J '-^/ <-- HARVARD UNIVERSITY Library of the Museum of Comparative Zoology ^^s7 BREVIORA MUSEUM OF COMPARATIVE ZOOLOGY Harvard University NUMBERS 351-379 1970-1971 CAMBRIDGE, MASSACHUSETTS, U.S.A. 1971 Edited by Penelope Lasnik CONTENTS BREVIORA Museum of Comparative Zoology Numbers 351-379 1970 No. 351. A new species in the nomeid fish genus Psenes from the equatorial eastern Pacific. By Richard L. Haed- rich. 7 pp. June 12. No. 352. The Chaiiares (Argentina) Triassic reptile fauna VII, The postcranial skeleton of the traversodontid Massetognathus pascuali (Therapsida, Cynodontia). By Parish A. Jenkins, Jr. 28 pp. July 10. No. 353. Anolis jacare Boulenger, a solitary anole from the Andes of Venezuela. By Ernest E. Williams, Osvaldo A. Reig, Pablo Kiblisky, and Carlos Rivero-Blanco. 15 pp. August 10. No. 354. Taxonomic and ecological notes on some Middle and South American lizards of the genus Ameiva (Teiidae). By Arthur C. Echternacht. 9 pp. September 18. No. 355. Generic relations and speciation patterns in the Cara- caras (Aves: Falconidae). By Francois Vuilleumier. 29 pp. November 30. No. 356. On a new species in a new earthworm genus from Puerto Rico. By G. E. Gates. 11 pp. November 30. No. 357. A review of the fossil Pelomedusidae (Testudines, Pleurodira) of Asia. By Roger Conant Wood. 24 pp. November 30. No. 358. South American anoles: Anolis apollinmis Boulenger 1919, a relative of A. biporcatus Wiegmann (Sauria, Iguanidae). By Ernest E. Williams. 11 pp. November 30. No. 359. The swimbladder as a juvenile organ in stromateoid fishes. By Michael H. Horn. 9 pp. November 30. No. 360. Mammals from the early Cenozoic of Chubut, Argen- tina. By George Gaylord Simpson. 1 3 pp. Novem- ber 30. No. 361. Additions to knowledge of the Argyrolagidae (Mam- malia, Marsupialia) from the late Cenozoic of Argentina. By George Gaylord Simpson. 9 pp. November 30, No. 362. Addition to knowledge of Groeberia (Mammalia, Marsupialia) from the mid-Cenozoic of Argentina. By George Gaylord Simpson. 17 pp. November 30. 1971 No. 363. Non-specificity of host-selection in the ectoparasitic snail Odostomia (Menestho) bisuturalis (Say) (Gas- tropoda: Pyramidellidae). By Robert C. Bullock and Kenneth J. Boss. 7 pp. January 8. No. 364. A new scincid lizard from Bougainville, Solomon Islands. By Allen E. Greer and Fred Parker. 1 1 pp. January 8. No. 365. Characters and synonymies among the genera of ants. Part IV. Some genera of subfamily Myrmicinae (Hymenoptera: Formicidae). By William L. Brown, Jr. 5 pp. January 15, No. 366. Pulsed sound of the porpoise Lagenorhynchus australis. By William E. Schevill and William A. Watkins. 10 pp. January 15. No. 367. Micromischodus sugillatus, a new hemiodontid characin fish from Brazil, and its relationship to the Chilo- dontidae. By Tyson R. Roberts. 25 pp. January 15. No. 368. Structural habitats of West Indian Anolis lizards I. Lowland Jamaica. By Thomas W. and Amy Schoener. 53 pp. January 29. No. 369. Lithophaga aristata in the shell-plates of chitons (Mol- lusca). By Robert C. Bullock and Kenneth J. Boss. 10 pp. January 29. No. 370. Ecological observations on a little known South American anole: Tropidodactyliis onca. By James P. Collins. 6 pp. March 31. No. 371. A new species of bromeliad-inhabiting galliwasp (Sauria: Anguidae) from Jamaica. By Albert Schwartz. 10 pp. March 31. No. 372. The paleontology and evolution of Cerion II: age and fauna of Indian shell middens on Curacao and Aruba. By Stephen Jay Gould. 26 pp. March 31. No. 373. The Chanares (Argentina) Triassic reptile fauna. VIII. A fragmentary skull of a large thecodont, Liipero- suchus fractus. By Alfred Sherwood Romer. 8 pp. March 31. No. 374. The fishes of the Malaysian family Phallostethidae (Atheriniformes). By Tyson R. Roberts. 27 pp. June 15. No. 375. Structural habitats of West Indian Anolis lizards II. Puerto Rican uplands. By Thomas W. and Amy Schoener. 39 pp. June 15. No. 376. Podocnemis venezuelensis, a new fossil pelomedusid (Testudines, Pleurodira) from the Pliocene of Venezuela and a review of the history of Podocnemis in South America. By Roger Conant Wood and Maria Lourdes Diaz de Gamero. 23 pp. June 15. No. 377. The Chanares (Argentina) Triassic reptile fauna IX. The Chaiiares Formation. By Alfred Sherwood Romer. 8 pp. June 15. No. 378. The Chanares (Argentina) Triassic reptile fauna X. Two new but incompletely known long-limbed pseudosuchians. By Alfred Sherwood Romer. 10 pp. June 15. No. 379. The Chanares (Argentina) Triassic reptile fauna XL Two new long-snouted thecodonts, Chanaresuchus and GuaiosLichus. By Alfred Sherwood Romer. 22 pp. June 15. INDEX OF AUTHORS BREVIORA Museum of Comparative Zoology Numbers 351-379 1970-1971 No. Boss, Kenneth J 363, 369 Brown, William, Jr 365 Bullock, Robert C 363, 369 Collins, James P 370 Diaz de Gamero, Maria Lourdes 376 Echternacht, Arthur C 354 Gates, G. E 356 Gould, Stephen Jay 372 Greer, Allen E 364 Haedrich, Richard L 351 Horn, Michael H 359 Jenkins, Parish A., Jr 352 KiBLisKY, Pablo 353 Parker, Fred 364 Reig, Osvaldo a 353 Rivero-Blanco, Carlos 353 No. Roberts, Tyson R 367^ 374 RoMER, Alfred Sherwood 373. 377, 378, 379 ScHEviLL, William E 366 Schoener, Thomas W. and Amy 368. 375 Schwartz, Albert 37I Simpson, George Gaylord 360, 361, 362 Vuilleumier, Francois 355 Watkins, William A 366 {j'li^^^'''' . BREVIORA Mmseuim of Compsirative Zoology Cambridge, Mass. 12 June, 1970 Number 351 A NEW SPECIES IN THE NOMEID FISH GENUS PSENES FROM THE EQUATORIAL EASTERN PACIFIC^ Richard L. Haedrich- Abstract. Psenes sio n. sp. is based on five specimens 23-66 mm SL from the equatorial eastern Pacific Ocean. The new species belongs to the species-group (others are P. pelliicidiis, P. maculatus, and P. arafurensis) with large, laterally compressed, knifelike teeth in the lower jaw. P. cya- nophrys and P. whiteleggii have small conical teeth in both jaws. The new species is characterised by its light color, long pelvic fins, two weak anal spines and 23-24 rays, 18-19 pectoral finrays, and 36-38 vertebrae. In a recent unpublished yet widely distributed manuscript (Haedrich and Horn, 1969), a new species of Psenes was included in a key (p. 36). It was stated explicitly that use of the name in the key did not constitute publication, and it was indicated that a formal description would soon appear in a review of the entire genus. Other responsibilities, however, have virtually brought this work to a halt, and early completion of the review does not seem likely. Therefore, in order to avoid the nomenclatural debacle that I see developing, this note has been prepared describing the new species. The specimens upon which the new species is based were made available by J0rgen Nielsen, and are housed in Universitetets Zoo- logiske Museum, Copenhagen (ZMC). They were collected by Dr. Nielsen on Step-I, a cruise conducted by the Scripps Institution of Oceanography. The manuscript has been read by Richard H. Backus and Giles W. Mead. Figure 1 was drawn by E. Leenders. 1 Contribution No. 2486 from the Woods Hole Oceanographic Institution. 2 Woods Hole Oceanographic Institution, Woods Hole, Mass., and Museum of Comparative Zoology, Harvard. 