females 133 to 143, mean 138; males 129 to 143, mean 137); subcaudals 57-84, mean 70 (females 57-70, mean 64; males 62 to 84, mean 75) ; maximum number of dorsal scale rows in both sexes 21-23 (among the 50 specimens, only six vary from 23 — three have 22 and three have 21). Size. Our largest specimen (MCZ 129290, a female from near Horse Pen Creek in the "Up Trent" region of Ocracoke Hammock, collected 16 June 1971, by Numi C. Spitzer) meas- ured 1224d= mm (48 in.) in life. This snake is illustrated in Figure 1. The smallest specimen examined is 139 mm in total length. This snake (a male) is the runt of a litter of 24 the rest of which vary from 193 mm to 216 mm (mean 204 mm). These are young of a female (MCZ 129251) from Portsmouth, Ports- mouth Bank, that measures 1040 mm, total length, and has a stump tail 1 25 mm long. Comparisons. Melanism occurs sporadically in several parts of the range of the nominate subspecies (sipedon). Occasional individuals of that widely distributed complex may be as dark dorsally as t)pical specimens of williamengelsi, and some large adults may be almost plain black or very dark brown. Ventrally, however., there is great variation in such dark snakes. In some. 6 BREVIORA No. 400 the under surface may be virtually plain black, but in others there may be bold black half-moons with or without deep red centers, a longitudinal midventral stripe of white, yellow, orange, or red may be present, or the entire belly may be dusted with gray or black, etc. Among the many thousands of specimens of A^. s. sipedon that we (collectively) have examined, we have never encountered a melanistic population that combines all four diagnostic characteristics of williamengelsi. Melanism in other populations is local or restricted to a few individuals, usually large adults. INTERGRADATION Intergradation between williamengelsi and nominate sipedon is extensive. The fresh- and brackish-water marshes of Pea Island, which is the northern end of Hatteras Island, and on Bodie Island, just north of Oregon Inlet, have yielded a considerable number of specimens from intergradient populations. Among 13 from the Bodie Island marshes, five appear to be typical sipedon, four are indistinguishable from williamengelsi, and four are in- termediate. Among 27 specimens from Pea Island, eight show the pattern characteristics of sipedon, six those of williamengelsi, and the other 13 are intermediate. We have 11 additional specimens from the zone of intergradation between sipedon and william- engelsi. A specimen from east of Manteo, Roanoke Island, is virtually typical A^. s. sipedon; another from the same locality is close to williamengelsi. Of two from Waves, Hatteras Island, one is intermediate, the other virtually williamengelsi. A speci- men taken eight miles north of Avon, Hatteras Island, has all the characteristics of williamengelsi. Another, from 4.3 miles north of Buxton, Hatteras Island, is typical of A^. s. sipedon. From the Buxton area, the southernmost limit of the zone of in- tergradation, one specimen appears to be typical of sipedon, two are typical of williamengelsi, and two are intermediate. Two specimens from Hatteras Island, without precise data, but pre- sumably from north of Hatteras Village, are A^. s. sipedon in most characters. A specimen from Creed's Hill, Frisco, within the Buxton Woods region but well west of Buxton proper, is in- termediate. We consider the following preserved specimens ( all from Dare Co., North Carolina, except as indicated) to be intergrades be- tween Matrix sipedon sipedon and A^^. s. williamengelsi: BODIE ' 1973 THE CAROLINA SALT MARSH SNAKE 7 ISLAND: AMNH 108839, 108842-50; Duke University col- lection (DU) 288. HATTERAS ISLAND: MCZ 129294, 8 mi. N of Avon; DU 227, 229, and MCZ 129296, Buxton; MCZ 129293, 4.3 mi. N of Buxton; NCSM 11800, Cape Hatteras; DU 228, Creed's Hill; USNM 72571-72, NE of Hatteras VU- lage; DU 242, and MCZ 129301, Waves. PEA ISLAND: AMNH 108835, 108837-38, 108840-41, 108851-64 and MCZ 129309-10; DU 231, 1/4 mi. S of Oregon Inlet; DU 226, 230, 2 mi. S of Oregon Inlet; MCZ 129295, 4 mi. N of Rodanthe. ROANOKE ISLAND: Carnegie Museum (CM) 23074, E of Manteo; DU 120, 2 mi. SE of Manteo. MAINLAND: Hyde Co.: AMNH 89149, Engelhard. Another specimen of Natrix sipedon sipedon is now available from the peninsula between Pamlico and x\lbemiarle Sounds, an area in which members of this taxon apparently are quite un- common. This is NCSM 11475 from Slades Creek, near Slades- \iUe, Hyde County. The onl\- other specimen of the nominate race reported from the peninsula is AMNH 43427 from Lake Phelps (Conant, 1963: 26). HYBRIDIZATION Hybridization with Natrix fasciata is also a major feature of A^. s. willia7nengelsi biology. An apparently typical but excep- tionally dark female Natrix fasciata. from Ponzer, on the Hyde County mainland (AMNH 88077), produced a Litter (AMNH 88078-101 ) that exhibits a variety of characteristics ranging from those of the mother f fasciata) to those of typical N. s. williain- engelsi. The pale marking passing through the angle of the jaws matches the \-ertical bar of Natrix sipedon in some, whereas in others it is comparable with the longitudinal stripe of Natrix fasciata. The dorsal markings \'ary from the posteriorly stag- gered middorsal and lateral bars of sipedon to the bands that are continuous throughout the length of the body in fasciata. All members of this litter are very dark. (See Conant, 1963 : 30, for further details.) The mother, as noted, is a ver\' dark A^. fasciata; we do not know the relevance of visual cues in mate selection in these forms of Natrix; perhaps they are important. A hybrid swarm occurs at Mullet Pond, on the Shackleford Banks, in a freshwater ecosystem of largely mainland ecological affinities. Among 22 specimens from Mullet Pond, four (in- cluding MCZ 46688, the type of "engelsi") are morphologically typical of Natrix fasciata, three are more or less typical of wil- 8 BREVIORA No. 400 liamengelsi, and the others, Including young from three litters and totalling 15 specimens, are intennediate. Snakes from Mullet Pond, Shackleford Banks, are: AMNH 75800-04 + 13 un- tagged; DU 10, 234; California State University, Long Beach (CSULB) 263-64. Apart from the Ponzer litter and Mullet Pond specimens, we have examined one other snake from the North CaroHna main- land which is apparently a hybrid. It is NCSM 11799 from Lennoxville Point, 3 miles east of Beaufort, Carteret County. We are of the opinion that much of the apparent introgres- sion manifest along the shores of Pamlico Sound and associated bodies of water is a direct consequence of the massive disturb- ance of habitats by the great cyclonic storms that sweep ashore rather frequently in eastern North Carolina. To visit the marshes after a hurricane, as Lazell did, following Ginger in early Octo- ber, 1971, is to see convincing evidence that salt water can be driven far inland by winds and tides. Conversely, as the result of torrential rains, vast quantities of fresh water move seaward after each tempest has passed, carrying many animals and plants of the freshwater biota with them on debris borne by the flood. Some of these organisms certainly must make landfalls in the salt marshes. Mayr, Linsley, and Usinger pointed out long ago (1953: 102) that hybridization between related species may oc- cur in localities where the natural ecological balance has been badly disturbed by human interference. Conant (1963: 16- 23) marshalled evidence indicating introgression between sipe- don and fasciata along the Fall Line in the Carolinas where mankind has been active. Surely hurricanes are equally disturb- ing to the ecological balance, at least in some localities. The distribution of specimens examined is shown in Figure 2. We have made no attempt to show localities for Natrix fasciata. which probably occurs in \irtually every body of fresh water on the adjacent mainland. Some specimens on which we report pat- terns and scale counts are not included on our lists. A few snakes were discarded because of their poor condition, and one escaped near where it was collected. ECOLOGY Natrix sipedon ivilliamengehi is closely associated with marsh grasses of the genus Spartina and the rush, Juncus, wherexer these form salt and brackish marshes. On Ocracoke Island, wil- liajnengelsi is common in salt marshes adjacent to Pamlico Sound, 1973 THE CAROLINA SALT MARSH SNAKE 9 along the tidal creeks that carry brackish water among the dunes, and also in the fresh water of land-locked ecosystems such as that which occurs at Island Creek. The density of the water where some of the snakes were collected varied from that of the Sound proper, 1.018, to Island Creek at .999. The density at five collecting stations sampled in May and June, 1971, ranged from 1.007 to 1.015; the average density was 1.013. The relation of density to salinity is dependent on tempera- ture. The water samples taken at the times the snakes were col- lected varied from 25° to 29°C; the average was 27°C. Salin- ities of seven habitats sampled (including the Sound, Island Creek, and the above five) \aried, therefore, from 2.7 to 24 grams per kilogram ; the average was 20.5 g/kg. In the laboratory, specimens of A^. s. williamengelsi fed vora- ciously on the minnow, Fundulus heteroclitus, the frogs, Hyla cinerea, Hyla squirella, and Rana pipiens, and the toad, Bufo fouieri. All of these are abundant on Ocracoke. We have the impression that these snakes will eat almost any fish or amphibian small enough to swallow, as is also the case with the nominate race. E\en after a full week of water star\'ation, our captives would not drink salt water, but they eagerly drank fresh water. Pettus ( 1 963 ) obtained similar results with Gulf Coast salt marsh snakes, Matrix fasciata darki. N . s. williamengelsi is a fa\'orite prey of marsh hawks, Circus cyanens, and is no doubt taken by other raptors. The numer- ous species of egrets and herons that frequent the Carolina salt marshes in all probability eat these snakes regularly, but we ha\e not observed this. Many specimens have mangled or abbre\iated tails that suggest predation by birds or the crabs that are com- mon in the habitat. The junior author, on field trips o\'er several years, has found A^ s. williarnenoehi abundant from May to September in the op- timal habitats south of Cape Point, Hatteras Island, on the Outer Banks. In the Buxton Woods area, north and west of Cape Point, Agkistrodon piscivorus is common, but it is not known from the Banks to the southward, at least to Cape Lookout. Natrix sipedon is scarce in the Buxton Woods area, where it may fall prey to the cottonmouth. The only other water snake known from the Outer Banks is Natrix taxispilota: it is confined to fresh water in the Buxton Woods area, and is scarce. It should be pointed out that a large part of the range of N. s. Willi ayn en gels i is within National Park borders. The blanket 10 BREMORA No. 400 protection provided for all animals includes this snake. Collect- ing is illegal within the Cape Hatteras National Seashore, ex- cept by specifically designated Park Service employees. ACKNOWLEDGMENTS We are indebted to staff members of the Cape Hatteras Na- tional Seashore Recreational Area for their cooperation with the junior author during his field work, especially to Clay Gifford, Park Naturalist; David Fletcher, Ranger; and John A. Musick, Collaborator. Specimens were lent from several study collec- tions by, respectively, Charles M. Bogert and Richard G. Zweifel of the American Museum of Natural History, Neil D. Richmond of the Carnegie Museum, Joseph R. Bailey of Duke University, Richard B. Loomis of California State University, Long Beach, Ernest E. Williams of the Museum of Comparative Zoology, and William M. Palmer of the North Carohna State Museum. Wil- liam L. Engels and John B. Funderburg loaned us material from their private collections and permitted us to deposit it in the North Carolina State Museum and the American Museum of Natural History, respectively. The senior author is grateful to Dr. Bailey for making it pos- sible to visit the Shackleford Banks. Harold A. Dundee fur- nished useful information, and Hobart M. Smith has read the manuscript and made a number of helpful suggestions. The junior author was actively aided in the field by James Woessner, John R. Alexander, Paul Elias, Edward James, Peter Rabinowitz, Numi C. Spitzer, and Dr. Musick. Meristics, meas- urements, and other data were assembled by Norma Rothman, Constance A. Rinaldo, and Miss Spitzer. Mrs. Rothman ob- tained many specimens for the senior author's studies on the genus Natrix in conjunction with her own field work in North Carolina. 1973 THE CAROLINA SALT MARSH SNAKE 11 LITERATURE CITED Barbour, T. 1943. A new water snake from North Carolina. Proc. New England Zool. Club, 22: 1-2, pi. 1. CONANT, R. 1961. A new water snake from Mexico, with notes on anal plates and apical pits in Natrix and ThamnopJns. Ameri- can Mus. Novitates, No. 2060: 1-22. 1963. Evidence for the specific status of Natrix fas- ciata. American Mus. Novitates, No. 2122: 1-38. Engels, W. L. 1952. Vertebrate fauna of North Carolina coastal islands: Shacklefoi-d Banks. American Midland Nat., 47: 702- 742. Ma^-r, E., E. G. Linsley, and R. L. Usinger. 1953. Methods and Principles of Systematic Zoology. New York, N.Y.: McGraw- Hill Book Co. 346 pp. Pettus, D. 1963. Salinity and subspeciation in Natrix sipedon. Capeia, 1963: 499-504. 12 BREVIORA No. 400 N s r- 4J o 'So CD ^H » ?.* -§ (—• ^ r^ 4J o ,^*^ o o V. ?H :z; X o Ah •S •^ CM d QJ :?: OJ QJ > cc ■r! , , r~i >> 0$ QJ a; o w rH r-* r-I QJ QJ > Tfi ?; hH CD 1— < p QJ % g o yA o; o , -u rt > c^ ?H r^ rt M QJ o O QJ «*-! 1 c3 ^ rt _S 9 m ^^ o O p ?: C 0) CiC u M QJ tH ^ a; < OJ P fn o 3 ^ o r- fc/j ^ — ^ rt E o C3 O i (M -t CI (M C<1 «(H Cvl 1973 THE CAROLINA SALT ^L^RSH SNAKE 13 LEGEND • N. s_ williomenqeisi o N. s sipedon and intergrades with williamenqelsi X N. s. williamenqelsi hybrids with N. f. fasciata Roanoke\^ \ " and 15 Ocracoke Portsmouth Shacklefo Bank Cape Lookout Bodie Island Pea Island Waves Avon 'Buxton Hatteras N Figoire 2. Map showing distribution of Natrix sipedon wilUa'nv- engeh'i and intergrades between it and A'. ,s. sipedon and hybrids be- tween wUliamengelsi and A^. fasciata fasciata. ^- -CMP. 200, MAR 1 Q fn.— B R E V I O R A Rloseiim of Coniparatiye Zoology CIambridge, Mass. March 6, 1973 Number 401 THE CHANARES (ARGENTINA) TRIASSIC REPTILE FAUNA. XVIII. PROBELESODOX MLXOR, A NEW SPECIES OF CARNIVOROUS CYNODONT; FAMILY PROBAINOGNATHIDAE NOW Alfred Sherwood Romer Abstract. A small species of Probelesodon, similar to P. lewisi but of smaller size, is described as Probelesodon. miyior sp. nov. Because of its advanced nature, particularly in the development of a squamosal socket for lower jaw articulation, it seems advisable to remove Pi-obainognatJius from the Chiniquodontidae and to erect for it the monotj^iic new family Probaino.crnathidae. Probelesodon minor sp. nov. Holotype: La Plata Museum 64-XI-14-18 (field no. 138 pt. ). A skull and jaws (Figs. 1, 2). From the Chaiiares Forma- tion, about 4 km north of the mouth of the Rio Chaiiares, La Rioja Pro\ince, Argentina. Diag7iosis. A small species of Probelesodon, al;)out half the size of P. lewisi, known skulls having a basal length of about 70 mm. In earlier papers on the carnivorous cynodonts from the Chaiiares (Romer, 1969, 1970), two forms were clearly dis- tinguishable, a large form described as Probelesodon lewisi, and a small, more advanced form described as Probainognalhus jenseni. At the time of publication, preparation of this carni\o- rous series had not been completed and it was assumed that all the small skulls pertained to Probainognalhus. \Vith further preparation by Head Preparator Arnold D. Lewis, it became obvious that this is not the case. Se\"eral small skulls and jaws are clearly of the Probelesodon type. All are apparently mature, and do not represent growth stages; all are about half the size of the "typical'' specimens of Probelesodon leivisi. and show a size difference too great to be sexual dimorphism. It is obvious 2 BREVIORA No. 401 Figure 1. Probelesodon minor, holotype skull in dorsal view, X 3/2. that we have a second, small, species of Probelesodon which is herewith described. In addition to the holotype the new species is represented by: MCZ 4100, a skull and jaws; MCZ 4099, partial skull and jaws; MCZ 3777, jaws and some postcranial scraps; MCZ 4102, in- complete jaws; MCZ 4163, postcranial materials. The holotype skull is incomplete in the premaxillary area, but total length was approximately 78 mm from snout to the level of the posterior end of the squamosal arch, 68 mm to the condyle. The lower jaw length is 64 mm. In MCZ 4100 the comparable figures are 73, 70, and probably 62 mm (the jaw is incomplete posteriorly). The materials of MCZ 4099, MCZ 3777, and MCZ 4102 are comparable in size. The postcranial material of MCZ 3777 (to be described in the next paper in this series) is comparable in nature to equivalent elements in P. lewis i except for smaller size, and the postcranial elements of MCZ 4163 are similar. In all major features the skull and jaws are comparable to those in P. lewisi; skull proportions are closely comparable and the sutural pattern is similar, except that in P. ?ninor the lacrimal and prefrontal are somewhat more expanded dorsally at the 1973 CHANARES GO^rP^ODONT.S Figure 2. Prohelesodon minor, holotype skull in lateral view, X 3/2. expense of the nasal. The secondary palate, as in P. lewisi, extends remarkably far back, exceeding in de\elopment e\en the contemporary Probainognathiis. As in Probainognathus jenseni the postero\entral flanges of the pterygoid are highly developed; also as in that species the posterior end of the lower jaw lies very close to the inner surface of the squamosal, and there was presumably a ligamentous connection, although no development of a '"glenoid" articular surface. As in P. lewisi there were four premaxillary "incisors," and there are, in the type and MCZ 4100, eight postcanine cheek teeth. As in P. lewisi these teeth are somewhat "hooked" — cur\ed backwards at their tips. Three "incisors" are present in the lower jaw, and seven "cheek" teeth behind the prominent canine. I ha\-e earlier suggested that Prohelesodon lewisi was ancestral to the larger Belesodon of the Brazilian Santa Maria beds. Closely related to the latter w'as the smaller, contemporary, Chiniquodon. Not improbably P. minor may ha\"e been an- cestral to Chiniquodon, a form of similar structure but of considerably larger size. Probaixogxathidae Fam. Nov. The carnivorous cynodonts from the Chanares Formation, Probainognathus and Prohelesodon. are both ob\iously ad\'anced forms, with an elongate secondary palate and other progressi\-e features. Similarly advanced are the Santa Maria forms, Chini- quodon and Belesodon, for which Huene (1944) erected the family Chiniquodontidae. In the past, I ha\e included both 4 BREVIORA No. 401 Chafiares genera in that family. TJiat this assignment is correct for Probelesodon seems certain, for that genus is obviously closely related to the Santa Maria forms. The position of Probain- ognathus, however, calls for further consideration. Like the proper chiniquodonts, Probainognathus is quite surely de- scended from the primiti\e galesaurid cynodonts of the early Triassic. But the progress has been in a somewhat different direction. The chiniquodonts proper are even more advanced in palatal construction than Probainognathus, but are apparently somewhat aberrant in dentition and in certain postcranial fea- tures to be described in a future number of this series of publications. Probainognathus, on the contrary, appears to be takinc; a direct course toward a mammahan condition and shows a very marked progressive situation in the de\elopment of a "glenoid" socket in the squamosal for lower jaw articulation. On the whole, it is perhaps best to separate Probaiiiognathus as the type of a new family Probainognathidae, characterized by progressive cynodont features, most important of which is the articulation of the lower jaw with the squamosal in mammalian fashion. Collection and preparation of Probelesodon and Probaino- gnathus were made possible by grants from the National Science Foundation. REFERENCES CITED HUENE, F.v. 1944. Die fossilen Reptilien des siidamerikanischen Gondwanalandes. Munich: C. H. Beck'sche Verlag-s. 332 pp. ROMER, A. S. 1969. The Chafiares (Argentina) Triassic reptile fauna V. A new chiniquodontid cynodont, Probelesodon letvisi — cynodont ancestry. Breviora, No. 333 : 1-24. . 1970. The Chaiiares (Argentina) Triassic reptile fauna VI. A chiniquodont cynodont with an incipient squamosal- dentary jaw articulation. Breviora, No. 344: 1-18. MAR 1 8 1985 B R E V I O R. A Museum of Comparative Zoology Cambridge, Mass. June 1973 Number 402 SPEGIATION IN THE GENUS OCHTHOECA (AVES: TYRANNIDAE)^ John W. Fitzpatrick Abstract. The nine Chat-Tyrants are small tyrannid flycatchers occurring in the Andes from Colombia and western Venezuela south to Bolivia and extreme northern Chile. Most of the species are di- vided into numerous isolated races. There are several distinct species- groups, and the species are highly synipatric. The complicated ranges appear to have arisen during the final three Pleistocene glacial periods in the northern Andes. During glacial peaks, the then-ex- isting species dispersed over wide ranges. They were divided during the subsequent interglacials. The three periods of dispersal and sub- sequent splitting first isolated the species-group precui-sors, then the species themselves, and finally the races of each species. The many isolated races of each species today are separated by geographic and vegetational barriers which probably affect a number of other mon- tane species as well. Introduction Several recent studies (B. Vuilleumier, 1971; F. Vuilleumier, 1969, 1970) have revealed consistencies in the ranges and specia- tion patterns of the fauna of the South American Andes. The patterns indicate that a large number of Andean species orig- inated and diverged in conjunction with the periodic glaciations of the Quaternary. Alternating glacial advances and retreats resulted in the complex sympatry between related species oc- curring there today. The avian genus Ochthoeca reflects this type of complex sym- patr)'. Eight of its nine species have widely separated races, and although several species occur throughout the central and north- em Andes, others are confined to smaller areas. Sympatry among species in the genus is considerable. Tublished by a grant from the Wetmore Colles Fund. BREVIORA No. 402 O. CINNAMOMEIVENTRIS O. RUFIPECTORAUS O. FRONTALIS O. FUMCOLOR Figure 1. Four members of the genus Ochthoeca. Species-groups diadema and oenayithoides are represented by O. frontalis and 0. fumicolor, respectively. About two-thirds life size. 1973 SPECIATION IN OCHTHOECA 6 Known collectively as the Chat-Tyrants, all species are tree- or shrub-inhabiting flycatchers of the middle- and high-altitude Andes. They occur in the upper subtropical, temperate, and paramo zones from the mountains of western Venezuela and Colombia south to western Bolivia and extreme northern Chile. Members of the genus have been collected at altitudes from 500 to 4314 meters, and the a\erage for the group is slightly over 2700 meters. They are small, generally chunky and large-headed flycatchers, and ail have a conspicuous supraloral stripe, or "eye- brow," varying in length and color (see Fig. 1). Their chief mode of prey-catching is gleaning insects from the vegetation, though several species also sit on conspicuous perches and hawk insects from the air (F. \'uilleumier, 1971). All species are nonmigratory and, with one apparent exception, they are com- mon within their respective ranges. The ranges of each species, plotted from their collection lo- calities, are shown in Figure 2. Based on these present ranges and on the Andean glacial exents during the Quaternary, I shall propose here a sequence of speciation for the group. The Genus The crenus Ochthoeca, as recognized b\- de Schauensee (1966), contains the nine species listed below. Two species are quite distinct from the others and ha\e widespread ranges. The remaining seven ha\e been placed in two species-groups by F. Wiilleumier ( 1971 ) . O. cinnamotyieiventris O. rufipcctoralis diadema species-group O. diadema O. frontalis O. pulchella oenanthoides species-group O. oenanthoides O. fumicolor O. leucophrys superspecies 1 . leucophrys 2. piurae F. Vuilleumier considered O. pulchella and O. frontalis as members of a superspecies within the diadema species-group. Although they are very similar, differing chiefly in the amount of yellow in the eyebrow, both species have been collected at BREVIORA No. 402 cirmamomeiventris pulchella furaicolor oenanthoides 111' leucophrys piurae Figure 2. Ranges of species in Ochthoeca. The diadema and oenanthoides species-groups are shown on the lower left and lower right, respectively. 1973 SPECIATION IN OCHTHOECA 5 each of four localities, and their ranges show considerable over- lap south of the Maranon River. I shall, therefore, treat them as sibling species within the group. One species, cinnamomeiventris, occurs in the moist forests of the upper subtropical and temperate zones. Its dark colora- tion is adapted to these dense, wooded habitats. Its bill is the broadest and shortest of the genus, reflecting its diet of aerial insects captured by hawking. O. rufipectoralis and the diadema species-group inhabit open forests of the temperate zones. Their bills are also short, but all are much thinner than that of cin- namomeiventris. O. diadema is the only species whose plumage is largely green. The oenanthoides species-group contains forms frequendy found over 4000 meters. O. fumicolor lives in moist paramo habitats, while oenanthoides and the leucophrys super- species occur in the dry puna scrub and grasslands (F. Vuil- leumier, 1971). Members of this group have long, thin bills adapted to their diet of insects gleaned from the vegetation. All ha\'e significantly larger bodies than the other Ochthoeca species, with relatively longer wings and tails. O. leucophrys is predomi- nantly grey, while the other two are buffy brown. Pleistocene Climatic Influences The zoogeographic history of these species seems closely re- lated to the periodic altitudinal lowering of their habitats dur- ing the Quaternary glaciations. Glacial climates affected the forest- and grass-life zones in two ways. First, the limits of each were lowered, creating connections or closer proximity between previously isolated habitats. During maximum glaciation, the life zones were lowered sufficiently to become nearly continuous along the entire mountain chains (B. Vuilleumier, 1971). Sec- ond, the valleys and lower mountain areas became more humid as the ice and glacial lakes surrounded them. This again resulted in greater proximity between the formerly higher and isolated humid life zones. During the glacial maxima, the lowering of the habitats permitted many avian species to colonize through the northern mountain chains along a north-south axis. During interglacial periods the life zones retreated to higher altitudes, thereby splitting again into isolated "islands." The deep valleys became dry, further decreasing the potential for gene flow be- tween neighboring high-altitude populations. Thus a species whose range had been widespread and nearly continuous during maximum glaciation divided during the subsequent interglacial. New races and species evolved as a result. 