2 BREVIORA No. 351 Portions of this work were supported by a United States Govern- ment Grant under the Fulbright-Hays Act, the Johs. Schmidt Fund, and National Science Foundation grant GB- 15764. Among stromateoid fishes, the genus Psenes is distinguished in having two dorsal fins with the first dorsal originating over or before the pectoral insertion, persistent thoracic pelvic fins, a deep to moderately elongate body, and teeth present in some species on the palatines and/or basibranchials but never on the glossohyal. The genus, its relationships, and the nominal species were treated in a general way by Haedrich (1967), though recent findings will modify this account somewhat. Within Psenes, there seem to be two species groups. One group (including P. cyanophrys and P. whiteleggii) is characterized by small, conical, slightly recurved teeth in both the upper and lower jaw. The other group (in- cluding P. pellucidus, P. arajurensis, P. maciilatus and the new species) has small, conical, slightly recurved teeth in the upper jaw and large, laterally flattened, bladelike teeth in the lower jaw. In recognition of the considerable contributions to marine ich- thyology made by the Scripps Institution of Oceanography, the new species will be known as Psenes sio n. sp. Figure 1 Material. Five specimens, 23-66 mm SL, all in ZMC; those marked with an asterisk* have been X-rayed: *1 spec, 60 mm SL, HOLOTYPE, Step-I sta. 23, 11°10'S 80°01'W, 17 Oct. 1960, 2250-2345 hrs., 0-90 m, 5' net, surf. temp. 17.8°C. PARA- TYPES: *1 spec, 66 mm SL, Step-I sta. 80-1, 1°24'S 94°55'W, 2/3 Dec. 1960, 2335-0230 hrs., high-speed net, battered. 2 spec, 26 & 44 mm SL, Step-I sta. 73-1, 4°22'S 95°04'W, 2 Dec. 1960, 0200-0500 hrs., high-speed net, very battered, smaller spec, cleaned-and-stained. *1 spec, 23 mm SL, Step-I sta. 80, 1°59'S 94°55'W, 2 Dec. 1960, 2200 hrs., dipnet-nightlight, surf. temp. 22°C. All specimens are immature. Diagnosis. An elongate, compressed, light-colored Psenes with large, compressed close-set teeth in the lower jaw, long pelvic fins, two weak anal spines and 23-24 rays, and 36 to 38 vertebrae. Description. Individual proportions and counts are presented in Table 1. The body is elongate, the maximum depth of larger specimens 1970 NEW NOMEID FISH 3 being around 30 per cent of the standard length, and is very com- pressed. The caudal peduncle is tapered, compressed, and some- what elongate. The musculature, though firm, is translucent, par- ticularly along the anal fin base and over the viscera. The two dorsal fins are scarcely divided. The first dorsal fin originates over the edge of the opercle and comprises 10 to 12 thin brittle spines, the second originates just behind mid-body and comprises 23 to 25 long rays. The entire fin folds partially into a very shallow groove and terminates behind the end of the anal fin. The anus is at mid- body, in a slit. The anal fin commences shortly behind the anus under the third or fourth ray of the second dorsal, and is composed of two weak spines and 23 or 24 long rays. The muscles for ele- vating the anal rays can be clearly seen, as can the basal elements of both median fins. The pectoral fin is long and fairly broad, with 18 or 19 rays; its base is incUned about 45° to the vertical. The pelvic fins are very long; they insert under the middle or end of the pectoral fin base and extend beyond the anal origin, and are composed of one short spine and five long branched rays. The caudal fin, broken in most specimens, is apparently long and forked; the small elements preceding the principal rays extend well forward on the peduncle. The cycloid scales are very small, and do not appear to extend significantly onto the bases of the median fins. The scales are extremely deciduous, and most are gone; the count of scale pockets along the lateral line of the holo- type is ca. 85. The skin is thin; the subdermal mucus canal system is but httle developed, and the body pores are very small or wanting. The head is around 35 per cent of the standard length, its profile sloping. The skin of the top of the head is naked, and pores are clearly visible, particularly those over the head of the hyomandibu- lar. The eye is of moderate size, located a little more than its diameter from the tip of the truncate snout, and does not enter into the profile of the head. There is apparently no adipose tissue around the eye, but somes does extend forward from the front of the eye to surround the nostrils. The two small nostrils are located much nearer to the tip of the snout than to the eye. The end of the maxillary is below the anterior border of the eye, but the angle of the gape is well before the eye. The premaxillary is not pro- tractile. The lacrimal bone is large and transparent, and covers the top of the maxillary. The teeth are uniserial in the jaws. The teeth in the upper jaw are small, conical, slightly recurved, and spaced; the teeth in the lower jaw are large, at least twice as long 4 BREVioRA No. 351 as those in the upper jaw, compressed and knifeUke with very small cusps, and very close-set. The vomer, palatines, and basibranchials appear to be toothless. The oral valves are prominent. The oper- cles are very thin; their margins are either entire or set with ex- tremely fine spinules. The striated opercle has two very weak flat spines; the angle of the preopercle is rounded but does not bulge backward. The gill-rakers are moderate, blunt, about half the length of the filaments, and bear fine teeth on their inner edges; the rakers are spaced, about 15 on the lower limb of the first arch. The pseudobranch is well developed, but there are no rudimentary rakers below it. The light yellow thymus is clearly visible. There are six branchiostegal rays. The color in alcohol is tan, darker on the back than on the sides. There is a suggestion of three brownish vertical bands on the after part of the body in some specimens. The first dorsal fin is dark, but all the other fins are whitish. The dark lining of the gill cavity shows clearly through the transparent opercles. The dark peri- toneum shows clearly through the thin abdominal wall. The in- side of the mouth is fight yellow, and the eye is grey. The skeleton in general is very light. This is particularly ap- parent in the thin transparent dermal skeleton. There is no supra- maxillary bone. The supraoccipital is but little developed. There is a wide opening between the cleithrum and the coracoid. The pelvic bones reach to the cleithrum. There is a large foramen in the scapula. The postcleithrum can be plainly seen through the body wall, and it extends to the lower margin of the body. There are 36 to 38 vertebrae, including the hypural; about 12 to 15 verte- brae appear to be precaudal. In the tail, there are two autogenous haemal spines, four hypurals, two paired uroneurals, and three epurals. The second and third hypurals are broad triangular plates much larger than any other caudal element. The first hypural bears an hypuropophysis. Three free interneurals precede the dorsal fin. The first dorsal interneural supports two spines. The two anal spines are weak. Remarks. Most species of P series are widely distributed in the tropical and sub-tropical parts of the world ocean. P. sio, how- ever, is quite restricted, and has been found only in the eastern Pacific from about 11° S (holotype) to perhaps 10° N (R. Rosen- blatt, additional Scripps specimens, in lift.). P. sio most closely resembles P. rnaculatus, an apparently antitropical species known only from sub-tropical waters in the North and South Atlantic. Within the range of Psenes sio, only P. cyanophrys is known to occur for sure. This species may be distinguished from P. sio 1970 NEW NOMEID FISH 5 by its small conical teeth that are similar in both jaws (as men- tioned above), its color pattern of fine horizontal stripes (P. sio has about three indistinct vertical bands), its greater maximum depth (43-52% SL vs. 29-41% SL in P. sio), its generally greater number of median fin rays (D 24-29 vs. 23-25, A 24-28 vs. 23-24), and its fewer vertebrae (31 v^. 36-38). As yet unrecorded from the eastern tropical Pacific but known from the western parts of that ocean are Psenes pellucidus and P. arafurensis. The former has more median finrays and verte- brae (D 27-32, A 26-31, vert. 41-42) than P. sio, the latter fewer (D 18-22, A 20-22, vert. 31). Psenes whiteleggii, from the Indian Ocean and Australia, has conical teeth in both jaws and low median finray and vertebral counts (D 17-20, A 17-18, vert. 31-32). The Atlantic P. macu- latus, the species most similar to P. sio, has slightly fewer median finrays and vertebrae (D 22-24, A 22-24, vert. 35) and more pectoral finrays (21-22 vs. 19 in P. sio) and anal spines (III vs. II). The preanal distance is 58-63% SL in P. maculatus, and 51-54% SL in P. sio. LITERATURE CITED Haedrich, Richard L. 1967. The stromateoid fishes: systematics and a classification. Bull. Mus. Comp. Zool., Harvard, 135(2): 31-139. Haedrich, Richard L., and Michael H. Horn. 1969. A key to the stromateoid fishes. Woods Hole Oceanographic Institution Ref. No. 69-70, September 1969, 46 pp. Unpublished Manuscript. (Received 2 April 1970.) 6 BREVIORA No. 351 to to (in /-i « O < . "^^ ^ "^^ - H -o < ^ a (U CU 1 1 <^ 1 1 + r4 X »o <^ a\ 00 00 rj r<^ 00 .~ ON VD \o ON r^ O -H HH k-H .-^ X II II ••.■•—■_• .~ - 00 T^ _' H-' =» Z" X " ^ + C/5 E E •^ r- r- >/-, GO U-, r 1 't 1 "^ m in m 1 o _; ^ (N m vo "'"1 <^i lO r«^ ri m o6 ^' x' ^' >r^ m „ m ri — 1/-, ^_ -^ GN I- ri /^ oo vr-, rl t I ON in >r^ r^ *"' '~' on + ^' jl ?^ CL S] S rt in NO >/-i K-; 1/^ (^ I — — ■ od ^' K>' „• _ „ "n x Qa.r-ND(^ — — rir^-^ ?<— ,oo>n 52 -^ I ""^ 53 -S _ .5 c S c J= ?? 1- >- ^ Wl r^ '^ C/l ^ ■ -o •?„ P fe o 1), o c .E ^ "E ^H - « e x: -4— t a f > S 1) c 3 ■a E 3 D- c E a (/5 c 3 O cS bS^niio^g'2c:y"'Fi B 1^ ^ -H ^ Ul 1970 NEW NOMEID FISH m U -a c c u N o 00 GO b e c .0 >>> c 0, (4-1 o u a, >. ■♦-• "o 3 00 BREVIORA Miiseimi of Cointipairsitive Zoology Cambridge. Mass. 10 July, 1970 Number 352 THE CHANARES (ARGENTINA) TRIAS5IC REPTILE FAUNA VII. THE P05TCRANIAL SKELETON OF THE TRAVERSO- DONTID MASSETOGNATHUS PASCUALI (THERAPSIDA, CYNODONTIA) Parish A. Jenkins, Jr.^ Abstract. The postcranial skeleton of Massetognathus pascuali is described from a single nearly complete individual and four disarticulated specimens; manus, pes and pubis could not be described from the available material. M. pascuali has a postcranial skeleton morphologically similar to that in Pascualgnathns polanskii (an earlier traversodontid) and in African Triassic cynodonts. The basic skeletal pattern of all these forms differs from that in Exaeretodon sp., an advanced traversodontid. The specialized, imbricating ribs of M. pascuali are intermediate in form between those of P. polanskii (which are diademodontid in form) and those of Exaere- todon sp. (which essentially have lost the cynodont specialization). Although Richard Owen made the first study of a cynodont more than a century ago, a detailed knowledge of cynodonts has accumu- lated only recently. First known only from Africa, representatives of this advanced group of mammallike reptiles have now been recovered from Asia and North and South America. The most specialized and, during the early and middle Triassic, the most abundant cynodonts were the herbivorous gomphodonts. Trans- versely broad, multicuspid postcanine molars with heavy occlusal wear (often to the point of effacing the original crown pattern) are characteristic of the group. Romer (1967) recognized three gomphodont families: Diademodontidae, Traversodontidae, and Trirhachodontidae. Massetognathus pascuali is a traversodontid and is one of two species described by Romer (1967) from the ' Department of Anatomy, Columbia University, New York, New York 10032 2 BREVIORA No. 352 Chanares Formation of Argentina. Other undoubted traversodon- tid genera include Traversodon from the Santa Maria Formation of Brasil, Exaeretodon, Proexaeretodon and Ischignathus from the Ischigualasto Formation of Argentina, and Scalenodon and Scal- enodontoides from the Manda and Molteno beds, respectively, of Africa. Pasciialgnathus polanskii from the Puesto Viejo Forma- tion of Argentina, originally classified as a diademodontid (Bona- parte, 1966a; 1966b), is now interpreted as a primitive traverso- dontid on the basis of tooth morphology (Bonaparte, 1967, and personal communication). The present account of the Massetognathus pasciiali postcranial skeleton is based upon the disarticulated remains of at least four individuals and an almost completely articulated skeleton of a fifth individual, all about the same size and preserved in a nodule several inches thick and two and one-half feet in diameter. The 1964-65 expedition of the Museo de la Plata and Museum of Comparative Zoology collected the nodule from the Chanares Formation in the Chanares-Gualo region of western Argentina. All the skeletal material is catalogued as No. 3691 in the Museum of Compara- tive Zoology. Although the available postcranial material of Massetognathus pasciiali is incomplete, there are two reasons why even a partial account is important. First, some workers regard cynodonts as ancestral to mammals (Crompton and Jenkins, 1968; Hopson and Crompton, 1969). Gomphodont cynodonts are certainly not di- rectly related to mammals, but they undoubtedly possessed a de- gree of biological organization that at least approximated that of their carnivorous-insectivorous relatives (from some form of which mammals were derived). Thus all cynodonts are relevant to evalu- ating the reptile-mammal transition. Second, most published in- vestigations of cynodonts have been restricted to cranial anatomy. Well-preserved, generically determinate postcranial skeletons are rare. Of South American genera, only Belesodon (von Huene, 1935-1942), Exaeretodon (Bonaparte, 1963a) and Pasciialgna- thus are known from relatively complete skeletons; the preservation of known Belesodon material is poor, however, and Exaeretodon appears to be morphologically aberrant in comparison to the gen- eral pattern known from other cynodonts. Postcranial material associated with Traversodon and Chiniquodon (von Huene, 1935 1942) is only fragmentary, although Romer (1969) has recently described