6 BREVIORA No. 402 Within the central and northern Andes, two particular cli- matic features affected speciation in high-altitude populations. First, the western slope from southern Ecuador through Peru and northern Chile now receives very little rainfall. Presumably this was the same in the past, and the area remained generally unavailable to the forest-inhabiting Ochthoeca species. Only one species, leucophrys, occurs in this arid region today. Sec- ond, two large areas of low, dr\' vegetation occur within the range of this genus. The first is the low, semi-arid valley of the upper Magdalena River, which lies between the Eastern and Central Cordilleras of Colombia, creating a significant barrier between a number of montane taxa on either side (F. Vuil- leumier, 1969). Five Ochthoeca species now occur in northern Colombia and northwest Venezuela. Only that species inhabiting the lowest altitudes does not show a major break and/or sub- specific differentiation in this area. A second, equally efTective barrier occurs in the region of the upper Maraiion River in northern Peru. The low mountains in this area create a wide break between the extensive high-altitude regions of Peru and Ecuador. The ranges of all nine Ochthoeca species reach this divide, and six show major breaks at the barrier. Furthermore, evidence exists that warmer Pleistocene interglacial tempera- tures resulted in even drier conditions still less favorable to alpine life than today's climate (B. Vuilleumier, 1971). Hence these dry barriers isolated populations more effectixely in the past than they do today. These two major divides thus created three large, separate areas of montane conditions north of Bolivia which appear to have been the primary centers for speciation in Ochthoeca, and certainly for other species as well. Speciation in Ochthoeca Geologists recognize four major world-wide glacial advances during the Pleistocene (e.g., Leet and Judson, 1971), but the first had relatively less effect on the northern Andes, since they were the last to attain their present height (B. Vuilleumier, 1971). Data on high-altitude species in the northern Andes in- dicate that they were derived predominantly from central An- dean birds (F. Vuilleumier, 1970). Species originating in these central mountains dispersed northward during the three major glacial periods affecting the northern mountains. The precursor of the genus Ochthoeca seems therefore to have originated in the central Andes, and became isolated during the first Pleisto- 1973 SPECIATION IN OCHTHOECA to a 15 3 ■ss~5 hr ij - » r C w ^H •r: 00 CO 1— 1 LO ^H Lf O ^- th oa ■^ © HH fl^ •»» 9 -1 1 <■ cc a 1 ' r->i ■ U ■:■: MAR 1 8 1965 B R E V I O R A Museum of Comparative Zoology CAMBRrocE, Mass. June 1973 Number 404 NEW STUDIES ON A MONTANE LIZARD OF JAMAICA, A NO LIS RECONDITUS Robert Hicks^ Abstract, Data on the natural history of Anolis reconditns, an arboreal ignanid lizard, living in the montane mist forest of east central Jamaica, is presented. The animal uses the crown of trees both for sleeping and as a refuge when frightened. It pei-ches fac- ing down on open tree trunks, generally about 12 feet off the ground. Its displays appear to vary a great deal from individual to individ- ual. This animal seems to use a combination of two tyioically anoline foraging methods: 1) active searching and 2) motionless perching. In addition, reconditus displays a characteristic reported in no other Jamaican anole: it does not interrupt its daily activities because of rain. The lizard appears to be less specialized than its Jamaican congeners. This may be related to its apparent isolation fi-om con- generic competition. INTRODUCTION Of the seven species of anoline lizards in Jamaica, Anolis re- conditus is apparently the rarest and certainly the least studied. This relatively large {ca. 90 mm snout-vent length), montane reptile was first described in 1959 by Underwood and Williams from a single male specimen. The animal is usually green, brown, and beige. It is mottled and striped laterally and dorsally, often with a prominent small green patch just behind each eye. The type specimen was collected in 1953 in St. Thomas par- ish, on the edge of the Blue Mountain rain forest region in east central Jamaica. In 1961 and 1962 A. Stanley Rand collected five Hzards at Hardwar Gap, nine miles west of the type locality. In 1965, J. D. Lazell captured five additional reconditus, in- cluding the first females, in the Hardwar Gap area; he pub- lished a paper (Lazell, 1966) that added substantially to the '219 Steeplechase Road, Devon, Pennsylvania 19333 2 BREVIORA No. 404 scanty data on coloration, squamation, and behavior of this liz- ard. Robert Trivers in 1969 collected, marked, and released several specimens at the Green Hills field station of the Institute of Jamaica, one mile northeast of Hardwar Gap. In 1970, I spent the months of April and May in Jamaica observing all seven species of AnoHs on the island for varying periods of time. I spent two weeks with Robert Trivers, who was working on A. garmani, and two weeks with Thomas Jens- sen, who was studying A. opalinus, A. lineatopus neckeri, and A. sagrei. Between April 26 and May 28, I lived three weeks at the Green Hills field station, near Hardwar Gap. During this time I caught, marked, and observed 32 reconditus. I saw at least another two dozen. The data that follow are the results of my study. HABITAT AND RANGE During my stay at Green Hills, I discovered some specific new localities in the Hardwar Gap region which have substan- tial populations of this species. The population that Lazell (1966) describes as living in the garage near the field station, I found in the garage and on a stand of trees just adjacent to and east of the garage. I also saw one animal living under the eaves of the Denoes' house, which is just east of the field sta- tion and across the road. One of my main study areas, referred to hereafter as "Steep Bush," was located about one-quarter mile southwest of the field station along the road to Hardwar Gap. The road is cut into the mountainside, which slopes 50° or more. The viney, densely wooded "montane mist forest" (Asprey and Robbins, 1953) is called "bush" by the local people. The area where I worked on seven difTerent days is on the top side of the road. I concentrated on a roughly square area that runs along the road for about 300 yards and extends up the mountainside an equal distance. The trees there are generally 30 to 50 feet tall, with no branches below 20 feet. Several feet of their root systems are exposed as they cling to the steep mountainside. Moss covers the rocks, much of the tree trunks, and the fallen trees and limbs. There are also 15-foot tree ferns, vines, and saplings. The canopy is quite dense and everything, including the deep leaf litters, is very damp. The trees are generally plane trees. Wild yacca (Podocarpus urbanii), milkwood (Sapium jamai- 1973 ANOLIS RECONDITUS Map 1. Hai'dwar Gap-Green Hills vicinity. Note collecting sites for reconditus marked by triangles. Open squares are buildings. cense), and trees known to the local people as "fleura" are most abundant. In Steep Bush, I caught ten lizards and made many observa- tions. However, movement for an observer is extremely difficult in this "montane thicket" (Asprey and Robbins, 1953), and, owing to the mazelike quality, resightings are nearly impossible. My best study area was a place, known locally as "Fairy Glade," one-quarter mile east of Hardwar Gap. It is on top of the ridge that lies to the east of the Gap and is the largest area of flat "bush" land that I could find within several miles of Hardwar Gap. The glade is about 10 acres in extent, and has a path 10 to 30 feet wide running through it which leads northeast to St. Catherine's Peak. Lazell (1966) quotes Rand as saying he found reconditus in "mossy, stunted forest — nearly 'elfin woodland' — at Hardwar 4 BREVIORA No. 404 Gap." This is an excellent description of the Fairy Glade. Al- though generally found over 5000 feet, elfin woodland occurs as low as 2500 feet in the John Crow Mountains, which are east of the gap. Asprey and Robbins (1953) regard elfin wood- land as a faciation of montane mist forest. Fairy Glade repre- sents an intermediate stage between the montane mist forest of the steeper, lower slopes and the dwarfed, stunted, typically elfin woodland of the higher altitudes. The omnipresent moss is often six to eight inches deep on the ground and on the rot- ting fallen trees and limbs which litter the ground. The trees in the glade are generally of large diameter with full foliage but are occasionally twisted, unlike the tall, straight, branchless trees of Steep Bush. The growth of the glade could certainly be described as lush. An early botanical observer, in describing similar Jamaican vegetation, spoke of "the confusing abundance of species and the rich plant life in a muggy, glass- house atmosphere." (Asprey and Robbins, 1953) . I caught and marked 15 lizards along a 200-yard stretch of the path which runs through the glade. Most of my data are for these lizards. I also observed lizards at two other localities in the Hardwar Gap region : ( 1 ) in the northwestern part of Holywell Park, just west of the gap, and (2) in the "19 Mile Gulley," a steep, dark, damp, rocky ravine about four miles southeast of the gap and along the road to Kingston, just below the 1 9-miles-f rom- Kingston roadside marker. This latter area is several hundred feet lower than Steep Bush but has about the same physical character and vegetation. In half an hour on my final day in the Hardwar Gap area, I located four large males and one fe- male in this gulley. In Holywell Park, I saw only one animal, a male. With the, exception of Lazell's garage (see above), I was directed to all the places where I found reconditus by local Ja- maicans. Consultation with those local people who frequent the "bush" could in the future lead to a better understanding of the extent of the lizard's range. It seems likely that reconditus extends over a narrow, per- haps discontinuous band of montane mist forest running east and west of Hardwar Gap. The band is at least two or three miles wide in places. The Green Hills area is half to three- quarters of a mile north of the gap and the Newcastle area one to one and a half miles south of the gap. 1973 ANOLIS RECONDITUS 5 SIZE AND SEXUAL DIMORPHISM During ni\ month of study, I obscncd at \arious times more than 50 diflerent indixiduals. Thirty-two of these were cap- tured; 29 were measured. Of the 1 7 measured males, the largest had a snout— \ent length of 100 mm, the smallest 61 mm. The largest lower jaw (tip of snout to back of lower jaw), 33 mm, was that of a 98-mm male, the shortest, 20 mm, that of the smallest male (61 mm). The longest tail, 220 mm, was recorded on a 96-mm male; the shortest unbroken tail was 115 mm, on the smallest individual (61 mm) captured. Of the twelve females that I measured, the longest snout- vent length was 84 mm. This lizard had the longest lower jaw, 26 mm, of any female. The longest tail, 164 mm, was on a 77-mm lizard. The smallest female, smallest in each of the three measurements taken, was 62 mm snout-vent with a 19-mm lower jaw and an 121-mm tail. The body length of reconditus males is 1.17 times longer than females and the lower jaw of males is 1.19 times longer than females. Robert Trivers (personal communication) has found that for A. garmani in southwestern Jamaica, the adult males are 1.31 times longer than the adult females. COLOR AVhen describing A. reconditus, Lazell (1966) emphasizes a grey component in the coloration of the animal. According to my observations, a more accurate description of the lizard is that a tan or brown component predominates. I never observed any part of a lizard to be black or grey. Often parts of the liz- ard would be \ery dark brown. These appeared tan when the lizard changed to a lighter color phase. In my field notes I once described a cr\ptic male perched on a grey tree trunk as "brown, greyish mottled." However, I still maintain that close obser\-ation re\-eals only shades of brown, never grev. Interestingly, the populations in m\ main stud)' areas, three- quarters to one mile apart and ph)sicall\ isolated from each other only b)- hea\\ undergrowth and o\ergrown guUeys, seem to differ in color. The males of the Steep Bush area appear generally light brown with quite dark brown markings and stripes. They are green or slighdy bluish green just behind the forelegs. They can change to a mossy dark green or to a dark 6 BREVIORA No. 404 chocolate. With both these color changes, the markings fade out. The females of Steep Bush are lighter tan than the males, with prominent green patches on the head, just behind each eye. There is also a green streak just above each foreleg. The females can appear quite beige with the dark markings almost completely suppressed. Larger patches of light green on the upper body can appear with this lightening. The females can also display dark speckled markings over the tan, with green remaining only on the head and above the foreleg. The dewlap of the female, which is about half the size of that of the male, is yellowish green. The animals of Fairy Glade are generally darker in colora- tion than those of Steep Bush. The glade receives more sun- light than the bush forest and this may account for part of the color difference. Both males and females have the prominent green patches behind their eyes that characterize the females of the bush described above. In addition, the lower body of the males from just in front of the hind legs to the tip of the tail is fairly green, while the mid-body is mostly brown. One Fairy Glade female I described in my notes as being "a dull, mousey brown" with a bit of muted green on the shoulder and behind the eyes. I observed a 74-mm Fairy Glade female in a completely chocolate brown phase, with no mark- ings evident. She was delicately rosy red behind each leg, just where the leg joined the body. Females, as well as males, in Fairy Glade seemed also to have the power of males in Steep Bush to turn to a cryptic dark chocolate. In this phase the ani- mals have no markings evident. All color except the brown is suppressed. In Holywell Park, about a mile west of each of the two other areas, the one male I saw was very green over much of his body. As is typical of reconditus, his head and legs were basical- ly brown, but his trunk reminded me of the bright green of A. garmani. Markings were evident as patches of darker green on his back and sides. The striking shade of most of his body marked him as a member of possibly a third distinct population. These data are too scant for speculation about whether re- conditus occurs in distinct demes each with a specific color pat- tern. In view of the broken terrain, I may have been sampling separated segments of a continuum. More work is needed before a positive statement can be made about the local distribution and differentiation of reconditus. 1973 ANOLIS RECONDITUS Figure 1. Male perching- (Holywell Park). Note proximity of his perch to the hole in the trunk. See escape behavior. 8 BREVIORA No. 404 PERCHES Although I observed reconditus on a N^ariety of perches in a variety of positions on the perches, there was one tableau which I observed most frequently. The lizard would be about 12 feet up on a tree trunk or large-diameter branch (more than 4 inches) "in a head-down position, with the neck bent so that the head was nearly parallel to the ground" (Lazell, 1966). A. S. Rand first reported this stance to J. D. Lazell, but re- ported the perch height as from three to seven feet. I have two dozen observations of height and diameter of spe- cific perches, 1 1 for females and 1 3 for males. According to these I saw only four females and two males perched under seven feet. On the other hand, 16 of the perches were ten feet or over. The highest was about 16 feet. I have four additional entries with only perch diameter. I made many additional ob- servations for which I failed to record perch height or diameter. In such cases the lizard was perched, facing down, between 10 and 14 feet on a four-inch or larger diameter perch. I had come to expect them to be at this height and diameter and onlv if thev were not there did I make note of it. As stated above, the standard perch was of fairly large diam- eter. I recorded at least 14 different perches of eight inches or larger. The largest was about 18 inches. There were a few exceptions to the standard downward-fac- ing posture. I saw one male and one female who were perched for a long period of time facing up. I saw several perched cross- wise, coming out of rotten furrows or at a crook in a tree branch. In some cases, crosswise or upward-facing perches would be assumed for a minute or two as transitional positions between two different downward perches or between periods of mo\e- ment. One striking and possibly unique element in the site that re- conditus chooses as a perch is that in a great number of cases it is near a hole in a tree trunk or branch (See Fig. 1 ) . In the Hardwar Gap area with its frequent rainfall and lush vegeta- tion, it is common for a tree to have a partially exposed root system and many dead moss- and vine-covered limbs. Holes into the heartwood of a tree where a branch used to be are quite frequent. I counted no less than eight separate perch sites each within a few inches of prominent trunk or branch holes. One of these had three marked lizards, two females and a male, perched by it at different times. Another hole had two 1973 ANOLIS RECONDITUS 9 marked animals, one male, one female, who regularly used a perch next to it. Onh one lizard at a time was ever seen at any of the holes. Several times I saw lizards perched half in and half out of these furrows or holes. Rand ( 1967) reports that A. valencienni lays eggs communally in tree-trunk holes but apparently' no other anoline lizard uses them in just the same way as recondi- tus. It is possible that the ho'es are related to the foraging activ- ity of reconditiis (see foraging). They are certainly hiding places (see escape behavior). The fact that more than one individual uses the perch near a particular hole raises questions concerning the nature of the territoriality of this species. Perhaps there is only one male per hole and the females who use the perch are part of a group that resides within his territory. Robert Trivers (personal com- munication) has obser\ed for A. oarmani that indixidual males, especially large ones, seem to police territories containing several females. Each male appears to maintain exclusive copulatory rights with the females in his territory. Perhaps there is a sim- ilar situation with reconditus (also see home range and ter- ritorial structure). The holes raise another question. If a lizard is going to es- cape into a hole, instead of up a tree, he has to be sure that the hole is not already occupied. Indeed, the intimacy of a hole is a great deal more than that of the crown of a tree, and a liz- ard cannot see from a distance if he will ha\e company in a hole, the way he can at the top of a tree. Therefore it seems likely that at any time the lizards on a particular tree must have a good idea of where the other lizards on that tree are perched and where to go in case of danger. HOME RANGE AND TERRITORIAL STRUCTURE In an attempt to learn something about the habits of A. re- conditus, especially about its territoriality, I noosed and marked individual animals. After its capture, the sex of each lizard was determined, it was measured, given a number, and marked in two ways. For a permanent mark, part of one or two toes was clipped (see marking system at end for details). This allowed me to give each indi\idual a unique identification number. Also stripes of paint of various colors were painted on the lizard's back so that different indi\'iduals could be distinguished at a distance. The location of the perch where the lizard was 10 BREVIORA No. 404 caught was recorded and then the animal was returned to this site. The lizards were rarely held capti\e for more than fifteen minutes. During my study I resighted nine marked lizards. One, a 62-mm female, I saw seven separate times. All the resightings were in Fairy Glade. Although my data is scanty, it points to- ward a large male having a home range shared by two or more females, three large trees (30—40 feet X 12—18 inches) and a few saplings. In the heavy growth of Fairy Glade, the ground area of a range might be only 500 to 750 square feet. I have no data on whether reconditus defends these areas or any parts of them. Robert Trivers (personal communication) has found that males of A. garmani do defend territories, slightly larger in extent than the reconditus home range described here. Further study may reaveal whether reconditus also defends a ter- itory. On two very large adjacent wild yacca trees, each of which had two 12-18-inch trunks, I regularly saw a 74-mm female, a 77-mm female, and an 88-mm male. The larger female I saw on six different occasions, five times on the same tree, four times on the same perch. Once I saw her on the adjacent tree, oc- cupying the perch on which I had often seen the male. The male I also saw half a dozen times, four times on "his" tree and twice on the perch where I usually saw the female. The small female I saw only twice. The second time she was on the perch on the "male's" tree which had been used by both other lizards. The smallest female (62 mm) that I caught I saw six times at exactly the same perch. This was a small tree (four-inch diameter), five feet away from the tree on which I first caught her. She was always perched four feet ofT the ground next to the same furrow in the tree trunk. I saw an 81 -mm male three times, always within ten feet of where I first caught him. He was caught on a straight, branch- less tree {ca. 30 feet X 6 inches) and resighted on two adjacent trees, one a sapling (one- to two-inch diameter) and one a small tree (three-inch diameter) . I repeatedly saw a 95-mm male and a 75-mm female on or in the vicinity of two medium-sized ( 8 inches X 35 feet each ) , well-foliaged, adjacent trees. They spent most of their time at one perch, nine to ten feet up, next to a hole in the tree in which I first caught the male. One day I would see one on the perch, another day I would see the other. 1973 ANOLIS RECONDITUS 11 One large male (95 mm s-v; 30 mm jaw) I saw on three successixe days reveal either an enormous home range or rather atypical beha\ior. I caught him one day about 8 a.m. along the path, close to the entrance of Fairy Glade. Three hours later, I saw him again 12 feet further up the path. He must have crossed se\eral large trees on his way to this new perch. The next day I saw him on the other side of the path, about halfway between the two trees on which I had seen him the day before. Four hours later he was nowhere to be found. On the third day I spotted him at noon on the side of the path on which I originally caught him but ten feet further up the path than I had e\er seen him. I proceeded to follow him and photograph him for two hours as he covered an 80-foot stretch which contained many large trees. My field notes that describe this remarkable journey follow. May 17, 1970 Fairy Glade 1 2 : 00 noon — ^ # 5 (95 mm s— v ; 30 mm jaw ) Spotted 10 feet further up path (on side he was orig- inally caught on) than seen prexiously. 12:10 — Following #5. Moved up beyond tree where #11 (62-mm female) was caught. 12:35 — Much head bobbing after assuming new perch. Many ^ inch diameter sapling perches. 1 2 : 45 ■ — Ate some ants. 12:50- — Dewlapped 3 /a times. 1:10 — 5 dewlaps preceded by head bobbing. 1:15 — Moxed further up path and dewlapped 4 times with head bobs. Sitting out on tree fern fronds, 10 feet up. 1 : 25 — Taken up downward facing perch 1 2 feet up on 3 inch diameter tree. Brown, greyish mottled shading very much color of the bark. 1:50 — Still perched. Light rain. I left. #5's run started about 30 feet further up path than tree he was caught on. Run documented abo\e roughly followed path up farther into glade. Lizard covered 80 feet in an ap- proximately linear fashion. In that 80 feet were about 30 trees greater than Ij/o inches in diameter and many, many smaller trees. On other side of path where #5 was observed perching on May 16, this same 80-foot stretch contained six large trees. The dis- 12 BREVIORA No. 404 tance between the two rows of trees, i.e., distance across the path, was about 40 feet. Clearly much more work has to be done to establish a rea- sonable knowledge of reconditus' home range size or territorial structure. I caught no other Hzards, except the 62-mm female, in the vicinity of the trip of the large male described above. It is not impossible that all the ground he covered was "his." It is also true that my best resighting data is on females who seem to occupy one or two adjacent perches quite regularly. You might expect a female to spend more time in a specific spot in a male's home range while the male himself roams around more within the area. Perhaps I did not see a male regularly because he was much of the time out of my view, in some other part of his quite large home range. It is difficult to get an idea of the size of the reconditus popu- lation of Fairy Glade. If I had a good estimate, I would under- stand the dispersal of the species better. I only caught fifteen lizards, eight males and seven females, along the 200 yards of the path which I used for my study area. If that represents half or even possibly a third of the resident males, then it is possible that reconditus males have a large home range, some even as big as that of the particular large male I have described. I am quite certain that there were many more females in my area than I was able to catch because they are smaller, more cryptic, and have a less obvious display. Owing to the cryptic nature of this species, the general lush growth of the habitat, and the apparently quite timid character of the animals once they have been exposed to man (see escape behavior), I think it is very likely that I missed many more lizards than I caught. This would indicate that my conservative estimate of home range size, at the beginning of this section, is a reason- able one. INTRASPEGIFIG INTERACTION I rarely had more than one reconditus under observation at a time. On only two occasions did I observe interaction between two lizards, in both cases between a male and a female. The first interaction was one morning in Steep Bush. Initially the male ( ca. 90—95 mm ) was ten feet up on a tree trunk ( eight inches in diameter), facing down; the female (84 mm) was about seven and a half feet up on the same tree, also facing down, about 120° to the right of the male. After I had been 1973 ANOLIS RECONDITUS 13 watching them for half an hour, the male dewlapped once. Fifteen minutes later he ran up the tree two feet, dewlapped three times and assumed an upward-facing perch. After an- other {\\c minutes, the male ran down about a foot and toward the female slightly. He gave three head lx)l)s, paused and gave three more. His downward-facing perch was now about three and a half feet above and 60° to the left of the female. Ten minutes later, the female finally moved, running down and around the back of the tree. A bit of vine fell behind the tree and then she returned. Probably she was eating. For the next forty minutes the two lizards remained motionless. Then the male moxed down a few inches toward the female, but also to the left, putting more lateral distance between himself and her. He dewlapped three times, the second and third being from a half open position of the dewlap. Fixe minutes later the fe- male opened her mouth. After another fixe minutes, the male dewlapped again three times in a manner similar to his earlier series. Although the female appeared to display no direct response to the male, it looked to me as if the male was courting her. Neither before this occasion nor after did I see two lizards so close together. Because of the extreme peripheral vision of these lizards, it seems likely that the female cou'd have been watching the male's displays despite his position abo\'e and behind her. Although I was quite close at the time, nothing about the male's behavior suggested that he was reacting to me. He mav of course have been making territorial displays either in general or to spe- cific other hzards that I could not see. On the second occasion the object of the male's displa\' was e\'en more aml:)iguous. I caught a 74-mm female about ten feet from where an 88-mm male, previously caught and marked, was perched. Both lizards were on different trunks of the same large tree. A minute or so after I returned the newly marked female to the perch from which I had captured her, the male began one of the most elaborate displays I ha\'e e\er seen. He dewlapped five times, head-bobbed quickly twice, dewlapped once again, head-bobbed several more times, and then dewlapped once more while bobbing. Finalh he head-bobbed and then dewlapped an eighth time. Five minutes after this display the female assumed the exact perch and position she had occupied before I captured her. In the next fifteen minutes both the male and the female rotated 14 BREVIORA No. 404 their perches through 90°. Then twenty-five minutes after his grand display, the male resumed his original position and dewlapped two and a half times. The male's elaborate display was certainly in response to either myself or the female. If it was an aggressive display to- ward me, it was a minute or two after I had put the female back on the trunk and returned to my close-by observation post. Normally, if a lizard saw me he would sit and wait motionless, and then rapidly move away if I got too close. This makes me tend to think that the male was reacting to the female to whom I had specifically drawn his attention. The movements of the male and female described above after the male's display may also have been part of the courtship interaction. DISPLAY I observed no pattern to the displays of reconditus. In gen- eral the displays are, in regard to specific body movements, much like those of other Jamaican anoles. There is head-bob- bing; four-leg pushups; a front-legged, half pushup and pulled back, slightly cocked head preceding dewlapping; dewlapping combined with head bobbing; and sometimes, at the end of a series of dewlaps, a half dewlap pulse. On characteristic of the species is that the displays are slower and seemingly more "methodical," as I called them in my field notes, than those of other Jamaican lizards. I saw females display only three times; each time the lizard I was observing head-bobbed on assuming a new perch. One of these females I watched for nearly four and a half hours and saw only the single head-bob display. Males displayed much more frequently, generally dewlapping intermittently while they maintained a specific perch. One in- dividual that I watched for three hours dewlapped five or six times at various intervals, sometimes every ten to twenty minutes, sometimes every 30 to 40 minutes. Two different displays of possible courting males are discussed in intraspecifig inter- action. Although I recorded that a 95-mm male head-bobbed and dewlapped each time he assumed a new perch (see home RANGE AND TERRITORIAL STRUCTURE ), I also obsCrVCd malcS take new perches without display. Occasionally males directed what were probably aggressive displays toward me. These con- sisted of dewlapping once or twice. The elaborate display of one male just before he "went to 1973 ANOLIS RECONDITUS 15 bed" is worth describing. I had been watching this lizard con- tinuously for three hours. He had not moved for the last hour while darkness fell. Suddenly he shifted his downward-facing position on a three-inch trunk 30° to the right and began a rapid display: three head bobs, a dewlap, two more bobs, a dewlap, one bob, and finally two dewlaps. Seven minutes later he ran up two feet on the trunk, assumed a downward position, and again displayed: four pronounced head bobs followed by three dewlaps. Three minutes later he ran up two feet more, stopped and looked around. After a brief pause, he ran up six feet more, stopped, paused, and then continued up the trunk, disappearing in the dark foliage. It seems likely that males, and females to a lesser degree, dis- play to announce their position in their home range or territory to other members of the species in typically anoline fashion. But I could ascertain no pattern to this type of display, and my data is not extensive enough to identify consistent features of aggressive or mating displays. INTERSPECIFIC INTERACTION On only one occasion did I see reconditus in close proximity with a lizard of another species. In the stand of imported yac- cas, next to the garage where Lazell worked, I saw a large male A. garmani perched 20 feet up on a tree. A foot away on an adjacent tree, perched slightly lower, was a male reconditus that I was trying to catch. Contrary to LazelFs speculation (1966), it seems that the niches of garynani and recoriditus do overlap at least occasionally. On the steep path up to Fairy Glade from the Hardwar Gap road I also once saw a male garmani perched 30 feet up on a tree. I never saw any reconditus on the trail to the glade, but this garynani was less than 100 feet below the glade. It is pos- sible that the two species share the same habitat at the outer edge of Fairy Glade. I saw neither A. opalinus nor garmani in Steep Bush, inside Fairy Glade, or in the Nineteen-Mile GuUey, the three areas that I located with substantial reconditus popu- lations. It may be significant that of all the anoles in Jamaica, only reconditus lives for the most part without congeners (see discussion). ESCAPE BEHAVIOR Lazell (1966) describes the "wary and timid behavior" of 16 BREVIORA No. 404 reconditus. I too found this to be true, but usually only after the lizards had become "wild," as the local people refer to the lizards after thev ha\e become fris:htened of man. Generallv. "naive" lizards, individuals who have had no contact with man, are quite easily approached, and with a little care easily caught. Owing to their cryptic nature and restricted range, they are often hard to find, but once you find them, you usually can catch them. However, after you have caught, measured, and marked them once, even if you can find them again, which is dif^cult, you will have an incredibly hard time recapturing them. One day in Fairy Glade I saw a 75-mm female which three days pre- viouly had been caught and marked. She edged her way into the trunk hole next to her perch. I was 20 or 30 feet away \vhen this happened, and it suddenly occurred to me that the reason I was resighting few marked lizards was because they would see me coming and hide. I only made one recapture, that of a 78-mm female whom I had originally caught and measured, but who had escaped before she was marked. I attempted to recapture her half a dozen times before I succeeded. She grew more and more warv. It is not surprising at all that Lazell ( 1 966 ) had trouble catch- ing the lizards in the garage and under the house. Most Ja- maicans fear lizards and try to kill them or drive them away whenever they encounter them. Lizards are especially dishked around human dwellings. LazelFs lizards were probably quite afraid of humans. I often saw reconditus next to the garage he describes in his paper, in a stand of imported yacca trees. Although I tried more than a dozen times to catch some of them, they would in- variably escape up the straight 40-foot trees. I finally caught one 77-mm female on the roof of LazelFs garage. I was very lucky indeed to do so. I heard reports that the sons of the care- taker of this garage threw stones at the lizards on the trees next to the building. The boys were afraid of the harmless animals, and I could not persuade them of the folly of their actions. The lizards around the garage were definitely "wild." Although Lazell in his paper agrees with Underwood and WilUams (1959) that reconditus usually escapes "under objects and into holes," I found this to be only partially true. In the great majority of cases, I observed reconditus, when attempting to evade my noose, to go up the tree in typical Anolis fashion. On four specific occasions, three times with males and once with 1973 ANOLIS RECONDITUS 17 a female, I watched lizards go into trunk holes, conveniently next to and at least as high off the ground as their perches, to hide from me. I nexer chased a lizard down to the ground or under the exposed roots of a tree, seemingly an adequate hiding place. They invariably went up. REACTION TO WEATHER The "moisture-laden easterly trade winds," which blow all year long bring the Hardwar Gap area of the Blue Mountains over 200 inches of rain annually. During much of the fall and spring, there is daily rain. Although there are occasional dry davs, it usually rains for several hours during the day and some- times for several days continuously. The area is described as "montane mist forest" because of the mist or fog which envelops it every day. The mist remains even during the rain and lasts for 30 to 70 percent of the daylight hours, often forming shortly after dawn and remaining until an hour before sunset (Asprey and Robbins, 1953). Because of the weather conditions and also because of the full canopy of the forest in which the lizards live, it is very difficult to get any idea of how reconditus reacts to heat or sunlight. In all my observation time, I very rarely saw a lizard perched in direct sunHs^ht. This was not because the lizards chose to a\oid it by picking shady perches, but because the sun was rarely shin- ing; if it was sunny, the canopy usually prevented direct sun from striking the lizards. On the few occasions when I did see sunlight on the lizards, they seemed to ha\e no apparent reaction to it. Basking would have been futile since the sun rarely shone more than a few minutes, if that, in one place and, as I said, seldom shone at all. Further study should be carried out during the summer months, which according to the local people are drier, often 80° or more, and sunny. It would be interesting to see if the sun ever gets hot enough in the bush to drive the lizards to seek shade and damp. The local name for reconditus is "water lizard." The animals seem to prefer the damp, often living in mountain-side gulleys where water flows during heavy rains. But to truly grasp the significance of the local name, you have to watch one in a drench- ing downpour: water streams off the body of the lizard, while, with no apparent concern, it remains passively perched. On three separate days of heavy rain I watched a total of seven in- dividuals display no apparent reaction to the precipitation. Four 18 BREVIORA No. 404 of these were females; three were males. One of the females was active during an entire hour of downpour, but her move- ments seemed to have nothing to do with seeking shelter. She was also quite active before the rain and during its tapering- off stages. On numerous other occasions I saw lizards in light rain show no concern for the weather. Only once did I see a female run down a tree and under the exposed roots during a heavy rain. She stopped every few feet during her descent to head-bob. A few minutes before, I had unsuccessfully tried to capture a large male and chased him up into the crown of the same tree. It is not clear whether the female's actions were related to the heavy rain or to the frightened male I chased up the tree. In any case the tolerance for rain which rcconditus seems to have is unusual in anoles. Robert Tri\'c^=, in his extensixe studies of A. garmani and A. vaJencienni in jcii.iaic?, reports (personal communication) that it is nearly impossible m rainy days to find even one individual in an area w'tli a laige, well-studied population. However, Thomas Schoencr (personal communi- cation) has seen A. sagrei perched in the rain. Still, every ob- server of anoles that I have talked to has corroborated the usual pattern as avoidance of the rain. Since A. opal inns is smaller and A. garmani is larger than the "water lizard,'' apparently size cannot be the determining factor in tolerance for rain. The habitat inxoh'ed is undoubtedly im- portant. The tolerance that reconditus has seems to me a rea- sonable adaptation in an environment where wet weather is so frequent. A great deal of energy would be wasted if the lizard had to mo\'e to shelter every time it rained. FORAGING In my many hours of obser\'ation, I saw only three instances of feeding bv reconditus. An 84-mm female was perched se\en and a half feet off the ground, facing down, on a 45-foot X 8-inch tree. I had been watching for half an hour when suddenly she turned 180° around and seemed to take something off the bark with her mouth. Then she "chewed" (opened and closed her mouth) four times and then turned back around to resume the stand- ard downward-facing perch. Half an hour later she ran around the back of the tree and down. A bit of \ine fell from the back of the tree. Then she returned. All of this took place in about a minute. Presumably she had gone after a bit of food. 1973 ANOLIS RECONDITUS 19 On another occasion I watched a 74-mm female traverse six feet of a rotten furrow in a horizontal tree trunk. The trunk was about eight inches in diameter and four feet al)o\c the ground. The Hzard licked the rotten wood inside the furrow, perhaps "tonguing" ants. Then, after assuming a horizontal perch fac- ing out of the furrow at one end of the trunk, the female snapped out of the air a mosquito which was buzzing around her mouth. Ten minutes later, she caught another, larger, flying insect, pre- sumably a fly. Half an hour later she withdrew from this perch back into the furrow and proceeded to tra\erse the length of the trunk again, inside the furrow. She made the trip quite slowly, moving from side to side as if searching for something. The final instance of feeding I observed was by a 95-mm male. I saw him eat some ants while he was making a long, relatively fast-mo\'ing journey of 80 feet through densely forested Fairy Glade (see field notes in home range and territorial struc- ture ) . The eating of ants appeared at the time to be a digres- sion from the business at hand, that is, his long, steady trip up the path. He ate only once in two hours. However, it is not impossible that the whole trip was a food search. Robert Trivers has shown (personal communication) that A. valencienni feeds by acti\'ely searching for prey. Generally Ja- maican anoles maintain a perch until they see a food item, and then go after it. Perhaps the feeding habits of reconditus include both searching and waiting. More study is necessary to deter- mine whether this is true. SLEEPING It seems likely that reconditus, just as A. garmani and se\'- eral other Anolis species, sleeps in the foliage of the crown of trees. The first lizard that I caught at Hardwar Gap was seen at 7:50 a.m. walking down the trunk of a 45-foot tree. The sun was up, of course, but had not yet reached the incredibly steep, densely forested northwest mountainside of Steep Bu'^h. It is possible that this 96-mm male was coming down from his sleeping perch to a lower daytime perch. On two later occa- sions, both at sunset, I put individual lizards "to bed" to see where they went. The first time, I watched a large male {ca. 90 mm) for more than three hours. He maintained until sundown his perch place 12 feet above ground with only occasional sideways movements from the downward-facing perching position. Then, in a period 20 BREVIORA No. 404 of 12 minutes, as darkness was rapidly falling, he went through a flurry of activity and elaborate displays (see display). This culminated in his disappearance into the higher foliage. Two weeks later, I watched a 62-mm female at sundown. She suddenly changed to a crosswise perch from the unchar- acteristic upward-facing perch she had maintained all day. The perch was four feet above the ground on a small tree. Ten minutes later, she jumped to a nearby straight sapling and looked around. Four minutes later she ran up a few inches and looked around. Then she disappeared up the small tree. It seems rea- sonable that a female of this size, indeed the smallest I caught, would sleep in a sapling such as this, a great deal smaller than the large tree, where, as noted above, it is likely that a big male sleeps. DISCUSSION Anolis recondilus is unique and at the same time unspecial- ized. The lizard is apparently isolated from congeneric compe- tition. This may be one of the keys to understanding the mor- pholog)', ecology, and behavior of the animal. E. E. Williams (Williams et al, 1970) makes such a point in his discussion of the solitary Venezuelan lizard Anolis jacare. Recondilus does not occupy any of the typically anoline struc- tural habitats defined by Rand and Williams (1969). Of the six categories they set up, it seems most like a "trunk anole." But it is unlike that category in that it often goes farther up the tree and occasionally down to the ground. Of course, being isolated as it is, there is no reason for it to be restricted to any one structural habitat, as is necessarily the case when several species live syntopically. In addition, the perch that recondilus most frequently takes seems a very reasonable adaptation for a solitary species. From 10 or 12 feet up on a trunk, the animal gets about as good a view of the surrounding forest as is possible and can reach either the ground or crown with equal facility. In the Lesser Antilles, islands with only one Anolis species are common. However, on the larger islands of the Greater Antilles, each of which supports several to many species, there is much overlap of congeneric range. Recondilus in Jamaica is in a situation similar to that of A. jacare on the continent of South America. It is on a "mainland island, just like jacare," at its Blue Mountain retreat. A detailed comparative study of A. recondilus in its montane mist forest and A. jacare- 1973 ANOLIS RECONDITUS 21 in its "Premontane Humid Forest" ^Williams cl al, 1970) might yield some interesting insights into the adaptations of isolated species. Thomas Schoener (1969 and unpublished data) shows that the male snout— vent length (mean of the largest third of his sample) for lizards inhabiting islands of the Lesser Antilles which ha\e only one species of Anolis ranges between 60 and 94 mm. Reconditus, at 97 mm, would fall only a bit beyond the high end of this range. From the few indi\iduals of jacare that have been measured, male size seems to be about 75 mm. Reconditus is possibly the only Jamaican anole which actively searches as well as motionlessly perches in attempts to obtain food. Other than A. sagrei, it is unique among anoles in its ability to function normally in the rain. It does not fit into a Rand and W^illiams' perch category, but instead ranges over se- veral of the structural habitats that they define. No other anole, except A. valencienni (Rand, 1967), is known to crawl into large holes in the tree trunks on which it perches. These features of the beha\ior of reconditus suggest perhaps that the animal is adapted to its habitat in a more general way than some of its aparently more specialized congeners. Recon- ditus appears less restricted in foraging methods, in ability to cope with the weather, and in perch and refuge space than al- most all other anoles. Further study of these and additional aspects of the lizard's behavior and ecology might reveal whether it is indeed less specialized than its congeners. Possibly because reconditus is isolated from congeneric com- petition it is free to fully exploit its habitat. If this is the case, it may be that within its restricted montane mist forest habitat, it can, living alone, use more of the resources of the environ- ment through generalized rather than specialized adaptations. AN EXPLANATION OF THE MARKING SYSTEM I marked lizards at the two main study areas. Steep Bush and Fairy Glade (see habitat and range and map for loca- tion of these places). In each area I used a numerical mark- ing system for permanently identifying individual lizards. A single toe or, in the case of the large toes, just the terminal joint, was clipped off the lizard's foot. The toes were numbered as described below : Starting with the lizard on its \-entral side with its dorsal side facing you, the toes on the front left leg (your left) were 22 BREvioRA No. 404 numbered consecutively 1 to 5 from smallest to largest. The toes on the right front leg (your right) were labelled 6 to 10 from smallest to largest. The toes on the left hind leg were labelled 11 to 15 from smallest to largest. Finally the toes on the right hind leg were labelled 1 6 to 20 in a similar man- ner. Should anyone catch any of these marked animals at Hardwar Gap, I would greatly appreciate his sending me size data (lower jaw, snout-vent, and tail lengths) and any other inter- esting features about the lizards. I can be reached care of Pro- fessor E. E. Williams, Museum of Comparative Zoology, Cam- bridge, Massachusetts. ACKNOWLEDGEMENTS I want to thank especially Ernest E. Williams for introducing me to the study of anoline ecology and behavior. This project would not have been possible without his constant advice and support throughout all stages of the work. Robert Trivers' contributions were invaluable. Not only did he help with details such as initial identification of reconditus in the field but also he generously contributed his ideas and his field techniques. Thomas Jenssen gave me much help and ad- vice while I was in the field. Preston Webster and A. Stanley Rand both provided extensive and valuable criticism on earlier drafts of this paper. I want to thank C. B. Lewis, Director of the Institute of Jamaica, for allowing me to use the Institute's field station at Green Hills for the duration of my study. This study was partly supported by NSF grant B 019801x to Ernest E. Williams. REFERENCES AsPREY, G. F., AND R. G. RoBBiNS. 1953. The vegetation of Ja- maica. Ecol. Monogr., 23: 359-412. Lazelx., J, D., Jr. 1966. Studies on Anolis reconditiLS Underwood and Williams. Bull. Inst. Jamaica, Sci. Ser., 18 (pt. 1): 1-15. Rand, A. 1967. Communal egg laying in anoline lizards. Herpeto- logica, 23(3) : 227-230. , AND E. E. Williams. 1969. The anoles of La Palma: aspects of their ecological relationships. Breviora, No. 327: 1-19. 1973 ANOLIS RECONDITUS 23 SCHOENER, T. W. 1969, Size patterns in West Indian Ayiolis liz- ards: I. Size and species diversity. Syst. Zool., 18: 386-401. Underwood, G., and E. E. Williams. 1959. The anoline lizards of Jamaica. Bull. Inst. Jamaica, Sci. Ser., 9: 1-48. Williams, E. E., O. A. Reig, P. Kiblisky, and C. Riv^ero-Blanco. 1970. Anolis jacare Boulenger, a "solitary" anole of the Andes of Venezuela. Breviora, No. 353 : 1-15. " «>• c. Ob- B R ^ J V I MAR 1 8 )9g5 iTr" iiseuiii of CoBiparal" TJ^ d" y us ISSN 0006-9698 Cambridge, Mass. September 20, 1973 Number 405 A FOSSIL TRTONYGHID TURTLE FROM SOUTH AMERICA Roger Conant Wood^ and Bryan Patterson Abstract. A hyopkistron of u larg-e trioiiychid turtle from the middle Pliocene (Huayquerian) Urumaco Formation of northern Venezuela constitutes the first indisputable record of a member of this family from South America. The reported occurrence of a tri- onychid from the late Cretaceous or Paleocene of Patagonia {Trionyx argentina Ameghino 1899, iiomen dubiuni) cannot be substantiated. The species represented by the Venezuelan fragment was presumably a waif immigrant from Central America that was unsuccessful, very possibly owing to competitive exclusion, in establishing an enduring foothold in South America. Living trionychid turtles are known from Africa, Asia, the Indo- Australian archipelao;o and North Americ;i. Paleontologi- cal e\'idcnce indicates that this group formerly had an c\'en more extensixe distribution. Fossil forms are known from parts of Africa, Asia, and North America where they no longer occur, and remains are found in European Tertiary deposits as well. Until now, with one dubious exception (discussed below), no members of the family, living or fossil, have ever been encountered in South America. Discovery of an uncjuestionable trionyc;hid fragment from late Tertiary sediments in Venezuela during the summer of 1972 is therefore of considerable interest. The specimen (Museo de Ciencias Naturales, Caracas, no. 238) was reco\ered from the upper part of the lower member of the LIrumaco Formation about twcKe kilometers WS^V of Urumaco and .75 kilometer north of Kilometer 153 on the oil pipe line running from Punta Gorda to the Paraguana Penin- 'Stockton State College, Pomona, New Jersey. 2 BREvioRA No. 405 sula. It was found on the surface, together with parts of a pelomedusid that was in the process of weathering out. The Urumaco Formation (Gonzales de Juana ct al., 1970: 612—613) is now believed to be of Huayquerian age (Pascual and de Gamero, 1969; Wood and de Gamero, 1971). Outcrops are restricted to a relati\'ely small area in the northern part of the state of Falcon. The vertebrate fauna, for the most part not )et described (see, in addition to the publications cited above, Royo y Gomez, 1960, and Sill, 1970), consists of numerous aquatic reptiles (some half dozen species of both crocodilians and tur- tles), a testudinine, fish, and \'arious kinds of mammals. The depositional environment was in the main a near-shore marine one, with fluctuations involving a variety of paralic and near- shore continental conditions. The find consists of a large, well-preserved hyoplastron (figure 1 ) . Its sculptured external surface, covered with anastomosing ridges, is characteristic of trion)chids. These rugosities, coupled with the distinctive outline of the bone, leave no doubt regarding the familial reference. Dimensions are as follows: width along the hyo-hypoplastral suture = 22.2 centimeters; length at nar- rowest part = 7.3 centimeters; greatest dorsoventral thickness along hyo-hypoplastral suture '= 2.8 centimeters. From the con- dition of the sutural surface, it is clear that the hyo- and hypo- plastra were distinct bones rather than fused as is the case in various genera of the family. The h)oplastron appears to be unusually thick for its size, the thickness/ width ratio being 0.126. Measurements of shell thickness are scarcely ever given in the literature. A few T/\V ratios can roughly be estimated from the figures and sporadic measurements given by Hay ( 1 906 ) for North American fossil trionychids; these range from 0.083 to 0.019. The hyoplastra of those living forms available to us are as follows regarding this ratio: Lissemys punctata, 0.087; Chitra indica, 0.081 ; Trionyx hurutn, 0.067; T. sinensis, 0.079; T. cartilagincus, 0.090; T. spinifcr (6), 0.107-0.064; T. ferox (4), 0.081-0.070; T. muticus (2), 0.077-0.049. This, of course, is a wholly inadequate sample, but as far as it goes it confirms our impression that the Urumaco hyoplastron is unusually thick. Further, the sample suggests that thickness decreases, relatively, with age; in all species of which we have more than one speci- men the smallest indixidual has the highest ratio and the largest the lowest. The thickness of the Urumaco specimen is not, we accordingly suspect, due to size. This hyoplastron is sufficiently distinctive for reasonable as- 1973 SOUTH AMERICAN TRIOXVCHII) TIRTI.E -*^>'.-':'ic Figure 1. Trionychidae, gen. et sp. indet. Ventral and posterior views of hyoplastron. Scale in centimeters. 