more complete limb bones of Chiniquodon. I have re- viewed the postcranial skeletons of African cynodonts (Jenkins, in 1970 MASSETOGNATHUS POSTCRANIAL SKELETON press). Of the Russian forms very little is known; some data is available for Permocyiiodon (Konjukova, 1946). Vertebral column and ribs The number of presacral vertebrae in Massetognathus pascuali is at least 23 and is here interpreted to be about 26. A hiatus in the lumbar series of the one nearly complete, articulated vertebral column is responsible for this uncertainty. Present are 7 cervicals, 164-? dorsals, and 6 sacrals. Judging from the length of several articulated but incomplete caudal series, tail vertebrae numbered at least 17 and probably as many as 25. A count of 26 presacral vertebrae in M. pascuali compares with 26 in Leavachia duven- hagei (Broom, 1948), 27 in Thrinaxodon liorhinus (Jenkins, in press), 28 in Exaeretodon sp. (Bonaparte, 1963a), 29 in Cynog- nathus crateronotus (Seeley, 1895 ) and at least 30 in a large cyno- dont identified tentatively as Diademodon sp. (Brink, 1955; Jen- kins, in press). Two separately articulating atlas arches appear to conform to the general pattern known from African Triassic cynodonts (Jen- kins, in press), but their preservation is too poor to allow specific description or comparison. The atlas intercentrum. transversely elongate and ventrally convex, bears a posteriorly directed process or lip (1, Fig. IB). A concave articular facet on the dorsal sur- face of this lip contacts a median convex facet on the anterior aspect of the atlas centrum (f a i, Fig. 6B ) . The lateral extremities of the intercentrum each bear, on their dorsal surfaces, a concave facet that articulates with the ventral aspect of one of the two occipital condyles. The atlas centrum (6 mm long) is synostosed Figure 1. Vertebral elements in Massetognathus pascuali. A. Cervical vertebrae, probably the third and fourth, in lateral view. B. First or atlantal intercentrum in ventral view with anterior margin toward the top of the page. C, Anterior dorsal ("thoracic") vertebrae; vertebra on the right is in lateral view, on the left posterolateral view. All X 1. Abbreviations: fo, costal fovea for rib head; 1. posterior lip of the intercentrum; p. para- pophysis; t, transverse process. 4 BREVIORA No. 352 to the axis centrum (8.5 mm long), forming a conspicuously large vertebra. On the dorso-lateral aspects of the atlas centrum are two facets, one for each atlas arch half (f a a. Fig. 6B). Protruding from the median aspect of the centrum between these facets is a small process (d, Fig. 6B) homologous with the dens or odontoid process of mammals. Elsewhere (Jenkins, 1969; in press) I have proposed that the mammalian dens originated as a neomorphic process from the atlas centrum of cynodonts. The occurrence of a dens in Massetognathus pascuali is unequivocal and supports the theory that both a dens and an atlas centrum (of which the dens was formerly thought to be a vestige) occur together among cynodonts. The axis centrum of the one articulated cervical series is about 9 mm long, 1 mm longer than the centra of the following cervicals. The axial prezygapophyses have been broken off on all specimens, but from the appearance of their narrow bases^ — no thicker than the laminae from which they protrude — they are vestigial. Axial postzygapophyses of normal size bear articular facets at an angle estimated to be 25° to the horizontal. The axial spine, a flat blade with a distinctly mammalian shape, exhibits a straight dorsal mar- gin that was probably slightly convex during hfe. The robust transverse processes are directed posterolaterally. Cervical centra are amphicoelous, as are all other centra in Massetognathus pascuali, with the exception of the atlas (only the posterior aspect bears a notochordal concavity) and possibly some of the smaller caudal vertebrae. In cervical vertebrae, the rim sur- rounding the notochordal concavity on each end is swollen, and this increases the concave curvature of the sides and ventral aspect of each centrum. A bulbous parapophysis is to be found on the ventrolateral aspect of each anterior rim (p. Fig. lA). If the rib head articulated on the apex of the parapophysis, as seems most likely on the basis of the large size of the process, then this condi- tion differs from that in species of Thrinaxodon and Cynognathus, in which cervical rib heads are situated intervertebrally. A median ventral keel, similar to that in other cynodont cervicals, traverses the length of each cervical centrum. Cervical pedicles are narrow anteroposteriorly and bear short, stout transverse processes (t, Fig. lA). Cervical spines, broken off on all but two disarticulated cervicals (Fig. lA), are trans- versely slender and recurved, and taper abruptly toward the apex. Of particular interest are the zygapophyses, which provide a basis for distinction between cervical and dorsal (anterior thoracic) vertebrae. As noted above, the axial postzygapophyseal facets are 1970 MASSETOGNATHUS POSTCRANIAL SKELETON 5 oriented at an angle of about 25° to the horizontal; the orientation of posterior facets on the third through sixth cervical are more than 25° — probably about 35°. The distance between their lateral mar- gins is approximately 7 mm. The anterior articular facets of the seventh cervical of course conform in orientation and spacing to those of the foregoing series. The posterior articular facets, how- ever, appear to be oriented at about 45°. Those on the succeeding (eighth) vertebra are nearly parasagittal — i.e., within a few de- grees of vertical — and are only 3.5 mm apart. The articular facet characteristics of the eighth vertebra, continued (with gradual modification) through the dorsal series, definitely estabhsh a dif- ferent pattern. Although transitional, the seventh vertebra most closely resembles other neck vertebrae and thus may be regarded as the last cervical. Similar changes in facet orientation and spacing, together with other morphological changes, occur between the seventh and eighth vertebrae of Tlvinaxodon liorhimis and Cynognathiis craterouotiis (Jenkins, in press). The condition in Massetognathus pascuali reaffirms the fact that the "mammalian" number of seven cervical vertebrae was already established in cynodonts. With the exception of the atlas intercentrum, no other cervical intercentra have been identified. The broad grooves formed by the rims of adjacent centra are evidence that intercentra were present — as in Thrinaxodon liorhimis, for example — and the state of dis- articulation accounts for their postmortem loss. The dorsal vertebral column in cynodonts is either a relatively undifferentiated dorsal series (as in most reptiles) or two series — "thoracic" and "lumbar" (as in mammals). In certain African Triassic cynodonts, separation of thoracic and lumbar regions may be made on the basis of distinct morphological specializations of the posterior dorsal (= lumbar) ribs (Jenkins, in press). In other cynodonts (e.g., Exaeretodon sp.; Bonaparte, 1963a) the trunk cannot be divided into thoracic and lumbar regions because the posterior ribs are