4 BREVIORA No. 405 surance that we are dealing with a representative of the group sometimes recognized as the subfamily Trionychinae (group I of Lo\'eridge and Williams, 1957: 414), but quite inadequate for further taxonomic refinement. Assessment of precise relation- ships within the family must await discovery of more complete material. The geographically nearest li\ing species of the family are Trionyx spinifer, whose range extends from the United States down into northeastern Mexico, and T. ater, which is confined to the Cuatro Cienegas Basin of central Coahuila, Mexico (Webb, 1962). Neither attains great size; in fact, the Urumaco form was clearly larger than any of the living American species. Some fossil trionychids {e.g., the late Paleocene Paleotrionyx quinni Schmidt, 1945) are known from North America, how- ever, whose size is comparable to that of the Venezuelan speci- men. The earlier record of a South American trionychid must now be considered. Ameghino (1899: 10) published a very sum- mary "diagnosis" of Trionyx argentina from the "Cretaceous of Patagonia (Guaranitic Formation)." In his only subsequent mention of it, he stated that Tryonix [sic] argentina had come from the "Sehuenian" (1900: 216). No locality was ever given. His "Guaranitic Formation" included marine and continental strata ranging from Cretaceous to early Oligocene ( Deseadan ) , and his "Sehuenian" formed part of it. On the basis of his last statement concerning the age it could be assimied that 7". argen- tina came from the type area of the "Sehuenian" (now the Mata iVmarilla Fonnation, Coniacian; Leanza, 1972: 695, 701) on the Rio Shehuen (or Sehuen or Chalia) in the west-central part of the province of Santa Cruz, and Feruglio (1949a: 257) belie\ed that it had. This is not certain, however. In 1898 (p. 121) Ameghino included in his "Sehuenian," or "Piso Sehuense," deposits occurring in regions other than the type area, among them the Rio Chico del Chubut. It would seem possible that T. argentina could have come from this gen- eral resrion, and there is some indication that it mav ha\e. In the introduction to his paper of 1899 Ameghino stated that all species proposed in it had been collected by his brother Carlos since the manuscript of his paper of 1898 (to which 1899 was a supplement) had been dehvered to the printer, which was done at the end of Julv, 1897 (letter from F. Ameghino to D. de la Fuente, August 7,' 1897; in Torcelh, 1935: 688). By that time Carlos Ameghino's collecting activities had centered on northern Santa Cruz and Chubut. Reporting on the results of 1973 SOUTH AMERICAN TRIOXYCHID TURTLE 5 his expedition of 1898-99 to the Golfo dc San Jorge, he \vrote (letter to Florentine, February 15, 1899; m Torcelli, 1935: 106) that he had found "more or less in the center of the gulf . . . characteristic fossils (sharks, chelonians, crocodiles, etc., the same as those of Sehuen and Lake Argentino)." These che- lonian remains could have included the material on which T. argentina \va.s based. Florentino's statement (1899) that a trionychid had just been found ("acaba de encontrarse" ) could be so interpreted. At any rate, years later, Staesche (1929) described turtle fragments collected, together with crocodilian remains, by von Huene near the center of the gulf, "mainly at Cabo Peligro," and se\'eral of these ans\ver to Ameghino's "diag- nosis" of T. argentina in one respect, \-ermicular sculpturing. Staesche believed the age of these specimens to be Late Creta- ceous, but the bed from which they came was the banco negro inferior at the base of the Rio Chico Formation or perhaps at the top of the Salamanca Formation Feruglio, 1949b: '11. fig. 87), both of which are now considered to be Paleocene in age. Turtle remains, according to Feruglio, commonly occur in this bed or beds in the region of the gulf. Against the possibility that T. argentina may have been reco\ered from this area, how- ever, is the fact that by 1900 (p. 119) Ameghino had begim to make a distinction between the "Sehuenian'" and the beds in Chubut containing Ostrea pyrotherionun (now the Salamanca Formation). As in the case of Niolamia argentina (Simpson, 1938), also described in 1899, no conclusion can now be reached as to the precise age and locaHty of T. argentina. although we regard Paleocene and the gulf as being the more likely alternative. The affinities are even more uncertain. Ameghino"s "diag- nosis" reads, in free translation, "Trionxx argentina n. sp., size small, surface of shell with \-ermicular sculpturing, ossification incomplete, without scute sulci. "^ No information was given as to the type material. This is wholly inadequate to define a spe- cies of Trionyx or e\en to demonstrate the familial assignment. Other turtles have what could be termed \'ermicular sculpturing ; \vhat is meant by "incomplete ossification"" is not clear; and the absence of sulci is not a certain indication that scutes were lack- ing. The specimen or specimens on which the taxon was based "'. . . de talla pequeiia y superficie de la coraza con esculptura vermicular pero de osificacion incompleta y sin surcos externos que indiquen la presencia de escudos ccrneos." 6 BREVIORA No. 405 would appear to have been lost; Wood failed to encounter them in the course of an examination of the turtle remains in the Ameghino collection. Consequently, we regard "Trionyx argen- tina" as a no?nen dubium and we feel that the existence of a late Cretaceous or Paleocene South American trionychid has not been demonstrated. The specimen described here is thus the only definite record of a trionychid for the continent. Representati\es of the family did obtain a foothold there, but for how long and in what man- ner? During 1970 Wood spent four months in South America examining collections of fossil turtles ranging in age from Cre- taceous to Pleistocene and nowhere saw so much as a fragment that could be identified as a trionychid. The extensive collec- tions made in the late Ohffocene and later Miocene of Colombia by University of California parties include turtle remains JDUt no representative of this family. Among the hundreds of turtle specimens in all stages of completeness seen by us in the Uru- maco Formation only this one trionychid fragment was found. ^ On the face of it, therefore, it would seem possible that the in- vasion \vas a Huayquerian event and only briefly successful in the geological sense of the word. Why this should ha\e been so is puzzling, for the great ri\er systems of South America encompass habitats surely suitable for members of the family. Simpson (1943: 423) has suggested that trionychids and che- lids may be ecologically incompatible. Chelids are clearly an old South American group in the sense of Dunn (1931). \Vood has in hand undescribed material from the early Eocene Casa- mayor of Argentina." Undescribed chelids have recently been discovered at Tremembe, in Brazil, now known to be a deposit of early Oligocene, Deseadan, age (Paula Couto and Mczzalira, 1971). Specimens occur in the Oligocene and in the Miocene of Colombia (Stirton, 1953: 614) and have long been known from Parana. We have good material of Chclus from the Uru- maco Formation, and additional undescribed material of this same genus has been found in Pliocene deposits along the upper reaches of the Jurua River in the territory of Acre, Brazil. The Chelidae was not the only family that could have provided 'It was collected midway in the season and a very close watch for others was maintained thereafter. ^Ameg-hino "diag-nosed" three species of Platemys from his "Guar- anitico" in 1899, but these records are vominn dnbia and the speci- mens evidently lost. 1973 SOUTH AMERICAN TRIONYCHID TURTLE 7 competition to inxading aquatic turtles.^ Pelomedusids, also Old South American forms, are known in the continent from the late Cretaceous on and are \ery numerous in the Urumaco deposits. The demonstrated failure of trionychids to maintain themselves in South America could be interpreted in terms of competitive exclusion. Chelids and pelomedusids ha\'e evolved together in South America throughout the Cenozoic, and in the course of this time no doubt parceled out between them a wide variety of habitats. That they would have completely saturated the con- tinent is unlikely, but such diversification would have lengthened the odds against an invader. The species represented by our Urumaco trionychid may well have had the misfortune to en- counter upon arrival a comparably adapted resident and have failed to compete successfully with it. It was, indeed, similar in size to the commonest Urumaco pelomedusid, Podocnertiis ven- ezuelensis Wood and de Gamero. Such mammalian remains as occur in the Urumaco Forma- tion are all of South American type, a finding in accord with Argentinian evidence suggesting a Montehermosan date for the joining of the American continents. On this basis, then, the \^enezuelan trionychid would appear to have been a waif immi- grant carried by ocean currents. Members of the family may well be rather prone to such accidents. There is some evidence that several of them enter brackish or even salt water (summary and references in Neill, 1958: 26-27). These would be par- ticularly liable to hazards of this kind, but tolerance for saline waters need not necessarily be prerequisite. The Afiican Trionyx tri unguis also occurs in rivers and swamps along the eastern Mediterranean littoral. Flower (1933: 754) has suggested that this extension results from individuals being swept to sea by Nile floods and carried eastward by the prevailing current. The ancestor or ancestors of the Venezuelan trionychid presumably came from Central America, the nearest land area, and arri\ed in South America at approximately the same time as did the extinct Cyonasua group of the Procyonidae. No trionychid now ^At the present time there is one area in which chelids and tri- onychids come together. This is southern New Guinea where the wide-ranging- trionychid PelocJichjs bihroni has established itself — i-ather recently in Darlington's opinion (1957: 210). There it is in contact wdth several chelids. So far as we are aware no infonnation is available on possible interactions, but direct competition would seem unlikely. P. bibrojil is a large species, much larger than any New Guinean chelid. 8 BREVIORA No. 405 lives in Central America — it is a decidedly odd fact of distribu- tion that all living North American species of Trionyx, in con- trast to a number of their Old World congeners, are extratropi- cal. None has been recorded fossil there, but in view of our abysmal ignorance of the Tertiary vertebrates of the region no weight can at present be given to their absence.' This note records a momentary success in range extension resulting from waif dispersal. Such transitory events must surely have been commoner than those more enduringly successful, but in the nature of things the chances are against their entry into the fossil record. They could have played a part in furthering distribution. In the case of organisms such as freshwater turtles each new toehold gained, for however brief a time, in a suitable environment would have provided a potential base for further waif dispersal. The distribution of pleurodirans, for example, may have been to some degree forwarded in this way. ACKNOWLEDGMENTS Our field work in Venezuela was made possible by National Science Foundation Grant no. GB-32489x to Patterson, and by the generous cooperation of the Escuela de Geologia, Universi- dad Central de Venezuela, and the Ministerio de Minas y Hidrocarburos. We are indebted to Profesora Maria Lourdes Diaz de Gamero, Dr. Clemente Gonzales de Juana, Dr. A. Bellizzia G., Professor John Gibson-Smith, Dr. Edgardo Mon- dolfi, Sr. Bias Gonzalez, and our companions in the field : Messrs. 'Webb (1962: 584) has credited Mullerreid (1943) with recording a fossil trionychid of unknown provenance from Mexico. Actually, that author imentioned seeing the internal mold of an unidentified turtle. The owner of the specimen, which has probably been lost, stated that he had found it on the Mesa de Capolla, a short distance SSW of Tuxtla Gutierrez, Chiapas. Cretaceous and early Tertiary sediments are present there, and Mullein-eid believed that the mold had come from the latter. With the exception of the Oligocene or Miocene testudinine Geochelone costarricensis (Segura) (Auffenberg, 1971), this is the only published record of a Tertiary turtle from Central America. (In addition, we know of undescribed material from two Tertiary localities in Panama; no trionychids are included among the specimens.) Aguilera (1907: 241) listed Tryonix [sic] sp. as occurring- with marine invertebrates in Turonian deposits at Peyotes, Coahuila. This is the only North American fossil record of the family south of the United States. It requires confirmation, and is in any event within the present range of T. spinifer emoryi. 1973 SOUTH AMERICAN TRIONYCHID TURTLE 9 Arnold D. Lewis (the finder of the specimen here discussed), Daniel C. Fisher, Robert W. Repenning, and Michael F. Stan- ford. The participation of Messrs. Fisher, Repenning, and Stanford was due, respectively, to the generous provision of a summer scholarship from the Department of Geological Sciences, Har\-ard Uni\ersity, and of grants from the Shell Oil Company and the Creole Petroleum Corporation. W^ood's work in South America during 1970 was supported by the National Geographic Society. REFERENCES CITED Aguilera, J. G. 1907. Apergu sur la geologie du Mexique pour servir d'explication a la carte geolog'ique de I'Anierique du Nord. Compte Rendu lOieme Sess. Cong. Geol. Internat., Mexico: 227- 248. Ameghino, F. 1898. Sinopsis geologico-paleontologica de la Re- publica Argentina. Segundo Censo de la Republica Argentina, 1: 111-255. . 1899. Sinopsis geologico-paleontologica. Suplemento (adiciones y correcciones). La Plata: Imprenta la Libertad. 13 pp. 1900. L'age des formations sedimentaires de Pata- gonie. An. Soc. Cient. Argentina, 50: 109-130, 145-165, 209-229 (part). AUFFENBERG, W. A. 1971. A new fossil tortoise, with remarks on the origin of South American testudinines. Copeia, 1971: 106-117. Darlington, P. J., Jr. 1957. Zoogeography: the Geographical Distribution of Animals. New York: John Wiley, i-xiii + 675 pp. Dunn, E. R. 1931. The herpetological fauna of the Americas. Copeia, 1931: 106-119. Feruglio, E. 1949a. Descripcion geologica de la Patagonia. Tomo I. Buenos Aires: Coni. i-xv + 334 pp. . 1949b. Descripcion geologica de la Patagonia. Tomo II. Buenos Aires: Coni. 349 pp. Flower, S. S. 1933. Notes on the Recent reptiles and amphibians of Eg>T)t, with a list of the species recorded from that kingdom. Proc. Zool. Soc. London, 1933: 735-851. Gonzalez de Juana, C, A. Salvador, R. M. Stainforth, G. A. Young, F. DE RiVERO, C. Martin Bellizzia, C. Petzall (eds). 1970. Lexico estratigrafico de Venezuela (segunda edicion). Rep. Venezuela, Minist. Min. Hydroc, Dir. Geol., Bol. Geol. Pub. Esp. No. 4: 1-757. Hay, 0. P. 1906. The fossil turtles of North America. Publ. Carne- gie Inst. Washington, No, 75: 1-568. 10 BREVIORA No. 405 Leanza, a. F. 1972. Andes patag-onicos australes. In. A. F. Leanza, Ed., Geologia Reg-ional Argentina. Cordoba: Academia nacional de Ciencias, pp. 689-706. LOVERIDGE, A., AND E. E. WiLLlAMS. 1957. Revision of the African tortoises and turtles of the suborder Cryptodira. Bull. Mus. Comp. Zool., 115: 163-557. MULLERREiD, F. K. G. 1943. Fosiles raros de Mexico III — Una tor- tuga fosil del estado de Chiapas. An. Inst. Biol. Univ. Nac. Mexico, 14: 623-624. Neill, W. T. 1958. The occurrence of amphibians and reptiles in saltwater areas, and a bibliogi-aphy. Bull. Marine Sci. Gulf Caribbean, 8: 1-97. Pascual, R., and M. L. Diaz de Gamero. 1969. Sobre la presencia del genero Eumegamys (Rodentia, Caviomorpha) en la foiTn- acion Urumaco del Estado Falcon (Venezuela). Su significacion cronologica. Bol. Infonnativo, Assoc. Venezolana Geol. Min. Pet., 12: 369-387. Paula Couto, C. de, and S. Mezzalira. 1971. Nova conceituagao geocronologica de Tremembe, Estado de Sao Paulo, Brasil. An. Acad. Bras. Cienc, 43, Suplemento : 473-488. ROYO Y Gomez, J. 1960. Los vei'tebrados de la formacion Urumaco, Estado Falcon. Mem. Ill Cong. Geol. Venezolano, 2: 506-510. Schmidt, K. P. 1945. A new turtle from the Paleocene of Colorado. Fieldiana: Geol., 10: 1-4. Sill, W. D. 1970. Nota preliminar sobre un nuevo gavial del Plio- cene de Venezuela y ima discusion de los gaviales sudamericanos. Ameghiniana, 7: 151-159. Simpson, G. G. 1938. Crossochelys, Eocene horned turtle from Patagonia. Bull. Amer. Mus. Nat. Hist., 74: 221-254. — ■ . 1943. Turtles and the origin of the fauna of Latin America. Amer. Jour. Sci., 241: 413-429. Staesche, K. 1929. Schildkrotenreste aus der oberen Kreide pata- goniens. Palaontographica, 72: 103-123. Stirton, R. a. 1953. Vertebrate paleontology and continental stratigraphy in Colombia. Bull. Geol. Soc. America, 64: 603-622. TORCELLI, A. J. (ed.) 1935. Obras Completas y Correspondencia Cientifica de Floi'entino Ameghino. Volumen XXI. Corres- pondencia Cientifica. La Plata: Taller de Impresiones Oficiales. 935 pp. Webb. R. G. 1962. North American Recent soft-shelled turtles (family Trionychidae). Univ. Kansas Publ. Mus. Nat. Hist., 13: 429-611. Wood, R. C, and M. L. Diaz de Gamero. 1971. Podocnemis vene- zitelensis, a new pelomedusid (Testudines, Pleurodira) from the Pliocene of Venezuela and a review of the history of Podocnemis in South America. Breviora, Mus. Comp. Zool. No. 376: 1-23. vlpO^ -,«■•' ■ ^^^I'S. CO MP. ZOOL LfPRARY B R A V I WW A iiseiim of Comparative Zoology us ISSN 000(1- '.)(i'.)S Cambridge, Mass. September 20, 1973 Number 406 TWO NEW LYGOSOXIINE SKINKS FROM NEW GUINEA WITH COMMENTS ON THE LOSS OF THE EXTERNAL EAR IN LYGOSOMINES AND OBSERVATIONS ON PREVIOUSLY DESCRIBED SPECIES Allen E. Greer, Jr. AiiSTKAcr. Two new species of the jasciatus species group of Spheno- murplnis are desciibcd from New Cuiiica: S. anotus is unique among its relati\es in lacking an external ear opening and i. inicrutympanus can be disiinguislietl 1)\ tlie greatly reduced size of its tvnipanuni. The earless l)gosomines are reviewed and it is noted that, while the loss of the external ear opening is apparently a prerequisite for a burrowing way of life, ii is inicertain whether the loss of the external ear in burrowcrs is a piimary adaptation to burrowing or whether it is a preadaptation inherited fioui a nonburrowing ancestor. Certain aspects of the morphology of the previously described but poorly known sjjecies Splicuoinorplius forbcsi, S. olii^olcpis and i'. schuUzei are discussed antl photographs of type specimens are provided. In 1964 members of the Se\enth Archbold Expedition col- lected two small scincid lizards on the Huon Peninsula which differed from all other Icnown skinks in New Guinea in having a scaly auricular depression instead of the more external ear opening. These were thus the first "'earless" skinks to come out of New Guinea. The specimens were sent to Dr. Richard Zweifel at the American Museum of Natural History, but he was unable to identify them and he put them aside in the iiope that more specimens would be forthcoming. Five years later two more very similar skinks were collected by Angus F. Hutton at Garaina and gi\en to Dr. Zweifel during his 1 9(59 trip to New Guinea. Dr. Zweifel recognized the great similarity between these two specimens and the Huon Peninsula specimens but, still being unable to identify them, he kindly turned them o\'er to me for further study. 2 BREVIORA No. 406 On close examination the two specimens from the Huon Peninsuhi pro\e to be members of the fasciatus species group of Sphenomorphus (^ Greer and Parker, 1967), but they difTer strikingly from all known members of this species group in being "earless." The two Garaina specimens are also clearly members of the fasciatus species group and are indeed similar to the Huon Peninsula specimens. They differ from the Huon Peninsula specimens, however, in having a relatively small, but nonethe- less distinct, tympanum instead of a scaly auricular depression, and they differ from all pre\iously described members of the fasciatus species group in the extreme reduction in the size of the tympanum. The Huon Peninsula and Garaina specimens are thus distinct enough from each other and from their closest relatives in the fasciatus species group to be described as new. Description of Two New Species Sphenomorphus anotus new species Figure 1 Hololype. American Museum of Natural History 95880; an adult collected on 5 May 1964 by Hobart M. Van Deuscn and Stanley O. Grierson in the Morobe District of the Territory of New Guinea at MASBA GREEK (Figure 4) at an elevation of approximately 2000 feet. For an account of this locality see Van Deusen (1966). Paratype. AMNH 95881; a badly mangled young juvenile collected by Van Deusen and Grierson on 7 May 1964 at the same locality as the holotype. Diagnosis. This species is a typical representative of the fasciatus species group of Sphenomorphus^ (Greer and Parker, 1967) except that it has a scaly auricular depression instead of the more usual external ear opening. In other words, it is the only "earless" member of the fasciatus species group known to date. Etymology. The species name anotus calls attention to the absence of an external ear {an — without and otus — ear). 'The diagnostic features of tlie fasciatus species group of Splieiioniorphus are as follows: digits and limbs usually well developed but the limbs generally not overlapping when adpressed to the body; frontal in contact with the two anteriormost supraoculars; generally four supraoculars; a single anterior loreal; no supranasals; usually a series of two or more paired nuchal scales; generally 36 or fewer scales around midbody, the scales of the paravetebral rows being larger than the scales of the more lateral rows; generally a postorbital bone that is usually long and thin. 1973 EARLESS SKINKS B Figure 1. Dorsal (A) and lateral (B) view of the head of the holotype of Sphenomorphus anotus (AMNH 95880) from Masba Creek, Huon Peninsula, New Guinea. Description of the holotype. A small attenuate skink mea- suring 48 mm in snout-\ent length with a complete tail mea- suring 68 mm in length; head bluntly conical; hmbs short, pentadactyl, widely separated when pressed against the body; uniformh' brown above and light yellowish brown below (in preser\ati\e ) . Rostral about as deep as wide and projecting well onto dorsal surface of snout; nasal large with external naris situated well forward and ventral in nasal scale; no supranasals; frontonasal wider than long and forming a short suture with rostral and a slightly wider suture with frontal; prefrontals separated; single anterior and posterior loreals; frontal slightly longer than wide and in contact with two anteriormost of four supraoculars; lower eyelid movable and scaly; frontoparietals and interparietal dis- tinct, approximately subequal in size, and measured together along midline, about equal in length to frontal measured along midline; parietals meet behind interparietal; two nuchal scales on left side and four on right; six supralabials, fourth situated most directly below eye ( Fig. 1 ) . 4 BREVIORA No. 406 External ear opening lacking and in its place an auricular depression completely lined with small scales; body scales smooth and in 26 longitudinal rows at midbody; scales of paravertebral scale rows slightly wider than other dorsal scales at midbody and numbering 71-72 in distance from parietals to midpoint of in- sertion of hind legs; medial pair of preanal scales enlarged; medial row of scales on underside of tail only slightly larger than adjacent lateral rows; 8-9 obtusely keeled lamellae beneath fourth toe; fourth toe covered dorsally by a single row of scales on distal third of length, by two rows of scales over medial third, and by three rows over proximal third (Group III of Bron- gersma, 1 942 ) . Color of the holotype. No color notes were made on the speci- men in life. In preservative, however, the dorsum is a uniform light chocolate brown while the venter is a light yellowish brown anterior to the vent and a slightly darker brown posterior to the vent. There is no sharp transition line between the dark dorsal color and the light ventral color. There is only the slightest trace of scattered dark spotting on the throat. Variation in the paratype. The single paratype of Spheno- morphus anotus is a small (snout-vent length = 24 mm), badly mangled specimen obviously of very young age. In the char- acters that can be evaluated it differs but little from the holo- type: there are 26 midbody scale rows, the fourth supraocular lies most directly beneath the eye, there are 3^ nuchals, and most importantly, there is a scale-lined auricular depression instead of an external ear opening. In color the paratype is similar to the holotype but it lacks the yellowish wash to the venter. Distribution. Sphenomorphus anotus is known only from the type locality on the Huon Peninsula of New Guinea (Fig. 4). Habitat. The Masba Creek locality where the two types were caught is in a "stretch of unbroken rain forest" (Van Deusen, 1966) and both animals were taken as the litter was being scraped level for the tents and work flys. Thus it would seem that, like other members of its species group, S. anotus is a cryptic burrower in the litter. Relationships. S. anotus appears to be very closly related to the following species, but a discussion of the relationships of both forms is deferred to the end of that species' description. 1973 EARLESS SKINKS Figure 2. Dorsal (A) and lateral (B) view of the head of the para- type of Sphenomorphus micro tympanus (MCZ 132767) from Garaina, New Guinea. Sphenomorphus microtympanus new species Figures 2, 3, and 5 (top) Holotype. AMNH 104076; an adult collected on 7 July 1969 by Angus F. Hutton in the Morobe District of the Terri- tor\' of New Guinea at GARAINA (Fig. 4) at an elevation of approximately 2300 feet. Paratype. MCZ 132767; same data as the holotype. Diagnosis. S. microtympanus is a member of the jasciatus species group of Sphenomorphus and in that it lacks an ecto- pterygoid process to the palatine running along the outer edge of the palatal ramus of the pterygoid it is most similar to the jasciatus subgroup of that species group (Greer and Parker, 1967). It differs from all members of its species group, how- ever, in having the tympanum both much reduced in size and decidedly more opaque (thickened?). Etymology. The name microtympanus calls attention to the relatively small size of the tympanic membrane of the species. Description. Since S. microtympanus is so similar to S. anotus just described, I will only give specific counts and mea- BREVIORA No. 406 1973 KAKI.KSS SKIXK.S 7 surcincnts f(jr >S'. yn'urolyynpdnus and will tl('s(iil)e im\\ tfiovc aspects of the species" niorpholog) in whicli it dilTers noti( cabU from S. anolus. Both the holotxpc and j:)aratype ha\e a snout \ent len^ili of 45 mm; in the paratype the tail is broken but in the t\jxj ii i.s complete and measures 60 mm. Both specimens of .V. tnirru- tympanus are a richer chocolate brown abo\c than .S'. anulus and in preser\'ati\e both lack the yellowish wash on the under- sides shown b\- the holotypc of S. anolus. These color differences may, however, be an artifact of preservation. There is a scaly auricular depression \er\- similar to the auri- cular depression of S. anolus, but at the bottom of the depression there is a small, opaque t)mpamnTi instead of scales a.s in S. anolus (Fig. 2). Both type specimens of S. microlympanus have four pairs of nuchal scales, and, in three out of the four cases, there are six supralabials with the fourth situated most directly below the eye; on the rioht side of the head in the paratype there are seven supralabials and the fifth is under the eye. There are 26 scale rows at midbody and the scales of the two mid-dorsal rows number 75 in the paratype and 71 in the holotype when counted from the parietals to the midpoint of the insertion of the hind legs. The subdigital lamellae on the fourth toe number 8-10. Color. In addition to being richer brown above and lacking the yellowish wash below, the type and paratype of S. micro- lympanus differ from S. anolus in ha\"ing a very noticeable brown wash on the throat and chest instead of a ver\' faint brown wa-sh limited to the throat as in the type of S. anolus. This wash is much more pronounced in the paratype of .?. microl\mpann'i than in the holot)pe. Dislribulion. S. microlympanus is known at present only from the type locality (Fig. 4). Habilal. According to Dr. Zweifel (letter, 14 February 1973), the ''undisturbed habitat around Garaina is rain forest on river terrace and foothills." Figure 3. '1 lie liolt)i\pe of Sl)linio»ini plnis niii rolynijxniu.s lAMXH 104076) from Garaina. Xcw Ciuinca. 