not specialized. Massetognathus pascuali has specialized "lumbar" ribs and clearly belongs to the first category. The exact number of thoracic and lumbar vertebrae is unknown because a complete vertebral column with ribs is not yet available. Therefore, thoracic and lumbar vertebrae will be given only a general description as anterior and posterior dorsals, respectively, although the ribs (to be discussed below) clearly give evidence of a differentiated series. The centra of anterior dorsal vertebrae are approximately 8.5 mm long; those of posterior dorsals are about 1 1 mm long (10 mm 6 BREVIORA No. 352 in the one articulated specimen). The costal foveae (fo. Fig. IC) of anterior dorsal vertebrae form a cleft for reception of the rib head, which is therefore intervertebral in position. On posterior dorsals the rib head articulates with a parapophysis (p, Fig. 2B) and is not intervertebral. Transverse processes are laminar and bowed dorsally on anterior dorsal vertebrae (t, Fig. IC) but are rodlike and round in cross-section on posterior dorsals (t, Fig. 2B). The posterior intervertebral notch is deep throughout the dorsal series, the anterior notch negligible or absent. In contrast to many cynodonts, anapophyses are lacking. Dorsal prezygapophyses are robust, extending to or slightly beyond the level of the anterior aspect of the centrum. Articular facets on the first dorsal vertebra are about 3.5 mm apart and on the ninth are 4.0 mm; an abrupt widening takes place at some point in the middle or posterior dorsal series, for the articular facets of the penultimate dorsal vertebra are about 7 mm apart. The narrowly-spaced facets of anterior dorsals are nearly vertical, whereas those wider apart on the pos- terior dorsals are oriented at angles of as much as 45°. Neural spines on anterior dorsals are narrow, but unlike cervical spines, do not taper significantly at the apex; they incline caudad at angles of about 30°. Spines on posterior dorsal vertebrae are broad an- teroposteriorly, leaving only a narrow gap between vertebrae; their inclination is only a few degrees caudad. Sacral vertebrae successively decrease in size posteriorly. The centrum of the first sacral is approximately as long as those of the posterior dorsals (about 10 mm), while the sixth and last sacral centrum is some 2 mm shorter. Massive synapophyses — repre- senting fused parapophyses and transverse processes — arise from the pedicles and from the anterior half of the side of the centrum. The zygapophyses, proportionately less robust than in the dorsal series, diminish in size on successively more posterior sacrals. In contrast to the orientation and spacing in the posterior dorsals, sacral articular facets incline nearly parasagittally and are nar- rowly spaced. At the last dorsal-first sacral articulation, the facets are 5.5 mm apart, but between the third and fourth and fourth and fifth sacrals they are only about 2 mm apart. At the fifth-sixth sacral articulation the trend is reversed; the facets are spaced 3.5 mm apart and appear to be inclined at 10° from the vertical. The trend toward less verticality of facets is continued into the caudal series. Sacral spines successively diminish in height and anteroposterior breadth. The spine on the first sacral is approxi- mately 8 mm in height and 7 mm in breadth (versus 8.5 mm height and 8 mm breadth for the last dorsal spine). On the fifth sacral i 1970 MASSETOGNATHUS POSTCRANIAL SKELETON 7 the same measurements are 5.5 mm and 4.2 mm respectively. This trend is continued into the caudal series. The apices of sacral spines tend to be oval in contrast to those of the posterior dorsals which are elongate and attentuated at each end. Caudal centra decrease in length from 6.5 mm at the first caudal to about 5 mm at the fifth. More posterior caudals preserved with MCZ 3691 are disarticulated, and their position cannot be posi- tively assigned. However, one isolated series of eleven caudals shows a decrease in centrum length from 5 mm (which is evidence that it is approximately the fifth caudal) at the first to 4 mm at the last (? fifteenth caudal). Other specimens show that at least the first five caudals bear synapophyses. Articular facets are inclined at approximately 45°, at least through the first five caudals. The width between the lateral edges of these facets decreases from 6 mm (between the last sacral and first caudal) to 4.5 mm (between the fourth and fifth caudals). The terminal caudal is unknown. Massetognathus pascuali probably possessed ribs on all presacral vertebrae, as in other cynodonts for which adequate material is known. Ribs were not found in association with the first six cervi- cal vertebrae, although the morphology of the transverse processes and parapophyses on the axis through sixth cervical is clear evi- dence of their existence. These features cannot be verified at pres- ent on the atlas. The ribs of the seventh cervical appear to have been shorter than, but otherwise similar to, those of the anterior dorsal series. Dorsal ribs in Massetognathus pascuali are of basically two types: in the anterior and middle dorsal series, a freely articulating rib of normal costal form and proportions, and in the posterior dorsal series, a fused rib with a Y-shaped distal end. Unfortu- nately, the transitional ribs between the two types are as yet unknown. The proximal ends of anterior and middle dorsal ribs are basi- cally triangular. The tuberculum and capitulum form two corners of the triangle; the shaft arises from the third (Fig. 3A). On the anterior aspect of this triangular surface is a slight crest (c. Fig. 3A) comparable to a similar feature on the anterior dorsal ribs of Cynognathus sp. (c. Fig. 3B. C). There is no other apparent fea- ture that makes this type of rib distinctive. The maximum widths of the shafts near their proximal ends are between 2.5 and 3 mm. The distal ends, as preserved, are approximately 1.5 mm thick. Ribs of this basic morphology (as opposed to the specialized posterior 8 BREVIORA No. 352 Figure 2. The posterior dorsal and sacral region in Massetoguathus pasciiali in A, dorsal and B, lateral views. X 1. Abbreviations: a il, acetabular facet of the ilium; a p, anterior process of rib shaft; b, ridge on dorsal aspect of rib shaft; f a p, articular facet on anterior process of rib shaft; gr, groove on dorsal margin of ischium; il, ilium; is, ischium; is t, ischial tuberosity; m p, medial process on base of ilium for articulation with ischium and pubis; p, parapophysis; p p, posterior process of rib shaft; t, transverse process. dorsal ribs to be described next) are associated with at least the first thirteen dorsal vertebrae. The following description of the specialized posterior dorsal ribs is based on the penultimate and last dorsal ribs preserved in articulation (Fig. 2) and in addition, a few disarticulated pieces of similar morphology. These ribs have a short shaft that bifurcates into a Y-shaped terminus with two processes (ap, pp, Figs. 2, 4B). The anterior process is broader than the posterior and bears, on the dorsal surface of its tip, a rather flat, round facet (f a p, Figs. 2A, 