1 lie specimen has a snout-veni length of \'i mm and a tail length of HO mm. 8 BREVIORA No. 406 ^::z^ ^> Figure 4. Map of eastern New Guinea showing the type localities of Sl)ln')Wiii()i j)hus aiiuliis (Masba Creek = dosed star) and of S. iiiicro- t\)iil>auus (Garaina = open star) . Tiie two species are known only from their type localities. Relationships between S. microtympanus and S. anotus. These two species are so similar in all aspects of their external mor- phology, sa\e for the nature of the external ear, that they are almost certainly each other's closest known lixing relati\es. Indeed, it looks as if S. anotus could have ea.sily e\ol\ed from a microtxm panusAxkt ancestor simply by ha\ing the scales on the sides of the auricular depression extend down and oxer the \ery small tympanum at the bottom of the depression. The relationship between these two taxa appears to be so close that I onginally thought it might be possible to describe them as the same species. To do so would ha\c required only that one bclicxe that the \ariation shown in the external ears of the specimens exists within a single species. But this kind of variation is unknown in better studied skink species [e.g., the skinks of the earless genus Heyniergis or the many species of Lerista, which ha\e minute ear openings), and to conclude 1973 EARLESS SKINKS 9 that it exists within or between the populations represented by these four specimens seems presumptuous.' Relationships with other skinks. Wiiiiin the fasriatus species group of Sphenomorphus there are only four other previously described species that arc like ynierotympanus and anotus in possessing the following suite of characters: relatively small size (maximum snout-\'ent length 55 mm or less) ; more or less uniformly dark dorsal color; a moderate number of midbod\- scales (overall range, 20-28; range of modes, 24-26), and a low number of subdigital lamellae on the fourth toe (upper limit of range not exceeding 16). All iouv species occur in New- Guinea and in the order discussed below they are forbesi, sehultzei. beaujorti, and olioolepis. Comparati\e data for these four species plus ynicrotxmpdnus and anotus are presented in Table 1. Forbesi. On the basis of palatal morphology, forbesi seems quite distant from ynierotympanus and presumably also anotus, although I ^vas not able to examine the palate of this last species. In forbesi there is an ectopterygoid process to the palatine which excludes the palatal ramus of the pterygoid from a posi- tion on the infraorI:)ital \'acuitv. In rnierotympayius and pre- sumably also in anotus there is no ectopter\goid process and the palatal ramus of the pter\ooid enters the infraorbital vacuity. The presence or absence of the ectopter\goid process may reflect a basic e\olutionar\' dichotomx' in the fawiatus species group ' soloynonis subsirotip \s. the jasciatus subgroup — fide Greer and Parker. 1967) and on this basis alone I would exclude 'Thcic is f)ne pooilv .iii.il\/rrei tens' (MHi'ii pnhlislied e\iden(e is nothing more than a hrief deseiiption and iinich disnission of one spednien ({\\v onl\ one fi(ini that lo(alit\) wliirli he said looked like ji(niii(nii( ii\ hut whiili lacked an e\lemal ear opening, the ke\ (haiacler of '^vaxdiim. I nlin himself exainiiu'd a total of oid\ six s|)((iinens of hoth spetinicais (llirte jxiiiiKitiif ii\ fiom one lo(alit\ and tlnee ^i(i\fmii<: ta~ _>- 0! i-l o o o" o z Z Z z >■ >i .^ o ~ ir. — X X 1<5 3 1^ cr-, — X C-. ■^ X !M ?i ?. s 1973 EARLESS SKINKS 11 forbcsi from close relationship with micrulyrnpanus and anolus. Schultzei. The palate of schullzei is similar to that of micrn- tyyyipanus in that it lacks an ectoptervgoid process, but other features of its morphology cause mc to exclude it from the close relatives of ynicrotympanus and anotus. The most notable of these features is the unique fusion of the first supralabial and nasal scales (see below), the medially meeting prefrontals (in most specimens), and the proportionately longer legs. Beaujorli. As far as I can tell, beauforti is known only from the type specimen (de Jong, 1927) and I have not seen this specimen. I feel, howexer, that the absence of nuchal scales and the medially meeting prefrontals are enough to make beau- forti an unlikely near relative of microtympanus and anotus. OUgolepis. In contrast to the preceding three species, oligolepis is in every way a perfect candidate for the closest living relative of ?nicrotympanus and anotus. The palate of oligolepis is very similar to that of rtiicrotyyyipanus and presumably also to that of anotus; all three species are similar in size and body proportions, and there is no significant difference in the general details of squamation. Oligolepis differs markedly from micro- tympanus and anotus only in having a well-defined external ear opening and ear canal at the bottom of which is a trans- lucent tympanum (Fig. 5), but since this kind of ear was undoubtedly primitive for the microtympanus^ anotus line, it simply serves to make oligolepis the closest living species, morphologically, to the ancestor of that line: oligolepis^ microtympanus-^ anotus. Comments on the Loss of the External Ear in Lygosomines With the desecription of Sphenomorphus anotus, the total number of known "earless" lygosomines comes to 33. This is about 5 percent of the total number of known species in the subfamily.^ Taking a \ery conservative view of the species relationships, I beliexe that these 33 species represent no fewer than ten dif- ferent lineages. Or, to put it another wav, the external ear has been lost at least ten difTerent times in the ex'olutionarv history 'This pcKcntase is vci v low compared to the other three subfamilies of skinks. All of the fcylinines (4 species) and acontines H") species) lack an external ear opening and just under 2") percent of tlu- scincines, of wliich there are a total of approxiuKUclv 182, are also "earless." 12 BREVIORA No. 406 1973 EARLESS SKINKS 13 of the living lygosomines. The species in these ten groups, along with their distributions and other pertinent data, are listed in Table 2. Unfortunately, it is difficult to say anything very conclusive about why the skinks in these different groups have Tost the external ear, but I can make a few comments and suggestions for further research along these lines. First, there is probably no one unifying reason for the loss of the external ear in all ten groups since there is nothing in the biology of these skinks beside the absence of an external ear that sets them apart from other lygosomines. It is true that all the earless species are in some sense cryptic in their habits but this is the rule rather than the exception for skinks. Second, the only outstanding ecological feature of any of the species in the list of earless lygosomines is that certain of the species, i.e., Isopachys, most of the australis group and perhaps the sumatrense group, appear to be the most confirmed burrow- ers among lygosomines. This fact indicates that the absence of the external ear is probably a prerequisite for an in-depth evolu- tionary commitment to burrowing life, but there is no way of knowing whether the loss of the external ear in these skinks was achieved as a primary adaptation to burrowing life or whether it was a preadaptation, i.e., originally evolved for other reasons in nonburrowing ancestors. The large number of ear- less lygosomines that show no exceptional procUvity to a bur- rowing life, e.g., the quadrivitattum group, Anotis mariae, and Ablepharus grayanus, would argue that the loss of the external ear could be as much a preadaptation to burrowing life as it is a primary adaptation. Third, Minton (1966) has suggested that, along with the ablepharine eye, the absence of an external ear opening in Able- pharus grayanus is a protective adaptation that allows this species to feed unmolested on the ants that are said to form much of its diet. This is an interesting idea, but to be con- \incing, it will have to be shown more rigorously than it can now be shown that A. grayanus is more of an ant specialist Figure 5. Lateral view of the head of Sphenomorphus microtympanus (top; paratype: MCZ 132767) , S. oligolepis (middle; syntype: BMNH 1946.8.3.47) . and 5. solomonis (bottom; syntype: BMNH 1946.8.34-37) . Note the relatively small external ear opening and small tympanum of S. micro- t\>iij>n)i)is compared to its dose relative S. oligolepis and its more distant species gioup relative S. solomonis. 14 BREVIORA No: 406 c" c E ^ O C 5 O o n ■^ ^ c Cj .^T" ^ ;^ ~ W ^ T" 3d J'. r: O "Eh ~ ^,, rj t/: ^1^ c ^ o "c? S^ be C = O %/i c CJ o «J o ^ r* o — *- *^ F— < ^J I_ C c c r: . o be V3 E Hf C '^ r3 ^ O -;^ u f/: c c n o rs — O r^ o o w • M c 7^ •:<: Vl _' ^— H rt c c O 'e E ^ ■"■ o — r: z^ (/: o c , . c/: ^ be ■y: ^ Sc ?— t *-< r: u > c« «! O o y: — ^ tf: rt o E '^ 'X * c — o c ly; O ;/; .iii ^ Cj Z. -f^ ^J 'u ^ ./: o ^ O a, "w 'm. sz w ..■ ^ ■^^ 5" <;*. r- (« V >-> c^". t/: *w o ^ ^ ~ CM ^ o r- be ? c o a ~ - v: - - be - - - ;£ .- C;^ - ^ r: = . ■ CO -c - — •- a" -S ^"^ _ ^ —. ly: ^ _ = o v-' j; o .s ~ c r^ o jr = 5 cr. .r: * — ""^"^ ^ ^' y. 1 >-. Cj ■~ :^ be C/j 1 1 "~ ' z' '• ^ be •^ be ^- ._= ■< .E — ■ - ^ ^^ ''^- '"" r; _o *; r3 cr "C ^ v. p < ■^ ^ ■§ o o o c 5c "5: ? s^ ' - i s a 5 s :s, Ch C3 •'-- be t/3 c C ^ c/2 c/; c S z ;z > ■-/; o o i* ;z 'o Z ^ o \^ r3 rt ;z z '^ rz S z: n - -z o o CJ = c C> J-. z:l ^> ^ ^> — /-«/ w c ^ c/; C O an "^ n.' ;: o ;; v fa' ^, = Z = 2 Z J7: U U o C ;j o >^ c/; :/2 oc ■Xl O 00 1^ 00 00 GO '/) -J" = X 00 X 'k- ^ — ^ ^^ — o ■ ■ v: ->: o ^^ ~ ■r- ^ <- o ^, ^ s ^ C- 2 2' 'be 5. £/: ti" i^ c ■^- ri K i 00 — _ -^ -2 2 •^ -w -o o c CO so C-5 1^ „ C"' 00 1^ 00 oo CO — ^" w ?. ■-!_ o '■J o t ~ •£ -^ c o a, o rt o ^ i o S Cm C: CI. o O ^ be to 1973 EARI.KSS S KINKS 15 o ;•<■ "!;! > -' C o -c c "^ - O r- cr. z _— -7, 4J Y', 2 _• ^ !r 2 c - c "5 tiC r.S. Thesis, University of Qucenslantl. 246 pp. BoL'LENGER, G. A. 1888. Descriptions of new reptiles and batrachians ob- tained by Mr. H. O. Forbes in New Guinea. Ann. Mag. Nat. Hist., Scr. 6, 1 (5) : 343-346. ■ . 1897. An account of the reptiles and amphibians col- lected by Dr. L. Loria in British New Guinea. .\nn. Mus. Civ. Stor. Nat. Geneva, Ser. 2, 17: 694-710. . 1900a. Descriptions of new reptiles from Perak, Malay Peninsula. Ann. Mag. Nat. Hist., .Ser. 7, 5(27) : 306-308. . 1900b. Descriptions of two new lizards from Selangor. J. Eombav Nat. Hist. Soc, 1.^(2) : 333-334. . 1903. Report on the batrachians and reptiles: pp. 131- 176 in N. .4nnandalc and H. C. Robinson. Fasticuli Malayensis. Zoology. Part 1, .303 pp. 1908. Report on the Gunong Tohan Expedition, May- Septcniber, 1905. IIL Fishes, batrachians and reptiles. J. Fed. Malay States Mus., 3: 61-69. 1914. An aiuiotatcd list of the batrachians and reptiles collected by the British Ornithologists' Union Expedition and the ■Wollaston Expedition in Duidi New Guinea. Trans. Zoo). Soc. London, 20 (.-)) : 247-274. 24 BREVIORA No. 406 BoLRRET, R. 1937. Notes hcrtctologicjucs siir I'lndochine frangaisc. XV. Lc'-zarcls ct serpents rcgiis au Laboratoire des Sciences Natuielles de rUnivcisite au cours de I'annee 1937. Descriptions de deux espcces et de deux varietes nouvelles. Bulletin General de I'lnstruction Publique (Hanoi) , No. 4: 57-80. . 1939. Reptiles et batraciens requs au Laboratoire des Sciences Naturclles de lUniversitc au cours de I'annee 1939. XVIII. Descriptions de quatrc especes et dune variete nouvelles. Bull. Gen. Instr. Publq. (Hanoi) , No. 4: 5-39. Brongersma, L. D. 1942. On the arrangement of the scales on the dorsal surface of the digits in Lygosoma. and allied genera. Zool. Meded., 24 (1-2) : 153-158. Brown, W. C., and A. C. Alcala. 1961. A new sphcnomorphid lizard from Palawan Island, Philippines. Occ. Pap. California Acad. Sci.. No. 32: 1-4. Bustard, H. R. 1964. Reproduction in the .'\ustralian rain forest skinks, Siaphos eqiinlis and Sphenomorphiis tryoni. Copeia. 1964(4): 715-716. Copland. S. J. 1952. A mainland race of the scincid lizard Lygosoma truncatum (Peters). Proc. Linn. Soc. New South Wales, 77(3-4): 126- 131. Dumerii,, a. 1851. Catalogue nicthodique de la collection des reptiles. Mus. Hist. Natur. Paris, iv -\- 224 pp. FuHN, I. E. 1969. Revision and redefinition of the genus Ablepharus Lichtenstein, 1823 (Reptilia, Scincidae) . Rev. Roum. Biol. (Zool.), 14(1) : 23-41. Gray, J. E. 1825. Synopsis of the genera of reptiles and amphibia, with a description of some new species. ,\nn. Phil., Ser. 2, 10: 193-217. Greer, A. E., and F. Parker. 1967. A new scincid lizard from the northern Solomon Islands. Breviora, No. 275: 1-20. GuNTHER, A. 1873a. Notes on some reptiles and batrachians obtained by Dr. Adolf Bernhard Meyer in Celebes and the Philippine Islands. Proc. Zool. Soc. London, 1873: 165-172. . 1873b. Notes on, and descriptions of, some lizards with rudimcntarv limbs in the British Museum, .^nn. Mag. Hist.. Ser. 4, 12(68) : 145-148. Jong, J. K. de. 1927. Reptiles from Dutch New Guinea. Nova Guinea, 15(3) : 296-318. KorsTEiN. F. 1926. Reptilien von den Molukken und den benachbarten Inseln. Zool. Meded.. 9: 71-112. Longman, H. A. 1916. Snakes and lizards from Queensland and the Northern Territorv. Mem. Queensland Mus., 5: 46-51. Mertens, R. 1965. Bemerkungen iiber einige Eidechsen aus .Afghanistan. Senck. biol., 46(1) : 1-4. MiNTON, S. A., Jr. 1966. A contribution to the herpetology of West Pakistan. Bull. Amer. Mus. Nat. Hist.. 134(2): 27-184. Peters. W. 1867. Herpetologische Notizen. Monatsber. Berlin Akad. Wiss., 1867: 13-37. 1973 EARLESS SKINKS 25 . 1873. Einc Mitlheilung iibcr neue Sauricr (Spaerioclactylus, Anolis, Phrynosoma, Tropidolepisma, I.ygosoma, Ophioscincus) aus Ccntralamtiica, Mexico unci Austialien. Monalsbcr. Berlin Akad. W'iss., 1873: 738-744. . 1S76. Uber die von S.M.S. Gazelle mitgebrachten Amphibien. Monatsber. Berlin Akad. Wiss., 1876: 528-535. Rooij, X. DE. 1915. The Reptiles of the Indo-.Austialian .\rchipelago. I. Lacertilia, Chelonia, Eniydosauria. Leiden: E. J. Brill Ltd. xiv + 384 pp. Smyth, M. 1968. The distriI)utioii and life history of the skiiik. Uemiergis peronii (Fitzinger) . Trans. Roy. Soc. South .Australia, 92: 51-58. Taylor, E. H. 1963. The lizards of Thailand. Univ. Kansas Sci. Bull., 44(14): 687-1077. TwEEDiE, M. W. F. 1940. Notes on Malayan reptiles. Bull. Raffles Mus., No. 16: 83-87. Van Dei sen, H. M. 1966. The seventh Archbold Expedition. BioScience, 16 (7) : 449-455. Vinciglerr.\, D. 1892. Rettili e batraci di Engano raccolti dal Dott. Elio Modigliani. Ann. Mus. Civ. Stor. Xatur. Genova, Ser. 2, 12(32) : 517-526. Vis, C. W. de. 1888. A contribution to the herpetology of Queensland. Proc. Linn. Soc. New South Wales. Ser. 2, 2(4): 811-826. VocT. T. 1911. Rcptilicn und amphibien aus New-Guinea. Sitzungsber. Geselisch. NaturforsMi. Ircundc 1911. No. 9; 410-420. ^'^^-'^. COMP. ZOOL B R E V I «>'* A MARVARO useuni of Comparative^^iMlogy us ISSN 0006-9698 Cambridge, Mass. September 20, 1973 Number 407 THE GHANARES (ARGENTINA) TRIASSIG REPTILE FAUNA. XIX. POSTGRANIAL MATERIALS OF THE GYNODONTS PROBELESODON AND PROBAINOGNA THUS Alfred Sherwood Romer and Arnold D. Lewis Abstract. Descriptions are given of postcranial materials of Probelesodon and Proba'nwgnallnis and a restoration of Probelesodon lewisi is attempceu. Because of the phylogenetic position of the cynodont therapsids as the probable ancestors of mammals, their structure is of great importance in the story of vertebrate evolution. A number of excellent studies have been made of cranial structures of c)no- donts. As regards the postcranial skeleton, there have been numerous descripti\e papers, but few which ha\'e attempted a broad study of cynodont skeletal materials from an evolutionary or functional viewpoint. Early essays of this sort were those of Watson (1917), Gregory and Camp (1918), and Romer (1922) ; a recent comprehensive work is that of Jenkins (1971; cf. also Jenkins, 1970). Postcranial remains of African cynodonts ha\'c been compre- hensively studied and summarized by Jenkins (1971); some data on Permocynodon of Russia have been given by Konjukova (1946). Of the South American cynodonts, descriptions of gomphodont skeletons ha\e been given by Bonaparte ( 1 963 ) for Exaerctodon, and by Jenkins (1970) for Alassetognathus, and Huene ( 1 944 ) has described a limited amount of post- cranial material of Traversodon. For the carnixorous c\nodonts of South America, all so far published has been the description of a partial skeleon of Belesodon by Huene (1944), and of a limited amount of material of Chiniquodon by Romer (1969). The purpose of the present paper is to place on record such data as are available on the postcranial skeleton of the car- 2 BREVIORA No. 407 nivorous cynodonts of the Chafiares Formation — Probaino- gnathus and Probelesodon. Together with their relatives, Chiniquodon and Belesodon, from the Santa Maria Formation of Brazil, they include the latest in time and most advanced of therapsids leading in a mammalian direction. As Jenkins has noted (1971), the postcranial skeleton of cynodonts shows in general a remarkable consistency of pattern, and hence little novelty is to be expected from the description of the Chafiares genera. Collection and preparation of Probelesodon and Pro- bainognathus were made possible by grants from the National Science Foundation. Postcranial Material of Probelesodon lewisi A major source of information is MCZ 3781 (field no. 79), the remains of a nodule that contained a nearly complete and mostly articulated specimen of this form, with a skull approxi- mately the size of the type. This is preserved as a slab, prepared on both surfaces. Photographs of the two sides (which we will call the obverse and reverse sides) are shown in Figures 1 and 2. Outhnes of the materials seen on the two surfaces are shown in Figures 3 and 4. Figures 3 and 4 were derived from the photographs and, owing to perspective, the two are not com- pletely superposable. Unfortunately, some parts of the speci- men were lost before collection; further, the specimen had undergone considerable weathering and penetration by grass roots, with the result of obscuring much detail and rendering interpretation difficult. MCZ 4002 (field no. 98), in addition to a good skull, included a fair amount of mostly disarticulated postcranial material; this material is, in general, in better shape than that of MCZ 3781. Several other concretions include postcranial materials that may be of Probelesodon, but for the most part such materials either add little to the data available in MCZ 3781 or MCZ 4002 or are of doubtful assignment. We may note, however, that MCZ 3801 includes a melange of bones, certain of which rather surely pertain to Probelesodon, notably a pair of excellent femora. Axial skeleton. At burial the specimen constituting MCZ 3781 appears to have had a complete vertebral column, articu- lated for the most part. However, before collection, a con- siderable portion of the dorsal vertebrae had been lost, leaving 1973 CYNODONT POSTCRANIAL MATERIALS Figure 1. Obverse side of the slab, MCZ 3781, containing a skeleton of Probelesodon leivisii. Slightly less than l^ natural size. BREVIORA No. 407 Figure 2. Reverse of the slab shown in Fig. 1. 1973 CYNODOXT POSTCRAXIAL MATERIALS Figure 3. Diagram of the structures on tlie obverse side of the slab shown in Figs. 1 and 2. Slightly less than I3 natural size. Abbreviations for Figs. 3 and 4: /, left; r, right; d, clavicle; /. femur; fib, fibula; /;, hu- merus; ic, iiiterclavicle; t, ilium; is, ischium; r, radius: sc, scapula; SI, first sacral; S5, fifth sacral; t, tibia; u, ulna. BREVIORA No. 407 0 a ^nxU^SS Figure 4. Diagram of the reverse side of the same slab. Slightly less than y^ natural size. Abbreviations as in Fig. 3. 1973 CYNODONT POSTCRANIAL MATERIALS 7 a small number of cendcal vertebrae present anteriorly, then, follo\sing a gap, posterior dorsals, "lumbars," sacrals and part of the tail. Although the articulated condition of the posterior part of the column and the forward continuation of a rib series in seemingh natural position indicates that much of the colunm was articulated at the time of burial, it is ob\ious that the column had been broken at about the posterior end of the cervical series. Se\'eral isolated \ertebrae, ob\iousl\- from the region of the break, are present abo\e and behind the skull. It is impossible to be certain of the exact presacral \ertebral count. As noted belo^v•, there is some question as to the position of the first sacral, but the articulated series anterior to this appears to include 14 posterior dorsals and "lumbars." For the most part the ribs associated with these vertebrae are preser\ed, and anterior to the \ertebrae present, this rib series is con- tinued for eight segments further to the 22nd segment anterior to the sacrum. The most anterior rib is not complete at its head, but has a moderately long shaft, and hence can be no farther forvvard than the posterior end of the cer\ical series. If 27 presacrals — 3. typical cynodont number — were present, this rib would pertain to \ertebra 6. Since presumabh the anterior cer\ical ribs were short, no more anterior position for this rib seems probable and we ha\e hence assumed for purposes of restoration the typical count of 27 presacral \ertebrae. In general the presers-ation of the vertebrae present is not good, but as far as can be seen the structure is of the general cynodont type. Remains of three \ertebrae are present close to the back end of the skull. The most anterior vertebra pre- ser\'ed has a neural spine (broken abo\'e) of considerable length anteroposteriorly, suggesting that we are dealing with the axis. Little more can be made out as to details of structure on the three further cervicals present in MCZ 3781 except for the presence of normal zygapoph) sial regions, well de\eloped trans- verse processes slanting outward, backward and downward, and in two of the three, neural spines that are relatively narrow anterposteriorly (the other — apparently the fourth in this series — -appears to have a widened spine). Isolated vertebrae that appear to be cervicals are present in MCZ 4002. As preserved, the centra are subcircular in end view, with a diameter of 11-12 mm, and a central length of perhaps 10 mm. Transverse processes are well de\eloped and extend strongly out from the arch bases. Posterior cer\icals and much of the dorsal region is missing 8 BREVIORA No. 407 in MCZ 3781, although represented by a few scattered vertebrae. A number of dorsal vertebrae are present in MCZ 4002. Those best preser\ed have central lengths of about 12 mm, and height of centrum of 12-13 mm. They are deeply biconcave; the cen- trum, as seen in end \iew, is oval in outline, with a width some- what less than the height. Both anterior and posterior margins of the centra are somewhat thickened ; presumably the capitulum was carried on the anterior rim, but there is little indication of a discrete articular facet. The transverse processes are stout but short, extending but little outward beyond the level of the arch base, and face strongly downward. The posterior zyga- pophyses are almost directly above the base of the transverse processes; anteriorly the arch bases extend far forward, so that the spine extends upward about opposite the front margin of the centrum. Of the series of 14 presacrals preserved in MCZ 3781, and mainly \'isible on the reverse surface of the slab, the second to fourth are seen in side view and are fairly well preser\'ed. The neural spines are relati\'ely short, rising to about 15 mm above the level of the zygapophyses, and broad anteroposteriorly. Centrum lengths are 12-13 mm. Beyond this point the \-erte- brae are seen from above, and the neural arches and spines are for the most part absent, revealing the neural canal on the surface of the slab. The series is continuous, except for the last two presacrals, which are disarticulated and damaged. Beyond this point the sacrals and anterior caudals are articu- lated. Here the vertebrae are seen from abo\e and somewhat to the left side. Eleven caudals and part of a twelfth are present. The neural arches, still well developed on the posterior sacrals, decrease in height and become more slender as we continue along the caudal series; the transverse processes, as preser\^ed and seen on the left, likewise decrease in length posteriorly. Length of centra, about 12 mm in the sacrals. decreases to about 8 mm on the last caudal preser\'ed. On the ob\'erse side are seen the rounded \entral surfaces of some of the vertebrae in the series. Beneath the articulated caudal series are obscure remains of additional caudal \ertebrae. On MCZ 4002 is found a series of seven articulated caudals, five of which are well preserved. The centra have lengths of 8 mm each; they are apparentlv somewhat compressed vertically. The height of the centra is al^out 5 mm; the neural spines are very short, and as preserved there is little trace of any transverse process. It seems certain that we are dealing with elements distal to the 1973 CYNODOXT POSTCRANIAL MATERIALS 26 25 24 23 22 21 20 19 18 17 16 15 14 13 12 II 10 9 8 7 6 4 Figure 5. Lengths of right ribs of Probclcsodon lewisi, MCZ 3781, from presumed \ertebra 6 to 26. Measurements are direct between the two ends of the parts preserved; rib lieads are complete only on ribs 14 and 15. X 2/3 natural size. tail "stump" seen on the slab, which strongly suggest the presence of a fairly long tail. No interpretable remains of anterior cersical ribs are pre- served. Of ribs of the left side, a few dorsals are present in isolated fashion on the reverse surface of the slab. Posteriorly, a few stumps of left rib are present in the posterior dorsal- lumbar series; these have a broad head but rapidly decrease to the base of a slender shaft. There is no indication of the interlocking expansions seen in various other cynodonts. The ribs of the right side are better preserved. Onh- on two ribs, which we interpret as presacrals 14 and 15, are the heads preserved; these are expanded proximalh' for capitular and tul^ercular attachments to the \ertebrae, but taper rapidly dis- tally, with no trace of proximal overlapping expansions. A proximal series of right ribs, commencing with that which we assume pertained to vertebra 6, are seen on both obverse and reverse surfaces posteriorly. The ribs attributable to \crtebrae in the posterior presacral series are seen on the reverse .surface, their heads concealed beneath the vertebrae. In Figure 5 we have plotted the lengths of ribs as preserved (measured directly between the two ends ) . Although we cannot be sure how much 10 BREVIORA No. 407 has been lost proximally or distally in most cases, we see a rea- sonable distribution in lengths, with an increase to a maximum in the anterior dorsal region followed by a gradual decrease as we enter the lumbar region and approach the sacrum. On MCZ 4002 a number of dorsal ribs are preser\'ed in their entirety, including the heads; they are similar in nature to numbers 14 and 15 on the slab and appear to be mid-dorsals of both sides with average lengths of about 10 cm. Although the usually imperfect nature of trans\'erse processes makes diffi- cult the interpretation of the orientation of ribs to the back- bone, the rib structure strongly suggests a broad, barrel-shaped abdomen. The lumbar ribs as preserved are slender (except for the presumed first presacral) and although the heads are concealed beneath the centra, there obviously was none of the interlocking found in the lumbar series of certain other cynodonts. In the sacral region some seven ribs on the left side show distal expansion (the first broken off from its \'ertebra). As the length of the associated right ilium indicates, fixe members of this series are sacrals; but which five? Possibly incorrectly, we assume that the first (whose vertebra is disarticulated from the sacral-postsacral series) is the last presacral (and we have used this assumption in making a count of presacral vertebrae). The first four, particularly, of the assumed sacral series are expanded at their distal surfaces for iliac articulation ; two sacrals are visible in ventral view, which better shows this prominent distal expansion. A slender rod of bone, extending fonvard, is attached without evidence of sutural separation to the outer end of the presumed fourth left sacral; this appearance, however, is probably due to the peculiar type of preser\'ation in this speci- men, and this rod is perhaps a fragment of the left ilium. Limbs a7id girdles. Remains of the shoulder girdle are present in MCZ 3781. The interclavicle is present, and on the obverse are the lower ends of the two clavicles. The interclavicle as preserv'ed (slightly incomplete anteriorly) measures 57 mm in length. A better preserved interclavicle (slightly incomplete dis- tally) is present in MCZ 4002 (Fig. 6b). This is of similar size, with a length of 59 mm, a width of 33 mm across the head, and, following a constriction, to a width of 1 1 