4B). Articulating with this facet is the end of the posterior process of the preceding rib. Presumably, the underside of the 970 MASSETOGNATHUS POSTCRANIAL SKELETON Figure 3. Proximal ends of dorsal ribs of A, Masseto^natliiis pascuali (X 2), and of B and C, Cynognatlms sp. (X V2) in anterior view. Ab- breviations: c, crest on anterior aspect of shaft (see text for details). posterior process also bears a facet. The rib as a whole projects laterad from the vertebral column and appears not to have had any ventral curvature. In lateral view (Fig. 2B) the shaft and point of bifurcation are at approximately the same level, but the anterior and posterior processes incline somewhat ventrally. A low, bony ridge (b. Figs. 2 A, 4B) runs obliquely onto the posterior process from the point of bifurcation where it is most prominent. This feature is comparable to a similar ridge on the ribs of Cynognatlms sp. (b. Fig. 4A) and other cynodonts; in the fourteenth thoracic Figure 4. Specialized posterior dorsal ("lumbar") ribs of various cynodonts. A. Cynognathus crateronotiis (British Museum of Natural History no. R. 2571), X V4. B, Massetogucithiis pascuali. XL C, Leavacliia duvenhagei (Rubidge Collection, Graaf Reinet, South Africa, no. 92), X V2. All dorsal views. Abbreviations: a p, anterior process of rib; b, ridge on dorsal aspect of rib shaft (of unknown function but probably homologous in the forms shown here); f a p, articular facet on anterior process of rib shaft; LI, L2, first and second lumbar vertebrae; p p, posterior process of rib shaft; T14, T15, fourteenth and fifteenth thoracic vertebrae. 10 BREVIORA No. 352 rib of C (T14, Fig. 4A), the ridge is merely a linear elevation on the flat costal plate. On successive ribs, however, the ridge becomes more prominent until, in the lumbar ribs (LI, L2, Fig. 4A), it reflects forward to contact the preceding rib plate (see Jenkins, in press). No such reflection is evident in Massetognathus pascuali, but the ridge morphology and general pattern of the process are nevertheless similar to that of the fourteenth and fifteenth thoracic rib plates of C Specialized, imbricating ribs are common but not universal among cynodonts. Known members of the earliest cynodont fam- ily, the procynosuchids, apparently did not possess this specializa- tion (see discussion below, however, for a possible exception). Galesaurids, typified by the well known Thrinaxodon liorhinus (Jenkins, in press), developed costal expansions on all presacral ribs. Members of three other families, e.g., Cynogmitlms craterono- tus (Cynognathidae; Seeley, 1895), Diadeinodon sp. (Diademo- dontidae; Jenkins, in press) and Cricodon metabolus (Triracho- dontidae; Crompton, 1955), possessed imbricating ribs only in the posterior dorsal region. The ribs in chiniquodontids (von Huene, 1935-1942) are as yet unknown. There remains only the Traver- sodontidae, which Bonaparte (1963b) characterized, on the basis of species of Exaeretodon and supposedly Traversodon, as lacking synostosed ribs with overlapping processes. For this and other reasons, Bonaparte interpreted traversodontids as probably having arisen from procynosuchids along a lineage separate from that an- cestral to all other cynodont families (whose members possess rib specializations). However, Massetognathus pascuali unquestion- ably possesses synostosed lumbar ribs with details comparable to the Cynognathus-Diademodon pattern. Pascualgnathus polanskii, now classified as a traversodontid (Bonaparte, 1967), has lumbar ribs that are unquestionably diademodontid in pattern. Further- more, von Huene (1935-1942; 137-140) described expanded ribs ("Facherrippen") synostosed to the lumbar vertebrae in Traverso- don stahleckeri. Crompton (1955) presented circumstantial evi- dence that the traversodont Scalenodon from the African Manda beds also possessed the expanded rib specialization. Yet Bona- parte (1963a) amply demonstrated that at least one traversodont, Exaeretodon sp., did not possess such specialization. Presacral ribs in Exaeretodon sp. are morphologically uniform and are more or less freely articulating (although the more "solid" attachment of the last three dorsal ribs, as described by Bonaparte, possibly rep- resents a vestige of a less mobile articulation typical of expanded ribs ) . In view of this unexpected association at the family level of 1970 MASSETOGNATHUS POSTCRANIAL SKELETON 11 forms possessing and forms lacking rib specializations, the taxo- nomic significance of this character should be reassessed. Further comment is reserved for the discussion below. Sacral ribs in Massetognathiis pascuali have an essentially con- fluent capitulum and tuberculum, a short shaft and an expanded distal end for articulation with the iliac blade. Proximally each rib is synostosed to its corresponding vertebra. If I may judge from the disarticulated condition of every known sacroiliac joint, liga- ments and cartilage must have been chiefly responsible for binding the ilium and sacral ribs. The concave distal ends of the sacral ribs conform to the gently convex internal surface of the ilium, but they do not appear to form any osseous interdigitation by which sacroiliac joints are commonly reinforced. Viewed from above, the distal end of the first sacral rib (Si, left side. Fig. 2A) is Y-shaped with processes directed anterolaterad and posterolaterad. Articu- lating with the dorsal surface of the anterolateral process is the posterior process of the last dorsal (lumbar) rib. The second sacral rib has the largest distal expansion; irregularly shaped and widest anteriorly, the expansion is 7 mm long in one well preserved speci- men. The third, fourth and fifth ribs bear more or less symmetrical distal expansions that are successively smaller caudally. The first four sacral rib shafts are oriented more or less laterally, the fifth slightly anterolaterally. The sixth and last sacral rib bears a shaft that is directed posterolaterally and a bifurcated, Y-shaped termi- nus resembling that of posterior dorsal ribs. Its iliac articular sur- face is narrow and strap shaped; the rib and vertebra could well be interpreted as the first caudal were it not for its position opposite the posterior tip of the ihac blade (S6, Fig. 2A). While it is appar- ent that the width between the distal ends of this pair of ribs (20 mm) is much less than that of the fourth and fifth sacrals (27 mm), the intervening gap could well have been completed by ligaments. This interpretation of the sixth sacral vertebra and ribs may be open to future modification, but on present evidence ap- pears to be the most probable. Only the first three caudal ribs are preserved. All are synostosed, their shafts directly posterolaterally. The first is only 8 mm long (Cdl, Fig. 2A), with two blunt processes on its terminus resembling a stunted version of the last sacral rib. As far as can be deter- mined, terminal processes were not developed on the second and third caudal ribs, which are 5 mm or less in length. Ribs on suc- ceeding vertebrae must have been very small and probably did not occur in the posterior caudal series. 12 BREVIORA No. 352 Shoulder Girdle Available interclavicles of Massetognatlius pascuali are incom- plete, although there is sufficient material to conclude that the morphology is very similar to that in Thrinaxodon liorhinus and different from that in Exaeretodon sp. In outline the interclavicle is cruciate with an elongate posterior ramus (pr, Fig. 5A). The entire bone, although basically a flat plate, is bowed ventrally from front to back. Two ridges, one longitudinal, the other transverse, divide the ventral surface of the interclavicle into quadrants. The two anterior quadrants (c c. Fig. 5A) are shallow concavities for reception of the proximal ends of the clavicles. The ridges are #^f prox Figure 5. A, The interclavicle and B, the right clavicle of Massetog- nathus pascuali, both in ventral view. X 1. Abbreviations: ca, concavity for acromion; c c, concavity for proximal end of the clavicle; dist, distal end of clavicle; f, ventral flange on distal end of clavicle; pr. posterior ramus of interclavicle; prox, proximal end of clavicle. most prominent at their intersection. With the exception of the posterior part of the longitudinal ridge, which gradually fades out, the ridges become more salient toward the margins. There is no evidence that the longitudinal ridge was a deep keel as in Exaereto- don sp. (Bonaparte, 1963a). The posterior ramus in M. pascuali is similar in length and form to that in galesaurids, and is unlike the very short ramus of Exaeretodon sp. The Massetognathus pascuali clavicle is robust. The proximal two-thirds are more or less straight (Fig. 5B), the distal third curving sharply posterodorsally. The broad, flat plate on the proxi- mal end articulates with the previously described concavity (c c. Fig. 5A) in the interclavicle. Along the ventral aspect of the distal 1970 MASSETOGNATHUS POSTCRANIAL SKELETON 13 third runs a flange (f, Fig. 5B) similar to that noted in African Triassic cynodonts (Jenkins, in press) and Exaeretodon sp. (Bona- parte, 1963a). This flange continues to the distal end where it contributes to the formation of a concavity (ca, Fig. 5B) for re- ception of the acromion. The clavicle is essentially identical to that in galesaurids. The scapula, coracoid, and procoracoid in Massetognathiis pas- ciicili are firmly synostosed, although the joints can readily be dis- tinguished (Fig. 6A). The scapular blade, elongate and narrow, bears a distinct concavity on its lateral surface — a fossa presumably for the supracoracoideus muscle, the infraspinatus homologue of mammals. The anterior margin of the blade is reflected sharply laterally, the posterior margin somewhat less so. An acromion process as such is not preserved on any of the scapulae; this ab- sence may be due to postmortem damage to a delicate process or to the fact that the clavicular concavity simply fitted to the convex edge of the anterior scapular base (ac, Fig. 6A). The scapular half of the glenoid is a hemicircular and shghtly convex facet that faces posteroventrally and somewhat laterally. The coracoid is basically triangular in lateral view (co. Fig. 6A). The posterior end forms an elongate, attenuated process terminated by a tubercle for the origin of the coracoid head of the triceps (co tr, Fig. 6A ) . The process is morphologically similar to the same feature in Pascualgnathus pokmskii and in African Triassic cyno- donts, but differs in form from that in Exaeretodon sp. as inter- preted by Bonaparte ( 1963a). Between the glenoid and the triceps tubercle, the superior margin of the coracoid is about 2 mm wide and is slightly concave from front to back. The slightly convex inferior margin is, in contrast, extremely thin bone, and, as a con- sequence, is invariably damaged postmortem. A saddle-shaped facet, concave dorsoventrally and convex transversely, constitutes the coracoid half of the glenoid. The procoracoid (pr, Fig. 6A) appears to be an irregularly shaped, flat plate, but in no available specimen are the free mar- gins complete. There are so many basic similarities in the scapu- locoracoid of Massetognathiis pascuali and African cynodonts that a complete procoracoid of the former would probably have the same oval shape characteristic of the latter. A crescentic depres- sion on the lateral aspect along the inferior margin may represent the biceps origin. Above lies the round procoracoid foramen (f pr. Fig. 6A) sculptured in a ventrolateral direction to facilitate passage of its nerve and blood vessels. The evidence as to whether the procoracoid participated in the glenoid is equivocal. I 14 BREVIORA No. 352 believe that the procoracoid probably supported articular cartilage at the very anterior extremity of the glenoid (see left glenoid, Fig. 6A), but there is no certainty of the procoracoid contribu- ting to the shoulder joint as there is for some African Triassic cvnodonts. 970 MASSETOGNATHUS POSTCRANIAL SKELETON 15 Figure 6. Elements of the postcranial skeleton of Massetogiuithiis pasciiali, drawn as preserved in situ. A, Incomplete left shoulder girdle and forelimbs seen from the left side. B. Lateral view of axis. C, Dorsal view of left humerus. D, Medial view of left radius and ulna. All X 1. Ab- breviations: ac, area of clavicular articulation (distinct acromion process not developed); cl. clavicle; co, coracoid; co tr, tubercle for coracoid head of triceps; cp. capitellum; d, dens; d f. distal flange on radius; dp, delto- pectoral flange; f a a, atlas centrum facet for atlas arch; f a i, atlas centrum facet for atlas intercentrum; f ec, ectepicondylar foramen; f en, entepicondylar foramen; f pr, procoracoid foramen; g, groove possibly representing teres major insertion or the origin of one of the humeral triceps heads; h, humeral head; hu, humerus; 1, ridge possibly representing insertion of the teres minor; 1 t, lesser tuberosity; p f, proximal flange on radius; pr. procoracoid; ra, radius; s, scapula; ul, ulna; ul f, ulnar flange. 16 BREVIORA No. 352 Forelimb Principal characteristics of the humerus of Massetognathus pas- ciuili are the relatively broad expansion of the proximal and distal ends, and the large deltopectoral flange. The proximal end of the shaft is bowed dorsally and the head is oriented to a more dorsal position. The well-rounded head (h, Fig. 6C) possesses greatest curvature along its dorsoventral axis. As preserved, the articular surface is confluent medially whh the lesser tuberosity (1 t, Fig. 6C) and laterally with the proximal margin of the deltopectoral flange. The greater tuberosity presumably arose in the mammalian lineage between the head and the proximal margin of the delto- pectoral crest, but in M. pasciiali there is no evidence of a distinct tubercle. The greatest width of the proximal end, from the lesser tuberosity to the region of the presumptive greater tuberosity, measures approximately one-third the total length of the humerus. The broad deltopectoral flange is slightly more than half the total length of the humerus. The free edge of the flange thickens and everts (laterally) at its proximal and distal extremities, but along the middle part is rather thin and flat. From the region of the pre- sumptive