mm back of the head, a modest increase in width distally. The structure is that typical of cynodonts in general. The anterior "quad- rants" for clavicular reception are deeply incised; a median keel is present between these quadrants but there is little de- 1973 CYNODOXT POSTCRANIAL MATERIALS 11 Figure 6. Probelesodon lewisi, a, left clavicle in lateral view; b, inter- clavicle in ventral view; c, ulna; d, radius of right side seen from extensor surface. All from MCZ 4002. X 1. \-elopment of a median ridge more posteriorly. The left cla\icle is well preser\"ed in MCZ 4002 (Fig. 6a). It is of typical cynodont build. Its length, measured directly between the ends, is 51 mm. The relati\ely thin shaft curves about 45° between proximal and distal ends. The dorsal end is somewhat expanded and excavated medially for acromial apposition. A well-defined ridge descends the posterior margin of the shaft and continues on to the posterior margin of the \-entral expansion for articu- lation with the interclaxicle. Of the primary girdle in MCZ 3781, an obscure fragment of the left scapula is present; the girdle of the right side is present, but is poorly preserved and incompletely seen (Fig. 7). On the re\erse side is visible the glenoid region and incomplete coracoids, in poor condition: on the ob\'erse is the medial sur- face of the scapular blade. This is, as in other chiniquodonts, imusually tall and narrow, with a length (probably incomplete) from the glenoid articulation with the coracoid of 55 mm, and a width at mid-height of 1 1 mi-i. The exposed medial surface of the blade is strongly convex in the transverse plane and obviously the external surface was strongly ridged longitudinally on both front and back margins. In MCZ 4002 a scapular blade was present, but is represented bv little but an imoression 12 BREVIORA No. 407 Figure 7. Probelesodon lewisi, right scapula and coracoids; the anterior coracoid poorly jsreserved, the acromial region of the scapula incomplete. From MCZ 3781. X 1. indicating that it was elongate and slender. Both coracoids are present, the posterior element complete and well preserved; the anterior coracoid is incomplete. In MCZ 3801 there is a scapular blade of Probelesodon type, narrow and deeply con- cave, externally with strong ridges posteriorly and (especially) anteriorly. As preserved, the blade is 48 mm tall, 12 mm wide at the summit, with the width decreasing to 6 mm below. Unfortunately, the scapula is incomplete ventrally, but there was obviously little acromial dexelopment and no indication of any beginning of a supraspinous surface. Both humeri are present in MCZ 3781, and are essentially complete although somewhat crushed, and a good left humerus is present in MCZ 4002 (Fig. 8). The bone in general corre- sponds well with that of typical cynodonts, as described in detail by Jenkins (1971). Lengths of the humeri of MCZ 3781 as preserved are 72 and 66 mm, that of MCZ 4002 68 mm. The 1973 CYNODONT POSTCRANIAL MATERIALS 13 Figiac 8. Probclesodon lexi'isi. Right humerus iu distal ventral, lateral, and distal dorsal views. From MCZ 4002. X 1. head of the bone is unusually expanded anterodorsally, so that the articular area extends into a "lip"' overhanging the antero- proxinial corner of the dorsal surface. The deltopectoral crest is somewhat more expanded than in typical cynodonts. The distal end of the bone is unusually broad; its width in MCZ 4002 is 43 mm, thus being about 63 percent of the length, in contrast with lower figures in most cvnodonts. This increased width appears to be associated with a greater expansion than usual of the entepicondyle. On the ventral distal surface the areas for radial and ulnar articulations are well ossified in MCZ 4002. In MCZ 3781 the right radius and ulna and left ulna and partial radius are present, but poorly preserved. In MCZ 4002 the right radius and ulna are well preserved (Fig. 6, c, d). In this specimen the radius has a length of 53 mm, the ulna 56 mm. Both bones conform well to typical cynodont structure. The radius is essentiallv columnar in shape, but as preserved some- what flattened between extensor and flexor surfaces. Proximally the l-)one is expanded and deeply cupped for humeral articula- tion. Distally the width of the bone is increased toward the medial margin and thickened on the flexor surface to make 14 BREVIORA No. 407 Figure 9. Probelesodon leivisi, right ilium and ischium, composite, MCZ 4002 and MCZ 3781. X 1. possible a broadly oval distal surface for carpal articulation. As in Thrinaxodon, a ridge, perhaps for biceps attachment, is present proximally on the flexor surface; a flattened area at its summit was presumably for apposition with the ulna. A ridge with a rounded summit descends the distal third of the medial margin, presumably marking the boundary between extensor and flexor muscle areas. The ulna has the typical cynodont pattern. As usual, the olecranon is unossified except for its base, the sigmoid notch, represented only by its basal portion. At the expanded dorsal end of the bone the medial area is somewhat concave for radial apposition; lateral to this area a pronounced ridge is present proximally, fading out rapidly below. The proximal part of the flexor surface is markedly concave, bounded externally l^y a distinct ridge. Below, the shaft of the bone narrows, to be only moderately expanded distally for carpal articulation. Deep to the trans\'erse plane of the shaft, however, a very strong ulnar crest is present, as in other cynodonts; this is conspicuously de\'eloped along the distal half of the bone. Remains of the manus of both limbs are present in MCZ 3781, but in poor condition. Little certain data can be made out concerning the carpus, except for the presence of three promi- 1973 CYNODONT POSTCRANIAL MATERIALS 15 nent distal carpals. All metacarpals are present on the left side, four on the right. Exact measurements are meaningless, owing to the imperfect nature of the material, but the a\erage length of metacarpals II-V is about 18 mm. A limited number of poorly preserved phalanges are present but in no case is there a complete digit. The digits are, however, sufficiently preser\'ed proximally to show that the vestigial "extra" phalanges present in primiti\e cynodonts were absent. In MCZ 3781 the incomplete upper margin of the left ilium is seen on the reverse side of the slab; the vertically oriented complete right ilium is seen on the margin of the slab, and a nearly complete left ilium is present in MCZ 4002 (Fig. 9). The iliac length is about 70 mm. For the most part the struc- ture is typical of that of cynodonts generally, but the posterior part of the blade is much more slender than in most cynodonts, suggesting that Prohelesodon is ad\'ancing toward the mamma- lian condition of forward migration of the gluteal musculature. The right ischium is present in MCZ 3781, and has a normal cvnodont form. Except for the area of iliac articulation of the right pubis, there are no identifiable remains of the pubis in available m.aterial. Hind limb material is poorly preserved in MCZ 3781. Only the proximal part of the left femur is present and the right femur (with a probable length of about 80 mm) is badly crushed and liT-oken. 0'"lv fragments are present of the left tibia and fibula. The rioht tibia and fibula are present but in such poor condition that little can be said exceot that they appear to conform generally to the usual cvnodont pattern. .A.s preserved their lengths are 74 mm and 63 mm (the fibula is incomplete proximally). In the melange of material in field number 188 (MCZ 3801 ) are t^vo well-preserved cynodont femora which seem quite surelv to belong to Probelesodon leivisi (Fig. 10). With a length of 63 mm each, they represent an animal smaller than MCZ 3781, but they are too large to pertain to Probelesodon minor or Probainognnthiis, and they differ in structure from the gomphodonts in the fauna. In nearlv all respects these femora are closelv comparable to the "PCynognathus" femur illustrated by Jenkins (1971, fig. 48) — even to the presence, part-way down the shaft, of a groove of vmknown nature. These femora, however, difTer markedly from all hitherto described cvnodont femora in the position of the greater trochanter. In all cvnodonts hitherto described the orcater trochanter lies some 16 BREVIORA No. 407 Figure 10. Probelesodon lewisi, left femur in dorsal, ventral and medial views. From MCZ 3801. X 1. distance down the posterior margin of the femur and faces as much laterally as proximally. In this specimen it is placed definitely farther toward the proximal end of the bone and its somewhat expanded tip faces nearly directly proximally. We have here a position splitting the difference between typical cynodonts and the mammalian condition. A femur of the chini- quodontid Chiniquodon described by Romer (1969, fig. 9C) shows a somewhat comparable proximal movement of the greater trochanter, but the specimen is not too well preserved. The feet are poorly preserxed. Of the left foot, a few dis- articulated metatarsals, phalanges, and tarsal remains are seen on the reverse side. The right foot is nearly complete but difficult to interpret. Astragalus and calcaneum can be made out on the obverse side. Two complete and two incomplete metatarsals are present; the two complete — perhaps the third and fourth — ^ are 17 mm in length as preserved. Although too much reliance should not be placed on measurements of such poor material, it would appear that here, in contrast to certain other cynodonts, front and hind feet were of approximately equal size. A number of phalanges are present, but except for 1973 CYNODONT POSTCRANIAL MATERIALS 17 u a, 18 BREVIORA No. 407 the certain presence of two phalanges on supposed digit IV (and the obvious absence of "vestigial" phalanges), we cannot be sure of their arrangement. Restoration. On Figure 11 we have attempted a restoration of the skeleton of Probelesodon. We believe that this restoration gives correctly the general appearance of the animal, although, as noted in the description above, knowledge is lacking in a number of regards, such as cervical ribs, the distal portion of the tail, the pubis, and the distal phalanges; even the probable presacral count of 27 \ertebrae is uncertain. The body was surely stockily built, and although we have included in the tail only those \'ertebrae in MCZ 3781, the evidence indicates that Probelesodon (and not unlikely other cynodonts as well) had a tail of rather good length. A notable departure from the situa- tion in many cynodonts is the almost complete absence of the overlapping of the ribs, seen at its height in Thrinaxodon. This type of structure has been discussed by Jenkins (1971: 76ff ) ; unless we assume that such a late cynodont as Probele- sodon has evolved independently from the very base of the cynodont stock, the situation, as Jenkins believes, is one in which there has been a return to "normal" rib structure. Presumably such cynodonts actively pursued their prey, but their limbs would seem to ha\e been relatively inefficient for speedy locomotion as compared with later mammals or con- temporary archosaurs (although an improvement over the pelycosaur condition). As in early tetrapods generally, the hind legs were longer than the fore, the combined length of femur plus tibia being about 30 percent greater than that of humerus plus radius. Again, as in primitive forms generally, the podials are longer than the epipodials; the humerus is about 30 percent longer than the radius, the femur about 15 percent longer than the tibia. Probelesodon may be compared with a cariid, Vulpes, of siinilar size, and with presumably somewhat comparable habits. We may use as a crude iDase for comparison the length of dorsal v'ertebrae (approximately the same length in Vulpes and Pro- belesodon). To give a rough estimate of comparative limb length we may use the length of humerus and radius for the front leg, of femur and tibia for the hind leg. In Vulpes this front leg measurement is 1 9 times the length of a dorsal vertebra, in Probelesodon, little longer than the len CO ^ CO a 1 1 1 1 CO 1 1 1 1 1 en csr o 1 T, 0; E « 5 Tf in t^ Ol CI Cf ■* t^ ift to "* - S en •*! CM eo GO CO CO 01 c^x CM ll e< ^ ^ o - rt ^ V3 3 !n XI u 1) _bC '> O o 2 - OJ o c a; U t^ to CM * ct! * B s Ctj pq ^ «i ^ ij o Nl ^ C3 C u Si pq OJ 3 O c/2 pq Si" (5 6 5^ U OJ P3 5 QJ rt M w H-^ d" 3 o S tsj 1— rt n JJ 3 rt Si s r3 0 ft, O "rt I ■J [2 % r3 (5 1973 AMERICAN UPOGEBIA 5 Table 2. Characters used in the analysis of Upogebia affinis and U. omissa, and indicated by niniiber in Fig. 1. Each character is referred to in the text and tables by the words in italics. Number Character 1 Number of ocular spines 2 Rostral ventral spines 3 Number of spines on epistome 4 Number of spines behind the cervical groove 5 Ventral abdominal spines 6 Serration of uropod distal edges 7 Number of uropodal spines 8 Style of dactylar teeth (PI) 9 Style of teeth on fixed finger (PI) 10 Carpal exterior lateral spines (PI) 11 Number of dorsal palni ridges (PI) 12 Proximal meral spine (P2) 13 Width of the rostral base 14 Rostral length 15 Length of eye stalk 16 Length of rostral lateral teeth 17 Length of sixth abdominal segment 18 Length of telson 19 Width of telson distal margin 20 Width of telson proximal margin 21 Length of fixed finger (PI) 22 Length of dactylus (PI) 23 Length of palm (PI) 24 Width of palm (PI) 25 Length of mcrus (PI) 26 Width of merus (PI) BREVIORA No. 408 /\ -N n vV-llS Figure 1. Diagrammatic presentation of characters analyzed, numbered as in Table 2. Underlined numbers are scores for the dentition patterns of fixed finger and dactylus. testing the position of the lines at this grand median rather than differences in slope (E. W. Fager, personal communication). When tested for sexual bias in terms of numbers of indi\iduals, none of the collections departed significantly from the null hy- pothesis of a binomial distribution (p = q=/4) at the 90 percent level. Assuming that the sex ratio is 1 : 1 in both species, these lots are not significantly biased in terms of sex. In order to minimize the effect of ontogenetic changes in morphology on interspecific comparisons, this study used the overall length of the smallest ovigerous female in each sample as a criterion for restricting the analysis to adults. This pro- cedure assumed that overall length, age, and maturity were highly correlated so that the probability of females longer than 1973 AMERICAN UPOOEBIA 7 this niininuini bciii"- adult was lan'e. The overall lene^ths of males versus females in each sample were such that a Mann- Whitney "U" test (Tate and Clelland, 1957: 89-91) revealed no significant difference at the 90 percent le\el. Because the samples were not biased in terms of sex (see above) nor were the sexes different in oxerall size, it seemed reasonable to extend the adult overall length minimum to males. Thus, an adult upogebiid was defined as an individual that was larger, and by inference older, than the smallest sexually mature female present in the sample. In two cases this criterion was set aside for prac- tical reasons. The available specimens of Upogebia omissa con- tained only two o\"igerous females (36, 44 mm), while the lengths of all specimens ranged from 1 9 to 44 mm ( Table 1 ) . Correa (1968) reports adults ranging from 27 to 47 mm. Those indix'iduals smaller than Correa's minimum were considered ju\-eniles, as Corre'a's range of adult o\'erall lengths was based on 106 o\'igerous females. The Wellfleet, Massachusetts, collec- tion contained no ovigerous females. The closest population of U . a/finis in overall length is that from Miami and its minimum (28 mm) was used. After the removal of subadults in this man- ner, subsamples for analysis were taken at random from samples of more than ten indi\iduals. Results Measured characters. To provide a quantitati\'e estimate of the variability within a species, Upogebia affinis and U. oyyiissa were analyzed by the regression of 14 measured characters on overall length. Each character was tested for sexual dimorphism by the comparison of 90 percent confidence limits erected about regression lines formed for each sex. For U . omissa none of the 14 characters differed significantly between sexes. In U. affinis fixed finger length (21) and palm width ( 24 ) were significantly sexually dimorphic (Figs. 2, 3); the remaining characters were not. Interspecific comparisons using the regression lines for each sex separately re\ealed no significant difference for either sex on any character. Regression fines formed from both sexes still showed no significant diflference between species on any char- acter, excluding characters 21 and 24. The characters measured contain infonnation about the shape of much of the animal. The results show the two species to be largely indistinguishable in gross morphology, making speculation about the origin and niche separation of these two partially 8 BREVIORA No. 408 3A y IJ 30 40 50 GO Overall length [min] Figure 2. Regression of the fixed finger length onto overall length for males (squares) versus females (circles) of Upogebia affinis. The upper triangle is the x-axis median for males; through it passes the best-fit median, regression line. The envelope of lighter lines are 90% confidence limits. The lower triangle marks the female x-axis median point with a similar set of lines. sympatric species interesting, but to little purpose until their natural history is better known. The analysis does point to prob- lems latent in the use of measured characters in this genus. One must quantify the variability and examine it comparatively be- fore any but the most ob\'ious differences in proportion are given taxonomic weight. The regression analysis confirmed one feature of taxonomic in- terest. Upobegia affinis has conspicuous sexually dimorphic chelipeds. In the males the cheliped is consistently more robust, larger, and better calcified than in the female. In U. omissa, while the males tended to be more variable about the regression line reflecting the occasional dimorphic indixidual as reported by Correa (1968), there was no significant difference between 1973 AMERICAN UPOGEBIA y 4 iji 41 SB Overall Lenath rmml Figure 3. Regression of the width of pahii (PI) onto overall length for male (squares) versus female (circles) Upogebia affinis. The upper triangle is the x-axis median for males; through it passes the best-fit median, regression line. The envelope of lighter lines are 90% confidence limits. The lower triangle marks the female x-axis median point witli a similar set of lines. 10 BREVIORA No. 408 sexes. Upogebia affinis is sexualh' dimorphic in palm width and fixed finger length; U. ornissa is not. This dichotomy helps to distinguish the species. Meristic characters. Despite this similarity of shape, there are differences between Upogebia affinis and U. omissa. The diag- nostic characters of these two species are differences in orna- mentation, as are those which distinguish the other species. Table 3 summarizes the results of an examination of 1 2 of these characters. In it one can see the type of individual and geo- graphic variability present in U. affinis and U. omissa, the fea- tures which separate them, and some of those they share. In delineating Upogebia affinis from U . omissa, clear disjunc- tions are most useful. Upogebia omissa has ventral abdominal spines and P4 is armed; U. affinis does not ha\e these spines. The other characters that show differences between species are less distinct. For a given character, each species has a different dominant state, though some individuals of each species exhibit the character state of the other species. The greater the fre- quency of the inappropriate character state, the less useful the character, but because of the obvious effect of the interaction between those populations contributing the most specimens to the relative frequency of a character state within a species, these frequencies were not tested statistically. De Man (1927) redescribed Upogebia affinis from a few Carolina specimens. His detailed description agrees with my material. He did not comment on variability beyond two lo- calities. Table 3 can be considered to supplement his description. In addition, the rostrum is not always longer than wide; the telson is rectangular to wider posteriorly; and the upper surface of the telson is not always punctate. De Man refers to reports of U. affinis from the "coast of Brazil, Mamanguape stone reef, Parahyba river . . ." On reexamination these specimens were found to be U. omissa (Table 1 ) . Correa (1968) described Upogebia omissa in detail, including its variability. Beyond those features already discussed, I found the following differences. The eyes are slightly shorter than the rostrum. I have examined a female that is 44 mm long versus a maximum of 35 mm given by Correa, Also, in the table pro- vided by Correa for comparison of Upogebia affinis with U. omissa, the distinction based on spines on the lower surface of the rostrum is not useful, as U. affinis from Venezuela lacks the spines. The protopods of the uropods bear two spines in U. affinis from Venezuela rather than one. 1973 AMERICAN UPOGEBIA 11 ^ T. 2 .'' " •- ^ ^ uTIsS-^'^': >r= - I ~^+ ^ + ^ ^ ^ - + < - - ~. -^ ^ '-S ^ ^ "-^ ^ ■ — r" f^ Q P '~ )^ O ^ |;., 3^(^^^ln■«t- ^ 2 o y «-- i; ci 15, CO* •I I i i I I a ^ ^ - ^1 + ^- 1 or 0. iM '^^ + 5 3 n ^ o" 3 - -I- " '^ '^'* ^ y n ^ si ii S « - 5^ — fcT £- = -£02 = r3 cr SP 3 5 or o ^ o « 5 < g *^ S. -H- ' -h' ' ^ Of *^ C-; + "-■ « S i- o t) « ~ "T" S w .. rs ■"■ ?? >H s •- i: „ w E ~ .2 _- ti ~ = ■>-■- b ^ o 2 y= ■=: " C « J u - ,^ ii -J - o ii^ '^ a ^ - '^ ~L^ '^- r'. -^ u: ^ ?\ 12 BREVIORA No. 408 The Species of Upogebia On the basis of the analysis of Upogebia affinis and U. omissa, I selected characters that had little within-species variability while setting off at least one described species from the others. These characters, coupled with several useful characters from the literature, are presented as a diagnostic matrix showing the interrelationships among the species (Table 4). Characters based on measurements were not included, because the lack of specimens of other American species made it impossible to prop- erly evaluate their variability, except that the ratio of PI fixed finger length to dactylus length, which is routinely given in the literature, was included. Table 4 reveals a portion of the interrelationship among the members of the genus in the Americas, and can be used to distinguish among the species. The information presented is from several sources: holotypes of Upogebia annae n. sp. and U. jatnaicensis n. sp., paratypes of U. omissa and U. rostro- spinosa, museum collections of U. affinis. The original descrip- tions of U. noronhensis Fausto-Filho 1969 and U. brasiliensis Holthuis 1956 were used. For U. spinigera (Smith 1871) the original description was supplemented by Holthuis' (1952) re- description. Similarly, for U. pugettensis (Dana 1852) de Man (1929) and Stevens (1928) were used. Upogebia longipollex was described ver\' incompletely and without figures by T. H. Streets (1871) from a Panamanian collection of J. McNeil. Lockington (1878) states that the material ". . . probably came from the Pacific coast of the isthmus." De Man (1928) speculated that U. longipollex might be a junior synonym to U. spinigera (Smith) if differences in spination of the pereopods were the result of differences of the ages of the specimens described. Holthuis (1952) synonymized U. longipollex with U. spinigera without comment. The results of this study indicate that leg spination, particu- larly the P2 meral spine (\'entral proximal spine of merus of pereopod 2), is diagnostic at the specific level. Streets, describ- ing spination, states, ". . . third article [carpus] . . . armed with spine above at distal extremity; remaining pairs [of legs] unarmed." U . spinigera has a P2 meral spine as well as spines on P3 and P4. It cannot be the same species as U. longipollex. Upobegia longipollex has been included in Table 4 as a good species and adjacent to U. pugettensis to which it seems to be most similar. It appears likely that after an adequate variational 1973 AMERICAN UPOGEBIA 13 c ^ ^ Ji -Ho . >— ) 3 c « t; <^ •a a 2 <« •^ S O t) < O o ~ o - >, ■-:: "^ I .--s CM ^ -—-«.. I' C4 §-.!=< s^^ ^ 5 "? -^ " I -. ^^. §<^ + + I I ^- - ♦^ -^ Lj .Li _I t"^ I ^ ^^ ^^ , 3 ^^ _t 1^ ^ ^ y < "06 „ .. . ^- r ::i^-. :. ^ 7^ \^ >r "^ ^ ^: 2:i «J ^ U «; S: ~ Si <; . E rj t^^ ™ ^ w V— ;i r* <- 't:i ^ = 5 ^ S = &H "^ ic i_ -*^ • P rt C jj o o a. be o til n .1 5 I 5 '^ "O O O i« c » I „ .^. ,^ ^ ^: .n _H 2 =^ I I I I - + <^ 10 CM rv. O I I I I r^ <%. « '^ • « 9" - -,». *^ * B R A J V I '^'^■'S. CCMP. ZOOL^ useiinti of Comparative Zoology us ISSN 0006-9098 Cambridge, Mass. September 20, 1973 Number 409 THE CLASSIFICATION OF THE COTINGIDAE (AVES) D. W. Snow Abstract. The treatment to be given to the Cotingidae in Peters' Check- list of Birds of the World is discussed. Both the limits of the family and the best ariangement of the genera within it are problematical, and it is stressed that the system to be adopted must be regarded as provisional. Previous attempts to classify the iamily are reviewed. Nine genera are discussed, whose allocation to the Cotingidae has been disputed. Four of these {Laniisoina, Plwcnicircus, LilMugiis and Riifncola) are considered best retained in tlie family on available evidence, and five (AttHa, Casiomis, Lanioccra, Rhytifitcrna and Xoiojjsayis) are considered to belong to the Tyrannidae. Doliornis and Zaratornis are not considered to be sufficiently tlistinct from A>ul)clion to be maintained as scpaiatc genera. Likewise, Platypsaris is merged with Pach\raviphtis, and Eratnr with Tityra. The sequence of genera adopted indicates a general progression from unspccialized types to highly specialized, sexually very dimorphic frugivores; but it is emphasized that this may not represent an evolutionary progression. Within the genera, there is a marked tendency for closely related forms to replace one another geographically. Many of these are too distinct to be treated as races of one species; moreover, in two cases where there is slight geographic o\erlap the forms concerned appareirtly behave as distinct species although they are very similar. To take accoimt of this general situation, liberal use is made of Mayr's concept of the zoogeographical species. On this basis the family is treated as containing 79 taxonomic species and 56 zoogeographical species. Before his death in 1955, J. T. Zimmer prepared a rough manuscript of the Pipridae, Cotingidae, Tyrannidae, and related families, to constitute \()lume 8 of Pcter.'i' Check-list of Birds of the IVorld. The section on the Cotingidae was reviewed in 1957 by James Bond, who made minor modifications, and it was this revised manuscript that was the basis of the classification adopted by Me)er de Schauensee (1966). Volume 8 of the Check-list, however, unfortunately remained unpublished. 2 BREVIORA No. 409 Having been invited by Melvin A. Traylor, who has ac- cepted responsibility for the final editing of this volume, to revise Zimmer's manuscript of the Cotingidae in the light of recent work, I here discuss the controversial points in the classi- fication of this difficult family at the species level, in particular the affinities of the genera incertae seclis, the limits of some of the species, and the sequence to be adopted in the Check-list. Zimmer, as is well known, was a meticulous worker with sub- species, and his treatment was reviewed by Bond, as already mentioned. It has therefore seemed unnecessary to alter his arrangement of subspecies except where modification is called for in the light of more recent published reviews. Hence in what follows no mention is made of subspecies except where they arc in\ol\ed in questions of specific status. The Cotingidae is a notoriously heterogeneous family, and successive systematists have despaired of producing a satisfactory classification of it. About half of the genera are monotypic, and several others are represented by a single superspecies. Many of these are such distinct forms that it is not possible to say to which other genera they are most closely related. Moreover, it has been doubted, by one worker or another, whether about a third of the genera should belong in the family at all. Recently there have been some very useful anatomical studies that have thrown new light on relationships within the family, and within the whole suboscine complex; but even so, it is not yet possible to make anything like a final classification of the family. What we have to deal with, it seems, is a collection of very diverse evolutionary lines whose common ancestry lies far back in the stock of forest-inhabiting, insectivorous and frugivorous birds that gave rise to the present-day manakins and tyrant- fl)'catchers, as well as the present-day cotingas. There is no certainty, and probably never will be, that even the "best" cotingid genera may not be more closely related to birds that are now placed in one of the other two families than they are to some other "good" cotingas. Nevertheless, some arrange- ment has to be made, and it should be based on the best avail- able evidence, and should be justified. It is the purpose of the present paper to justify the treatment to be given to the family in Peters' Check-list. THE LIMITS OF THE FAMILY COTINGIDAE Sibley (1970) has given a useful historical review of the attempts to classify the passerine birds, and it is not necessary to go over the same ground. For the present purpose it need 1973 CLASSIFICATION OF COTINGIDAE (aVES) 3 only be said that, ever since the foundations of the present arrangement of the passerines were laid by Nitsch, Aliiller, Garrod, and others, and systematized by Sclater (1888), there has been great uncertainty as to where the dividing lines should be drawn between the three main suboscine famihes of the New World, the Cotingidae, Pipridae and Tyrannidae (with which are associated the two very small families, Phytotomidae and Oxyruncidae) . Recently, moreover, Olson (1971) has widened the contro\ersv bv rex'iving Pvcraft's contention that the Old \Vorld familv Eurvlaemidae mav be the closest relatives to the Cotingidae. According to this view, the Eurylaemidae and Cotingidae are derived from the primitive suboscine stock, and the Tyrannidae, by implication, are the result of a more recent evolutionary radiation. It seems entirely possible, how- ever, that much of the detailed resemblance between the special- ized frugivorous members of the Eurylaemidae and Cotingidae, which Olson stresses, may be convergent; but whether the resemblance is the result of relationship or convergence, it does not resohe the difficulty under discussion, that among the less specialized members of the Cotingidae (as usually considered) there are genera that approach the tyrant-flycatchers and mana- kins more or less closely. Olson's suggestion raises an important general point, that it may be incorrect to regard the specialized frugivorous cotingids as the culmination of the adaptive radiation of the family. Pre- vious classifications of the cotingas ha\e usuallv started with the apparently less specialized types, and have ended with the specialized frugivores such as the fruit-crows and bellbirds, and this is a convenient sequence which is followed here too (p. 13) ; but it must be admitted to be no more than a matter of con- venience. The e\oIutionary interaction between specialized fruit- eating birds and the fruits to which they are primarily adapted, belonging mainly to a small number of plant families (Snow,, 1971b), has obviously been a very long one; in fact, the evo- lutionar)- histor\' of the frugivorous stocks may be older than that of the more specialized insectivorous types. Previous classifications. Sclater, in the Catalogue of Birds in the British Museum, vol. 14 (1888), recognized 31 genera in the Cotingidae, dividing them from the Tyrannidae and Pipridae on the basis of tarsal scutellation. The relevant part of the key is as follows: a. Tarsus exaspidean. a'. Toes nearly free (as in the Oscines) . Bill incurved, hooked 1. Tyrannidae 4 BREVIORA No. 409 Bill straight, pointed 2. Oxyrhamphidae [^Oxyruncidae] b'. Toes more or less united 3. Pipridae b. Tarsus pycnaspidean. Bill elongated, compressed, not serrated 4. Cotingidae Bill short, conical, serrated 5. Phytotomidae In this classification, all three famiUes were distinguished from the Eurylaemidae by the conformation of the manubrium of the sternum, which, as Olson (1971) has since shown, may show considerable intrageneric variation in the characters that have been used to separate families. Although he was aware of the fact that the main artery of the thigh is the sciatic in Rupicola and the femoral in all the other cotingas studied, Sclater did not consider this to be of sufficient importance to justify removing Rupicola from the family. He placed it in a subfamily, including with it Phoenicircus (whose thigh arteries had not been studied and still have not been) for no very convincing reason. Ridgway (1907), in prefacing his very detailed key to the family Cotingidae, pointed out that the nature of the tarsal scutellation was not so invariable within the famiUes as Sclater supposed. He followed Sclater's arrangement almost exactly, however, the only changes being to replace Laniisoma in the family (placed in the Pipridae by Sclater), remove Laniocera to the Pipridae, and add to the family eight genera that are now generally considered to be tyrannid and a ninth, Xenop- saris, which has subsequently been bandied between the two families. In spite of his caveat on the variability of tarsal scutellation, all these changes from Sclater's treatment were made on the basis of this character, except for Rupicola, which Ridgway placed in a separate family on the basis of the arteries of the thigh. Ridgway also divided Sclater's Tityra into two genera, Tityra and Erator, largely on the basis of tarsal scutel- lation. The next important classification of the family was that of Hellmayr (1929). Again, he followed Sclater and Ridgway closely as far as most genera were concerned. He agreed with Ridgway in including Laniisoma, but difTered from him in also keeping Laniocera in the family. He removed the eight genera of small flycatcher-like birds that Ridgway had brought into the family, as well as Xenopsaris. Like Ridgway, he treated Rupicola as a separate family. He did not subdivide the genus Tityra. 1973 CLASSIFICATION OF COTINGIDAE (aVES) 5 These three classifications agreed in large part, in regard to the genera admitted into the family. Ridgway's sequence, how- ever, departed radically from that of Sclater. Both sequences were simply the result of the keys which they devised, and they do not appear to have been intended to reflect any evolutionary development, such as from more primitive to more specialized types. Thus Ridgway began the family with the peculiar genus Phoenicircus and followed it with the vers- different genus Laniisoma simply because the first division of his key dealt with the union (or non-union) of the outer toe to the middle toe. It is noteworthy that these three authors all questioned the correctness of placing Attila (and Sclater also Casiornis) in the Cotingidae rather than in the Tyrannidae. Ridgway expressed the same doubts about Rhytipterna and Lipaugus and the fly- catcher-like genera, which he separated from the Tyrannidae "solely on account of the different character of the tarsal en- velope." Aleyer de Schauensee f 1966) followed Hellmayr fairly closely. He replaced Xenopsaris in the family, however, with a footnote indicating that it may be tyrannid, and he added the newly disco\'ered genera ^aratornis and Conioptilon, and the recently described genus Pseudattila. In his later book (1970) he made a more important change. As a result of recent anatomical research, he removed five genera and placed them in the Tyran- nidae. These were the controversial Attila (with Pseudattila), Casiornis, Laniocera, and Rhytipterna. He also modified HeU- mayr's order of genera in two ways. He placed Ampelion, ^aratornis, and Doliornis later in the sequence, associating them with the other Andean genera Pipreola and Ampelioides rather than with the eastern Brazilian genus Phibalura. Secondly, he placed Phoenicircus at the end of the family, after Procnias, instead of at the beginning. The debated genera. The eight genera that Ridgway alone included in the Cotingidae are now generally agreed to belong in the Tyrannidae, and they will not be discussed further {Microtriccus, Ornithion, Tyrannidus, Idiotriccus, Elainopsis, Hylonax, Ramphotrigon, Syristes). Apart from these, there are nine genera whose position has been debated. Lanhsoma Laniisoma was placed in the Pipridae by Sclater (under the name Ptilochoris) on the basis of its foot structure and tarsal 6 BREVIORA No. 409 scutellation. Ridgway replaced it in the Cotingidae, linking it with Phoenicircus, also on the basis of its foot structure. Curi- ously enough, among the other characters that he listed for these two genera, he gave for Phoenicircus "adult males with sixth and seventh primaries shortened, especially the seventh, which is strongly bowed . . .," and for Laniiso?na "adult males with sixth and seventh primaries normal"; but in fact males of Laniisorna have the seventh primary modified, as Sclater noted, though to a much less extreme degree than Phoenicircus. Thus Laniisorna and Phoenicircus agree in structural characters of both wing and foot, though they are very different in plumage. Laniisorna has not been dealt with in recent anatomical studies, being rare in collections, and nothing is known of its l^ehavior. Its distribution is similar to that of Phibalura, with a south- eastern Brazilian population and restricted populations along the eastern slopes of the Andes. It also resembles Phibalura in some points of plumage. On present evidence Laniisorna is best kept in the Cotingidae. It is in accordance with the Kttle that can be inferred about its relationships to place it between Phoenicircus and Phibalura, but this creates difficulties, as discussed under Sequence. Phoenicircus Phoenicircus seems to be a very isolated genus, in spite of the characters shared with Laniisorna, as noted above. All recent classifications have placed it in the Cotingidae, though Hellmayr (1929) remarked (footnote, p. 92) that it is "a genus of doubtful affinity, perhaps more nearly related to the Rupico- lidae or constituting a family by itself." Sclater also linked it with Rupicola, in the subfamily Rupicolinae, but on what real evidence is not clear. In his key to the subfamilies the designa- tion for the Rupicolidae, "Tarsi very strong; gressorial," refers only to Rupicola; in fact only the crest (very different in the two genera), the presence of modified primaries in the male (but difTerent ones, and many cotingid genera show such modi- fications) and the general colour serve as a very tenuous link between the two genera. The anatomy of Phoenicircus has not been investigated, and very little is known of its behaviour except that it appears to have some form of communal display ( Olalla, 1 943 ) . The united toes, the colour of the plumage, and the communal display all suggest that it may be an overgrown manakin. At least, this 1973 CLASSIFICATION OF COTINGIDAE (aVES) 7 possibility desen-es as much consideration from future workers as the more u^ial idea that it is cotingid; but until more is known about the genus it is preferable to retain it in the Cotingidae. Attila Superficiall)-, its narrow hooked beak, flycatcher-like plumage, general behaviour (Skutch, 1971 ), and habit of nesting in niches in tree trunks all tend to suggest tyrannid affinities for the genus Attila. Sclater, however, following Sundevall, placed it in the Cotinoidae, forming a subfamily Attilinae with Casiornis, on the basis of the tarsal en\elope. .Mthough he expressed doubt about the correctness of this, subsequent authors all kept Attila as cotingid, though usualh' with reservations, Ridgwa\- pointed out that the tarsus is not in fact typical of the cotingas. Meyer de Schauensee ( 1 966 ) followed the con\entional treatment in his 1966 work, but in 1970 transferred Attila to the Tryannidae, as a result of ^Varter's studies of the skull. ( Warter's work is not yet published, although cited with a publication date of 1966 in Meyer de Schauensee, 1966: 314, footnote.) The more recent studies of the syrinx b\- Ames i 1971 ) confirm the tyrannid character of Attila, whose syrinx is closely similar to that of Mxiarchus. It mav be noted that these two genera have the same kind of nest-site. There is little doubt that Attila should be remo\ed from the Cotingidae, and placed in the Tyrannidae near to Myiarchus. The anatomy of Pseudattila has not been studied, but it seems so close to Attila (not having been separated as a genus until 1936) that it may safelv be placed next to it. Casiornis Most of what has been said about Attila applies equally to Casiornis. Structurally it is close to Attila (in tarsus, beak, and syrinx [Ames]), but little seems to be known about it in life and its nest has apparently not been found. On present kno\\l- edge is seems best to place it close to Attila and remo\e it from the Cotingidae. Lan^ocera Sclater placed Laniocera in the Cotingidae, whWc notin<: that in foot structure it resenibles the Pipridae. Ridgway, on tlie basis of its foot structure fcxaspidean tarsus and extensi\ely 8 BREVIORA No. 409 coherent toes), placed it in the Pipridae, but it was replaced in the Cotingidae by Hellmayr in spite of its foot structure because of the general siniilarit\- of its appearance to Lipaugus and Rhytipterna. Recent studies of its osteology and svrinx, howe\-er, suggest strongh- that its correct place is in the T\ ran- nidae, presumably close to Attila. The genus is litde known in life, and no nest has been found. Rhytipterna There was general agreement in placing Rhytipterna in the Cotingidae, close to Lipaugus, until Meyer de Schauensee re- mo\-ed it to the Tyrannidae along with the genera Attila and associated genera, as the result of recent studies of the osteology and syrinx. This anatomical e\'idence seems too strong to be set aside. The \ery peculiar tarsal scutellation of Rhytipterna separates it from all other genera, whichever family it is placed in. Nothing seems to be known of the genus in life, but it mav be guessed that the spiky tarsal scutes are in some way connected with its nest-site or e\'en its roosting behaviour. In the former case they might be analogous to the horny processes on the plantar surface of the tarsi of some hole-nesting non-passerines such as jacamars (in which case there would be a further link with Attila, a niche-nester), or in the latter case to the tarsal serrations of the tree-roosting Tinamus. Field obser\'ations on all the genera that are here thought to be close to Attila are badly needed. Lipaugus There has been consistent agreement among systematists that Lipaugus is a cotinga, but on the basis of his syringeal studies Ames (1971) has recendy suggested that "Lipaugus may later be added to this group" (Attila and related genera, which must be transferred to the Tyrannidae). The only specimen avail- able for study, however, was damaged, and no firm conclusion was possible. The two closely related species L. vociferans and L. unirufus seem to be typical cotingas, so far as they are known in life. Males of L. vociferans display in leks (the advertising display being vocal, not visual), and the nest (L. unirufus, Skutch, 1969; 1973 CI.A.--..SIK1C.ATIOX OF COTINGinAE ''aVES) 9 L. vocijcrdiis. Willis, personal coninumication) is a minute stick platform such as other medium-lars^e cotingas (Xipholena, Procnias) build, but no t\rannids so far as known. Moreo\er, in spite of the dull o;re\, brown, or greenish plumage of most of the species, L. streptophorus has a brilliant magenta collar encircling the anterior part of the body, and \'olker (1952) has shown that the pigment in these feathers, as in the red feathers of other cotingids, changes to orange under mechanical pressure. In summar\-, the e\idence, though scant}', suggests that Lipaugus should be retained in the Cotingidae. The peculiarities of the syrinx may well be related to the extraordinan' de- \'elopment of \ocal display in the genus, and may thus be a specialization rather than an indication of relationship. In this connection it would be interesting to examine the syrinx of a species that has not been reported to ha\e an unsually loud voice, e.g., L. streptophorus. It is not surprising that Hellmavr was impressed b\' the general similarity of Lipaugus, Laniocera, and Rhytipterjia. In addition to their general resemblance they show a remarkable parallelism in geographical \-ariation. In each genus there is a grey form widespread in South America east of the Andes, which is replaced by a chestnut-brown form in Central .\merica and South America west of the Andes. Xenopsaris Sclater placed this genus in the Tyrannidae, as did most authorities of his time, although the single species, X . albinucha, was originally ascribed to the cotingid genus Pachyramphus. Ridgway placed it in the Cotingidae, noting that its tarsus is not exaspidean, thus not conforming to his main criterion for the Tyrannidae. Hellmayr removed it again to the T\rannidae, and Meyer de Schauensee replaced it in the Cotingidae. This genus has thus been switched back and forth four times. As Meyer de Schauensee notes, "anatomical in\estigation is needed to settle the disputed status of this gentis," but it is not abundant in collections and its anatomy is still unknown. If it is a cotinga, it is unique in its habitat; no other member of the family is found primarily in reedbeds and other rixerine vegetation, Init many t\rannids are. The e\idence for the t\rnn- 10 BREVIORA No. 409 nid affinities of Attila and related genera shows that undue reHance should not be placed on tarsal characters alone. The evidence is clearly inadequate for a proper decision, but on balance it seems best to place Xenopsaris in the Tyrannidae. RUPICOLA As mentioned above, Sclater kept Rupicola in the Cotingidae in spite of the fact that the main artery of the thigh is the sciatic (as in the Tyrannidae) and not the femoral (as in the other cotingas studied). It is uncertain what weight should be gi\'en to this character until the thigh arteries of all the cotinga genera have been studied, but it should be noted that there is apparently no other character which links Rupicola with the tyrant-flycatchers rather than with the cotingas. On the basis of the thigh arteries, Ridgway placed Rupicola in a separate family, and Hellmayr followed him presumably for the same reason. Subsequent authors have not been consistent: for instance, Meyer de Schauensee (1966) maintains the family Rupicolidae, while Sibley (1970) has replaced Rupicola in the Cotingidae. Sibley's decision was made before any evidence from egg- white proteins was available. Since then, he has analysed the proteins from a fresh ^gg that I sent him from southern Guyana (Snow, 1971c), and he writes as follows: "Your fine speci- men has been 'run' at least three different times with different comparisons, both in isoelectric focusing and also in the standard type of net charge electrophoresis. In all of these it was com- pared with excellent specimens of Pachyramphus, with Phyto- to7na, Manacus, at least one or more t^rannid flycatchers, an oxenbird, and an antbird. Without any question, in all of these comparisons, Rupicola agrees most closely with Pachyramphus. Thus I think it is entirely in keeping with the egg white evi- dence, which seems to me satisfactory, to include Rupicola in the Cotingidae." If Rupicola is retained in the Cotingidae, its isolated position in the family can be given recognition by placing it at the end of the sequence of genera, after the very difficult bellbirds (Procnias) . The two genera thus represent two distinct t)pes of specialized frugivores, in which the pure fruit diet is associated with marked sexual dimorphism and the evolution of elaborate courtship display. 1973 CLASSIFICATION OF COTINGIDAE (aVES) 11 RECOGNITION OF DEBATED GENERA Most cotinfrid a:enera are so distinct that there has been c^eneral agreement about their recognition, even though their affinities may be in doubt. There are, however, a few genera that have been recomized bv some authorities but not bv others, and these are discussed below. In a family like this, where rela- tionships are so hard to detect and the ratio of genera to species is so high, it seems far more satisfactory to use generic names, when possible, to indicate affinity rather than to emphasize minor difTerences between forms that are certainly quite closely related. Ampeliox, Doliornis and Zaratornis Taczanowski in his original description admitted the nearness of Doliornis to A?npeIion. but it w^as maintained as a separate genus apparently without dissent until Bond '1956) suggested that both it and ^aratornis should be merged with Ampelion. Ampelion and Doliornis are both Andean, occurring at high altitudes; they share the same plumage colours (greys and chestnut-browns), and they agree closely in their concealed red-brown nuchal crest. The main structural difTerence is that Doliornis has a narrower and less hooked beak than Ampelion. There has been no study of the internal anatomy of Doliornis, of which only two specimens are known from a single locality in Peru. It seems most likely that Doliornis sclateri has been derixcd from an isolated fragment of Ampelion stock, and the maintenance of the monotypic genus seems unnecessary. The best treatment of ^ar atom is is more difficult to decide. Bond recommended merging it with A?npelidn because of the general similarity of its plumage, especially to that of A. ruf axilla. The red-brown nuchal patch is present, though reduced. Its relict distribution in the Andes, like that of Doliornis, suggests an offshoot from early Ampelion stock. According to J. Farrand, Jr. 'personal communication), it is \"ery like Ayyipelion in general beha\iour. Farrand writes: "I spent several davs at Zarate in the Department of Lima with the late Dr. Maria Koepcke in 1964, and saw the species daily. In its general behavior it is \'erv like Aynpelion rubrocristata. In a manner verv reminiscent of that species it often 'pops up' suddenh' onto a dead snag and sits upright, looking about rather ner\ously. The flight of ^aratornis is very similar to that of Amfjelion. and both species approach a perch flying low and making a final 12 BREVIORA No. 409 upward sweep, rather like that of a shrike or kestrel." On the other hand, Lowery and O'Neill (1966) state that its skull, which they have examined, is "so distinct as to preclude making ^aratornis congeneric with Ampelion." Nevertheless, they place its skull next to that of Ampelion in what they describe as a "nicely graded series" of six genera, and the exact differences are not specified. Without knowing to what extent skull struc- ture can be modified in response to differences in feeding be- haviour in otherwise closely related forms, it is not possible to weigh the skull differences against the other evidence that suggests close relationship to Ampelion. Provisionally, I recom- mend merging ^aratornis with Ampelion. Pachyramphus and Platypsaris The closeness of these two genera has been obvious to every- one who has studied them, though nobody has Hked to take the step of formally merging them. Ridgway gave as distinguishing characters the greater size of Platypsaris, the less extreme special- ization of the modified ninth primary in the male, and the more cylindrical, less broad, beak. These hardly seem to warrant the recognition of a separate genus in view of the diversity of size and other characters within the genus Pachyram.phus. Moreover, the partly concealed white scapular feathers charac- teristic of Platypsaris are also present in Pachyramphus surina- mus. Both genera build gobular, bulky nests, with a side entrance, but that of Platypsaris is typically suspended from the tip of a drooping branch, while that of Pachyramphus is typicallv supported in a vertical or horizontal fork. The four species of Platypsaris are almost completely allopatric (see later, p. 21). They overlap widely with several species of Pachyramphus, and their greater size is presumably one of the means by which they avoid competing with them. None of the distinctions seem great enough for generic separation, and thev are best united under Pachyramphus, the older name, as Bond (1959) has already recommended. Tityra and Erator At first glance the three species in this group [cay ana, semi- jasciata, and inquisitor) seem very close to one another. Ridgway, who laid stress on structural characters that might be used to separate genera, phced the last-named species in Erator on the basis of its feathered lores and orbital region. 1973 CLASSIFICATION OF COTINGIDAE (aVES) 13 taxaspidean (not pycnaspidean) tarsus, and broader and flatter bill. Hellmayr and most subsequent authors have preferred to recognise a single genus, although Wetmore (1926, 1927) has upheld Erator for the reasons gi\en by Ridg\vay. It is almost inconceivable that this group of hole-nesting cotingas, whose plumage, modified ninth primary, general proportions, behav- iour, and calls are so similar, are not closely related and monophyletic. The presence or absence of areas of bare skin clearly need not be considered of generic importance in the Cotingidae (cf. Cephalopterus, Procnias) ; beak shape is pre- sumably adapted to feeding habits (though exactly how in these species is not known) ; while the difference in tarsal scu- tellation must be taken to indicate that \ariation in this character may e\-oh'e relatively easily without major modifica- tion of other characters. THE SEQUENCE OF GENERA As is true for all diverse families, no linear order can express relationships satisfactorily. This would be true even if the relationships were perfectly understood. In the Cotingidae we have a large number of isolated genera whose affinities are quite uncertain, and it seems best to use a sequence that keeps as close as possible to what has been customary while taking account of the more well-based conclusions from recent studies. Sclaters sequence, as already mentioned, followed directly from his key, and subsequent advances in the systematics of the family make it obsolete. Ridgway's sequence also followed directly from his key, and although it is to this extent artificial it has been generally followed. The reason for this is probably that it seems a more or less natural sequence — • except that it starts with Phoenicircus, but this genus would be anomalous in an\- position except the beginning or the end. x'\fter this awk- ward start, the sequence proceeds with five genera ( Laniisoma, Phibalura, Heliochera {^Ampelion) , Tijuca, and Ampelion {=Carpornis) ) which are undoubtedly cotingid but un- specialized — that is to sav, not specialized towards the large, frugivorous, highly dimorphic types that seem to be the culmi- nation of evolution in the family (but see the caveat on p. 3). They are not \ery specialized in beak shape, and have plumages in which yellows, greens and greys predominate. Moreo\-er, these apparently unspecialized genera are distributed peripherally to the main lowland forest region, which is the headquarters of 14 BREVIORA No. 409 most of the specialized friigi\ores. It is reasonable to regard these genera as closer to the early cotingid stock than the larger, brilliantly coloured fruit-crows and other bizarre forms that are placed towards the end of the generic sequence. Four genera follow {Porphyrolaema, Cotinga, Xipholena and Carpodectes), which consist of medium-sized speciahzed frugi- \orcs with striking colours (purple, wine-red, blue or white) in the male plumage and duller-coloured females. There then follow (omitting the genera that have been placed in the Tyrannidae by all other authors) three genera of small or very small frugivorous forms {Pipreola, I odo pleura, Calyptura), the attiline genera and Lipaugus, the tityrine genera, and finally the large and diverse fruit-crows with the very specialized Ijell- birds {Procnias) placed at the extreme end. Ob\'iously the sequence in which the middle groups should be placed is to a large extent arbitrary, but this order does broadly represent an advancing sequence along the line of specialization for which the family is noted. Probablv for this reason, Hcllmayr followed it almost exactly (except in so far as he exluded or included the debated genera). Meyer de Schauensee also followed it, except for one major change: he placed Aynpelion and the two rehted genera Doliornis and ^aratornis (^the latter not described when Hellmayr wrote) some- what later in the sequence, before Pipreola and after the recent!)- disco\ered Conioptilon. I iDcliexe that the basis of Hellmayr's sequence can be retained, but that some changes are needed in order that it should reflect as far as possible the most likely relationships in the light of recent knowledge. The suggested sequence, with annotations and justification of changes, follows: — ' Plinrnirir( //? "I retained at llic head of the list liecausc it is }- so isolated but, so far as it shows affinities with J other genera, may be linked with Lrmiisoma. Lnniisoma Pliihalurn Tijnra Cfi) pniiiis Ampclion (including Doliornis and 7.ar(itornis these four genera appear to be products of the same evolutionary centre (the SE Brazilian high- lands, from which the fust two have spread lo the eastern slo])('s of the Andes) shares some characters (cspeciallv nuthal crest) with Phihiilura 1973 Cr.ASSIFICATIOX OF COTIXOIDAE 'aVES) 15 Pil)reola Ainpelioides lodoplewa Calx j)t inn Lipfiugus Cliironlla almost aitainlv closclv related: perhaps an ollshoot frdin Aiulean stock sliariiig a rominon j aiucsti\ Avitli A III Ix'lioii. and more remotely \\'itli the four precedinji; SF. Rrai'iliaii genera. of uncertain statns: small forest hirds not ob- viously related to each other nor to anv other genera: svrinx of lodoplcuKi of tvrannid l\pc (Ames 1971) i pioi)al)l\ belong together: not ohviowslv rihited f to other genera Paclni'diii pints (including Plnlxpsniis Tit\ra Porphxrolaeiiia Cotinga Xiphohna Carpodectcs Couioplilon Gyiinwderus Haematodeius Q_uernla Pxroderus Cephaloptcrus Perissocepludus Pracnins l)rol)abl\ fairh closeh related (plumage colours, modified niiuh [)rimar\ of nuilcs, beak shape) : not ob\ioiislv related to other genera medimn-si/cd specialized frugisores. sharing some ])linnage and structural characters: listed together in this sequence hv Ridg\sav and later authors. Xipliolcna and Cm podcctcs appc;n- to he closelv related, on some characters (Snow, H'Tli: but Anu;s lias fcuind marked dillei cnces in svringeal structiue and puts Carpodectcs in a dilfcrcni group from the other three, with Qiiriuhi. I-inther evidence is needed before U|)setting the existing order, which seems satis- fac tor\' in other respects. lesemhles Cai podeclcs in .skull characters, and seems to form a link between the above group and ('•xiiniodcnis. ^vhich it resembles especially in tlie great cle\elopmeiU of jiowder do'ivn (Lowery and OXeill, 1960). large, niainlv frugi\c)r()us cotingas. including the so-called fruit-crows. The secpience is the same as that adopted bv Ridgwav and followed by ;dl lalei ;nithors. except that Gxiiiiiodcnis is placed first instead of near the end. This en- ables Couioplilon and Cxni'iodcius to be placed together, and does not \iolate anv other known or picsuiued ;i(rniities. The position of Piocnias at the end is in iiccord with its highh distiiut svrinx. Riipirola 16 BREVIORA No. 409 Species Limits A rather large proportion of all genera of cotingas consists of groups of parapatric forms (mainly or entirely allopatric forms that replace one another geographically). The differences be- tween the allopatric populations range from very slight to \ery marked, thus posing problems of e\ery degree of difficulty in the often arbitrary decisions that haxe to be made about specific limits. Each case needs to be treated on its own merits, but in fact there is too little e\idence in nearly every case to allow a reasoned decision to be made. It is noteworthy, howe\'er, that in two cases where verv similar forms come together {Phoenicircus carnifex and P. nigricollis, Xipholena punicea and X. lamelbpennis) they appear to behave as good species, admittedly on ver\' slender evidence (Haffer, 1970). The mem- bers of these two pairs are more alike than the members of most other allopatric pairs or groups of cotingas whose specific status is debatable. This should lead to caution in lumping allopatric forms into single species when only an arbitrary decision is possible. It is also very difficult, once one starts to lump allo- patric forms, to know where to stop; and awkward situations occur. Logically, if closely related allopatric forms are to be treated as conspecific, Xipholena atropurpurea should be treated as conspecific with one of the \'er\' closely related species which replace it geographically; but with which? As far as one can tell, it is somewhat intermediate between punicea and lamelli- pennis and there is no good reason to link it with one rather than with the other. For these reasons, I ha\'e been conservatix'C in my treatment of allopatric forms, retaining specific status for them unless there seems to be particularly good reason to merge them; and even this in\ol\'es subjectixe judgment with which it is easy to dis- agree. In order to make the classification gixen in Peters more useful for zoogeographic analysis, groupings of zoogeographic species and possible alternati\'e taxonomic treatments will be indicated in footnotes, where appropriate. Phoenicircus. The two very similar species, P. carnifex and P. nigricollis, meet in the region of the lower Tapajos and perhaps the lower Xingu and Tocantins (HafTer, 1970). It seems that the\' must exclude each other from their respecti\'e ranges, but information is quite inadequate to decide the point. There is no evidence for intergradation between them. On present evidence they are best kept as separate species. 1973 CLASSIFICATION OF COTINOIDAE (AVES) 17 PiPREOLA RiEFFERii and P. INTERMEDIA. Thesc two specics are closely related, differing chiefly in the terminal tail markings of intermedia, which are absent in riefferii, and, less strikingh , in other plumage characters. P. riefferii, which has a wide range in northern parts of the Andes, is replaced b\ intermedia in Bolivia and most of Peru. Hellmayr ^1929) treated them as specifically distinct, since specimens of both, showing no signs of intermediacy, have been collected near to each other in the Department of Libertad, intermedin at 2400 m and riefferii at 1200 m. Altitudinal data from elsewhere do not show inter- media as occurring consistently higher than riefferii, and in fact 1200 m is an exceptionally low altitude for either species; in the Department of Amazonas in northern Peru riefferii has been collected at 2300-2800 m, while further south intermedia had been recorded mainly at 2500-3000 m. Further informa- tion is needed on the distribution of these two species where they approach one another, but Hellmayr's opinion regarding their specific status appears sound. Pipreola arcuata is almost certainh a deri\ati\e of the riefferii- intermedia stock, from which it differs chiefly in its large size and barred underparts. It now o\'erlaps extensi\ely with them. Pipreola aureopectus, P. lubomirskii, P. jucunda, and P. pulchra. The last three of these ob\iously form a natural group, as Hellmayr recognized. P. aureopectus is more distinct, but pulchra forms a clear link between it and the other two. The male of pulchra is almost exactly intermediate in its crown colour, and the female is almost exactly like the female of aureopectus except for the lack of white apical margins to the wing feathers. All four species agree in the unique combination of red beak, yellow iris, and oli\e-grey feet. They replace one another from north to south in the Andes, in the order listed, with no overlap so far as known. It seems clear that thev should be treated as conspecific. Pipreola frontalis and P. chlorolepidota. The males of these two species at first sight appear \ery different, since the underparts of chlorolepidota are darkish green apart from the throat-patch, and those of frontalis are yellow. But the former .species retains a small \ellow area lateral to the throat-patch, where the feathers are brightest \ellow in frontalis, and the two species agree in the colour of the soft parts fin the adult male, white or vellow iris, orange or red beak and feet). Moreo\'er, the females arc almost identical in plumage. .\lthouQh they appear to be mainly allopatric, so far as can be told from the 18 BREVIORA No. 409 limited records, they overlap in southeastern Ecuador and prob- ably northeastern Peru, and so cannot be considered conspecific; but they ha\e probably speciated comparatively recently. lODOPLEURA FUSCA 0)1(1 I. ISABELLAE. ThcSC tWO fomiS replace one another, so far as known, but there are wide areas where neither has been found. Fusca has been found at f\vc localities in Guyana, one in Surinam, and one in eastern \'ene- zuela on the upper Caroni River; the nearest records for isabellae are from the upper Orinoco region, the Rio Negro and the middle Amazon (Obidos). Until more is known of the situation in the inter\'ening areas it seems premature to treat them as conspecific, as Hellmayr and others have suggested; they cer- tainly differ more than do Phocnicircus carnijex and P. nigri- collis. ft may be noted that lodofAeura pipra, the only other species in the genus, which occurs in southeastern Brazil, is also known from two specimens from Guyana. The possibility of long-distance migration by these small aerial cotingas of the trectops cannot be excluded. LiPAUGUS VOCIFERANS, L. UNIRUFUS and L. LANIOIDES. These three forms replace one another geographically. Hellmayr considered lanioides probably conspecific with vocijerans. but did not go so far as to merge them. No systematist has merged vocijerans and iinirufiis, although they are almost certainly more closelv related than vociferans and lanioides. L. rorifrrans and L. iinirufus between them occupy almost the whole of the lowland tropical forest region of Central and South America. L. vociferans occupies the greater part of the South American range, but does not occur north of the Orinoco (except near its mouth). L. unirujns does not extend east of the Alagdalcna \"allev in Colombia. There is thus an area in eastern Colombia and western Venezuela, containing some low- land forest, where neither occurs. The two species are \ery different superficially, since vociferans is all grey and unirufus all rufous-brown, but in beha\'iour and ecology the litde e\-i- dence there is suggests that they are rather similar, and the colour (lifTerence could be based on a small genetic difference. For the reasons gi\en in the introduction to this section I prefer, howe\"cr, not to treat them as conspecific. At first sight, /.. lanioides differs from vociferans less than docs unirufus. but the differences arc in fact probably more numerous and important. It is larger, with a browner rumn and undcrparts, and brownish fnot grey) wings and tail. It replaces vociferans in southeastern Brazil south of 20*^ S. The 1973 CLASSIFICATION OF COTIXCIIDAE (AVES) 19 separation at this point involves habitat as \vell as range, as vocijerans occurs in lowland tropical forest while la?iioides is found mainly in subtropical montane forest (but has also been recorded near sea-lc\el, perhaps as a result of \ertical migration) . Further, although little is known of lanioidcs in life it is hard to belie\e that it could utter the kind of piercingh' loud calls, in Icks, that are so characteristic of vocijerans where\er it occurs. It seems more likely that lanioides and the other mon- tane species of Lipaugus [strcptophorus. fuscocinereus, crypto- lophw: and subalaris) are comparatixeh' quiet birds, with a different social organization from voriferans and iinirufus. Hellma\r"s suggestion that lanioides and voeiferans are con- specific thus recei\es no support in the light of more recent knowledge. The genus P.achyramphus. \\'ith 1 1 species usually recog- nized (15 if Platypsaris is included . P(i< hxKnnphus is easih the largest cotingid genus. It is also unique in the famih" in the amount of geographical o\erlap between species; in extensixe areas up to five or exen six species may occur close together, even if not in exactly the same habitats. Some of the species ran be placed in groups that constitute zoogeographical species. The arguments f(;r and against recognizing larger specific imit^ than are currently recognized are discussed belo\\ . Pachyramphus castaxeus and P. cixnamomeus. P. castaneus differs from P. ( innamomeus principalh in its smaller beak and in the presence of a narrow grey stripe separating the chestnut of the crown from the paler brown sides of the head and neck. Othei-visc thc\ are closeh similar, and are the onh' two species in \vhi( h male and female are alike in plumage. MoreoNcr, both are characteristicalK found in lowkmd tropical forest, a habitat axoided b\ most of their congeners. P. casla- neus also ranges up into the subtropica.l zone. Between them, these two species apparentlv occupx' mo-^t of the tropical forest area of South and Ontral .\merica. In iioith- western \ cnezuela and eastern C!olombia, thi'ir ranges aj~)proach closeh : P. easlaneus occurs in the coastal mount, liiiv o{ north- ern \ cnezuela, and is rej)laced 1)\ /'. ( itniauion!: us round !/:ke Maracaibo and in the mountains inatus has a small black crown-patch in the male and a brown crown with no black border in the female. P. marginatus is a bird of lowland tropical forest, east of the Andes. P. albogriseus replaces it in limited areas of the Andes, west of the Andes, and in Central America, being found in a variety of habitats, tropical and sul^tropical. Where the two species approach one another, on the eastern slopes of the Andes of Peru and Ecuador and in northern Venezuela, P. albogriseus occurs only at high altitudes and P. marginatus in the lowlands. 1973 CLASSIFICATION OF COTINOIDAE ( AVES ) 21 Again, it seems clear that these two are related, but the little evidence a\ailable does not justifx' treating them as conspecific. Platypsaris. The close affinity of Pachyramphus and Plairpsaris, and the reasons for merging the two genera, ha\-e alreadv been discussed. Platypsaris consists of fi\e forms, usually treated as species, which are almost completely allopatric. Four of these forms between them occupy almost the whole of tropical and subtropical America except some montane and unwooded areas. They clearly constitute a zoogeograpical species, but the status of the constituent members is not clear. It is conx'enient, howexer, in discussing them to use the four specific names that are in general use. (Unfortunately, the specific name rufus is in use in both Pachyramphus and Platypsaris, and when the genera are merged it will be available only for the species cur- renth' known as Pach\ramphus rufus, which has priority.) The fifth form, P. niger, occurs onlv in Jamaica and is rather dis- tinct. The following discussion concerns only the four mainland forms. P. ininor, the central form, occupies the whole of the tropical forest of the Amazon basin and upper Orinoco system. To the south and east, P. rufus occurs in a wide \'aricty of more sea- sonal wooded habitats, tropical and subtropical. These two differ in a number of plumage characters in the male, and less strikinglv in the female; they differ also in size. There appears to be no intergradation between them, and moreo\-er their ranges ON'erlap in two areas, in northern Para and Maranhao in north- ern Brazil and along the base of the Andes in Peru and Boli\ia. These are areas where lowland tropical forest comes into con- tact with more open vegetation types (N Brazil) or with lower montane woodland (base of Andes). Their ranges also approach closelv, and may e\'en interdigitate, along the southern fringe of the main Amazonian forest in central Brazil. To the northwest, P. yninnr is isolated from P. hoyriochrous, also a forest bird, bv the sa\annas of \'enezuela and eastern Colombia. Neither species is found in the forested coastal mountains of northern \xnezuela. These two fonns differ slightK but constantl\- in both mal-^ and female plumage, and are much alike in size. In Central America P. hoDiot /nous and /'. aiihiiai', bolh in- habiting forest or more open woodland according to the habitats ax'ailable, are separated bv a gap between central Panama and central Costa Rica in which neither seems to occur. Tiie j:)oj:)u- lations on either side of this gap are \en- alike in male plumage. 22 BREVIORA No. 409 and in size. The females are more distinct, but some specimens of aglaiae resemble female hoynochrous closely. P. aglaiae is geographically and individually a \'ery \ariable species. Webster (1963), in a recent review of its variation, considered it con- specific with homochrous by modern standards, and the judg- ment seems sound. The evidence does not, however, warrant any further lumping within this group. It is obviously necessary to give specific status to minor vis-a-vis rufus. No critical test is possible of the status of minor vis-a-vis homochrous, but the constant plumage differences in both sexes suggest a similar degree of differentia- tion to that between such species as Lipaugus vocijerans and L. unirufus, whose ranges are also similar to those of minor and homochrous. It has been argued above that in the absence of evidence to the contrary it is best to give specific status to such fomis. The genus Cotinga. Six of the seven species usually recog- nized in this genus replace one another geographically: the Central American and northwest South American amabilis, ridgwayi and nattererii; maynana of the upper Amazon basin; cotinga of the lower Amazon-Guiana region; and 7naculata of eastern Brazil. (The seventh species, cayana, is more distinct morphologically and overlaps extensively with mayna?ia and cotinga.) The first three are ver\' similar to one another, and considered in isolation from the others might reasonably be treated as conspecific. C. cotinga and maculata are also very similar to one another, and Hellmayr suggested that they might be conspecific. They differ most conspicuously from the Central American forms by the deeper (not turquoise) shade of blue in the male and the more extensi\'e purple below. The sixth fomi, maynana. is at first glance rather distinct from the others. Hellma\r made the rather surprising suggestion that it might prove to be conspecific with the Central American forms, apparently on the basis of its similiar shade of turquoise blue. A more detailed analysis shows that these six forms exhibit a mosaic of characters that link one with another in many different ways f Table 1). The Central American group forms a unit, ha\ing in common six of the characters considered, and the two eastern forms constitute another unit, with se\en com- mon characters. C. maynana is certainly the most distinct form, having four peculiar characters, while none of the others has any character not found in at least one other form. In wing formula, however, maynana is closer to nattererii than the latter 1973 CLASSIFICATION OF COTINGIDAE AVES 23 ex be • S Hid n a, ^^ o w rt hJ P3 X! < (« h «*-i C « OJ rt u n ^ o ■a c o -a o TS a _3 g ^^ T3 CJ 5 to 2 t/5 r3 a. O >^ u i2 > CJ rt 3 o p p OJ 1J. a fc —' r^ ^ 15 3 rt o rt — ^ «-^ rt — H o CJ p CJ TS 3 > 4-1 ^_ ^J c; 9 'u 3 2 CJ 4-J p o _flj X a o -^ o ^v' o CU r> 3 3 ^ C, rt o «-t £ rt p oc t/i ^ OJ '55 . , c 1 'o 3 cr 3 4—1 ■4-1 .S 3 CJ CJ P I- o J ^5 -p ~ .SP o ;j •= trt (^ C C a "5 — ~ t^ « p c O •a i^ 2 ^ ^ ^ "o 3 «-) ^^ «J •a OJ "re o C ■^ rt >- J= ^ J^ £ .^ > U' o u o u CJ u K- ■*«j K r_v ra pj D ^ ^-' 0 a 3 r^ CJ « C o C S 3 — "- -- • — ^ 6 CJ t« D c O "2 rt •5" ?% W5 n 4-1 "tfl" bO ^^ o ^c a O 4-> ^ J= ^ ja! V E o a ^ > f 3 4-1 O x: 4-1 "o CJ -3 ■~ p ^ — a. 3 3 -^ r: O — -~ ^ rt O CJ be CJ P3 CJ ^ *c ^ ■^ (A o > ■3 r- .5 3 n >-- ^ ^ ^ V o r^ e c^ p u u u i) ■^ p •-* cd ca 3 o OJ rt (^ c a -^ ■" /-\ r^ S 3 ^ ~a -C >, CO <— 11 ^ Si CJ rt e- .a 3 a, ■a 1-. be r^ M^ •a r^ o 4) o « ^ t^ 1 CJ ci s .5 l^ VM bo c U1 o W5 _CJ u U-l *-i o T* n _a, £ ^ Q o 2 r* C3 rt ,__ 3 O CJ C T3 O 2 > — bJD .3 J D, o O on ■q a OJ 8 24 BREVIORA No. 409 is to the other two of the Central American group. Similarly, amabilis of the Central American group resembles the two eastern forms, cotinga and maculata, in its abbreviated 7th pri- mary, while ridgwayi, the remaining member of the Central American group, approaches them in the modification of its two outer primaries (but not in its 7th primary, which is normal) . It seems probable that the common ancestor of these six forms — perhaps during a period of contraction of the forests — was divided into a number of isolated populations, which evolved a number of small but consistent differences to which no obvious adaptive significance can now be attributed. The forest refuges suggested by Haffer (1970) provide obvious origins for five of the forms: amabilis in the Caribbean (Costa Rica) refuge, natter erii in the Choco refuge, maynana in either the Napo or the East Peru refuge, cotinga in the Guiana refuge, and macidata in the southeastern Brazilian refuge. The sixth form, ridgwayi, presumably e\ohed in a small forest refuge on the Pacific coast of Costa Rica and Panama, to which it is still confined. The rather exact geographical replacement of one form by another, and lack of any evidence of morphological intergradation, suggest that they have reached specific status to the extent of being reproducti\ely isolated, but that they are still not able to penetrate one another's ranges. In spite of the unequal differentiation of these forms, and the fact that they fall into two groups, with one isolated form, it is probably best to treat aU six as species, as has been the usual practice. The genus Xipholena. The status of the three forms has already been discussed, with the reasons for keeping them as three separate species. The genus Carpodectes. This genus consists of three allo- patric forms, of which two (nitidiis and antoniae) are very similar to one another while the third (hopkei) is a good deal more distinct. There is no doubt that together they constitute a zoogeographical species, but it is less clear whether they should be treated as one, two or three species. All three courses have been adopted or advocated. Slud (1964), however, gives some reasons for keeping nitidus and antoniae separate, and if this is done hopkei must also be gi\'en soecific rank. Such a treat- ment is in accord with the general policy adopted for other groups, as discussed above. The genus Cephalopterus. In accord with the general 1973 CLASSIFICATION OF COTINGIDAE (aVES) 25 policy adopted, as discussed above, the three allopatric forms of umbrellabirds should be kept as separate species. The differ- ences between them are at least as great as the differences between the overlapping forms of Xipholena and Phoenicircus. Literature Cited Ames, P. L. 1971. 1 he morphology of the syrinx in passerine birds. Bull. Peabody Mus. Nat. Hist., 37: 1-194, 21 pis. Bond, J. 1956. Additional notes on Peruvian birds II. Proc. .\cad. Nat. Sci. Philadelphia, 108: 227-247, 2 pis. . 1959. Fourth supplement to the check-list of birds of the West Indies (1956) . Acad. Nat. Sci. Philadelphia. 12 pp. Haffer, J. 1970. Art-Entstehung bei einigen Waldvogeln Amazoniens. J. Orn., Ill: 285-331. HELLMA'iR, C. E. 1929. Catalogue of birds of the Americas, Part VI. Field Mus. Nat., Hist., Zool. Ser., 13: 1-258. LowERY, G. H., AND J. P. O'NEILL. 1966. A new genus and species of cotinga from eastern Peru. Auk, 83: 1-9. Meyer de Schaiensee, R. 1966. The species of birds of South America. Narberth, Pa.: Livingston Pub. Co., Acad, Nat. Sci., xvii + 577 pp. . 1970. A guide to the birds of South .Vmerica. ^V'vuncwood, Pa.: Livingston Publishing Co. xvi -(- 470 pp., 50 pis., 21 figs. Olalla, a. M. 1943. .Algumas observagoes sobre a biologia das avcs e mamiferos sul-americanos. Pap. .\vuls. Dep. Zool. Sao Paulo, 3: 229-236. Olson, S. L. 1971. Taxonomic comments on tlie Emylaimidac. Ibis, 113: 507-516. RiDGVVAY, R. 1907. The birds of North and Middle America, Part IV. Bull. U.S. Nat. Mus.. 50: 1-973, 34 pis. ScLATER, P. L. 1888. Catalogue of the birds in the British Museum, Vol. 14, xix + 494 pp., 26 pis. Sibley, C. G. 1970. A comparative study of the egg-white proteins of pas- serine birds. Bull. Peabody Mus. Nat. Hist., 32: 1-131, 38 figs. Skutch, a. F. 1969. Life histories of Central .\mcrican birtls HI. Pac. Coast Avifauna, 35: 1-580, 28 figs. . 1971. Life history of the Bright-rumpcd Attila AttUa spadiceus. Ibis, 113: 316-322. Slud, p. 1964. The birds of Costa Rica. Bull. Am. Mus. Nat. Hist.. 128: 1-430, 2 pis., 1 fig. Snow, D. \V. 1971a. Display of the Pompadour Cotinga Xipholena puni- cea. Ibis, 113: 102-104. . 1971b. Evolutionarv aspects of fruit-eating by birds. Ibis, 113: 194-202. 1971c. Notes on die biology of (he Cock-of-the-rock (Rupi- cnla nipiroln) . ]. Orn., 112: 323-333. In press. Disti ibution, ecologv and evolution of the bell- birds {Procnias, Cotingidae) . Bull. Br. Mus. (Nat. Hist.) 26 BREVIORA No. 409 VoLKER, O. 1952. Der Lipochrome in den Federn der Cotingiden. J. Orn., 93: 122-129. Webster, J. D. 1963. A revision of the Rose-throated Becard. Condor, 65: 383-399. Wetmore, a. 1926. Observations on the birds of Argentina, Paraguay, Uruguay, and Chile. Bull. U.S. Nat. Mus., 133: 1-448, 20 pis. . 1972. The Birds of the Republic of Panama, Part 3. Smiths. Misc. Coll., 150: 1-631. 1 pi., 48 figs. Appendix Annotated species list of the Cotingidae; treatment to be adopted in Peters' Check-list of Birds of the World. (Zoogeographical species bracketed) [ Phoenicircus carnifex Phoenicircus nigiicollis Laniisoma elegans Phibalura flavirostris Tijuca atra Carpornis cucullatus Carpornis melanocephalus Ampelion rubrocristata Ampelion rufaxilla Ampelion sclateri Ampelion stresemanni Pipreola riefferii Pipreola intermedia Pipreola arcuata Pipreola aureopectus Pipreola frontalis Pipreola chlorolepidota Pipreola formosa Pipreola whitelyi Ampelioides tschudii lodopleura fusca *- lodopleura isabellae lodopleura pipra Calyptura cristata Lipaugus subalaris Lipaugus cryptolophus Lipaugus fuscocinereus Lipaugus vociferans Lipaugus imirufus Lipaugus lanioides Lipaugus streptophorus Chirocylla uropygialis Pachyramphus viridis Pachyramphus versicolor See discussion, p. 16 Formerly Doliornis sclateri Formerly Zaratornis stresemanni See discussion, p. 17 Includes lubomirskii, jucunda and piilcha (p. 17) See discussion, p. 17 See discussion, p. 18 See discussion, p. 19 ^^2 n 1973 CLASSIFICATION OF COTINGIDAE (aVES) 27 Pachyramphus spodiurus Pachyramphus lufus Pachyramphus castaneus Pachyramphus cinnamomeus Pachyramphus polychopterus Pachyramphus marginatus ..Pachyramphus albogriseus Pachyramphus major Pachyramphus surinamus {Pachyramphus aglaiae Pachyramphus minor Pachyramphus validus Pachyramphus niger Tityra cayana Tityra semifasciata Tityra inquisitor Porphyrolaema porphyrolaema {Cotinga amabilis Cotinga ridg^vayi Cotinga nattererii Cotinga maynana 'Cotinga cotinga Cotinga maculata Cotinga cayana {Xipholena punicea Xipholena lamellipennis Xipholena atropurpurea {Carpodectes nitidus Carpodectes an teniae Carpodectes hopkei Conioptilon mcilhennyi Gymnoderus foetidus Haematoderus militaris Querula purpurata Pyroderus scutatus {Ccphalopterus glabricollis Cephalopterus ornatus Ccphalopterus penduliger Perissocephalus tricolor pProcnias tricarunculata \ Procnias alba I Procnias averano I- Procnias nudicollis 'Rupicola rupicola ^Rupicola peruviana Total: 79 56 See discussion, p. 20 See discussion, p. 20 See discussion, p. 20 Includes homochrous; see discussion, p. 21 Possibly the first six species could be considered as one zoogeogiaphical species (p. 22) . See discussion, p. 24 See discussion, p. 24 See discussion, p. 25 All four Procnias species are closely related and replace one another al- most completely (Snow, in press) taxonomic species zoogeographical species ACME BOOK-'!'; :3.. INO. 100 CAM3RI0GE STr?££T CHARUESIOWN. MASS. ^^ Harvard MCZ Library _ 3 2044 066 302 795