greater tuberosity a low, bony ridge runs obliquely across the flange toward the shaft (1, Fig. 6C). An identical ridge on the humeri of certain African Triassic cynodonts has been interpreted as possibly representing the insertion of a teres minor (Jenkins, in press). On the posterodorsal aspect of the shaft is a groove pos- sibly representing the insertion of the teres major or the origin of one of the humeral triceps heads (g. Fig. 6C); Bonaparte ( 1966b) interpreted a rugosity at this site in Pascualgnathus polcmskii as the origin of the medial triceps head. The distal end of the humerus is triangular in dorsal view, its maximum breadth being approximately 40 per cent of the humeral length. Arising from the robust ectepicondylar region, a thin supra- condylar flange runs proximally as well as somewhat dorsally. The flange, pierced in its proximal half by a small ectepicondylar fora- men (f ec, Fig. 6A, C), becomes a low crest at the middle of the shaft and is continuous with the ridge (1, Fig. 6C) described above. A stout bar of bone arising from the entepicondylar region encloses an elongate, oval entepicondylar foramen (f en. Fig. 6A, C). The capitellum (cp, Fig. 6A) is bulbous and contributes to the thick- ness of the ectepicondylar region. The trochlea immediately adjacent is a broad, shallow groove; the principal axis of this groove is dorsoventral, as expected, but it is also slightly oblique — the dorsal part being more laterally situated than the 1970 MASSETOGNATHUS POSTCRANIAL SKELETON 17 ventral part. Morphologically, the humerus of M. pasciudi is essen- tially identical to that in Pasciuil^naihus polcmskii and galesaurids; apparent differences with galesaurids, e.g., the greater roundness of the head and capitellum, are due to the better ossification in M. pascuali and P. polcmskii. As Bonaparte (1963a) noted, the humerus of Exaeretodon sp. is more similar to the dicynodont or gorgonopsid pattern than to that typical of galesaurids, and thus stands in contrast to the conventional cynodont humerus of M. pascuali. The radius has a slight sigmoidal curvature (which facilitates its crossing over the anterior aspect of the ulna) and expanded proxi- mal and distal ends. The nearly circular proximal articular facet forms a shallow concavity, in which the greatest curvature is an- teroposterior (as is its reciprocal surface on the capitellum). On the posteromedial aspect of the proximal end is an excrescence that bears a facet (f u. Fig. 6D) apparently for articulation with the ulna. From this excrescence, a distinct flange (p f. Fig. 6D) runs distally to about the midpoint of the shaft. On better preserved material of African Triassic cynodonts, I interpreted a similar flange as possibly being associated with the biceps insertion and the radio-ulnar interosseous ligament (Jenkins, in press). Bona- parte (1963a) interpreted a similar feature in Exaeretodon sp. as marking the position of the interosseous ligament. The distal end of the radius expands gradually to the distal articular facet which is oval (long axis transverse) and shallowly concave. A distal flange ( d f. Fig. 6D ) arises near the midpoint of the shaft essen- tially as a continuation of the attenuating proximal flange described above. Beginning on the posterior aspect of the shaft, the distal flange takes a spiral course toward the lateral aspect as it enlarges distally. Its position is suggestive of the attachment of an inter- muscular septum separating flexor and extensor muscle groups. The ulna, like the radius, is sigmoidally shaped but is expanded only at its proximal end (Fig. 6D). The relatively shallow semi- lunar notch represents the typical cynodont condition — basically oval in outline, but with a rather straight medial margin and a nearly hemicircular lateral margin. An olecranon process is not present or at least was not ossified; the proximal end of the ulna, where such a process would be developed, is broad and rugose. The transversely narrow shaft of the ulna bears on its anteromedial aspect a flange (ul f. Fig. 6D) that extends from the semilunar notch to the distal articular facet. In all probability this flange rep- resents the ulnar attachment of the interosseous ligament. The lateral surface of the shaft reveals one large, spoon-shaped fossa proximally, and on the medial surface two fossae, one proximal, the other distal. All three fossae are wefl represented in African 18 BREVIORA No. 352 Triassic cynodonts, and I have proposed (Jenkins, in press) that they represent origins of various manual flexor and extensor mus- cles. The distal articular facet, convex from front to back, is broad anteriorly and narrow posteriorly and thus is triangular in outline. Both the radius and ulna of M. pascuali, as far as available mate- rial permits comparison, are extremely similar to their counterparts among galesaurids and in Pascualgnathus polanskii. Although definite similarities exist with the antebrachial elements of Exaere- todon sp., the essential identity of the M. /7<;/.vcwa//-galesaurid pat- tern is incontrovertible. Only an incomplete and disarticulated series of seven or eight carpals of Massetognathus pascuali is known, but these are so poorly preserved, and good comparative material is so scanty, that no constructive observations on the manus can be made at this time. Pelvis A complete pelvis of Massetognathus pascuali is not yet avail- able, although enough is known of the ilium and ischium for pre- liminary description and comparison with other forms. The ilium bears an elongate, vertical blade, spatulate in front and lanceolate behind (il. Fig. 2). The lateral aspect of the blade is concave, especially anteriorly. The shape of blade, as well as the relative proportions of the pre- and postacetabular regions, is most similar to that in Pascualgnathus polanskii and is comparable to that of galesaurids, cynognathids, and diademodontids; Exaeretodon sp., on the other hand, has an iliac blade quite unlike the foregoing (Bonaparte, 1963a). The base of the iliac blade in M. pascuali is constricted into a short neck, below which are medial and lateral processes. The medial process (m p. Fig. 2B) bears two articular surfaces — one each for the pubis and ischium — which intersect at an angle of about 150°. The lateral process bears a nearly circular, concave facet that represents the iliac contribution to the acetabu- lum (a il. Fig. 2B). The facet is oriented principally in a postero- ventral direction but with a slight lateral eversion. Only the dorsal half of the ischium is available for examination (is. Fig. 2). The concave acetabular surface faces anterolaterally and is oriented essentially vertically. The postacetabular part of the ischium constitutes a broad plate that ventrally meets its counter- part of the opposite side. A longitudinal groove (gr. Fig. 2) on the dorsal margin of this plate terminates posteriorly at an ischial tuberosity (is t. Fig. 2). Available pubes have been extensively damaged postmortem. 970 MASSETOGNATHUS POSTCRANIAL SKELETON 19 The pubic contribution to the acetabulum is considerably smaller than that of the ischium. Neither the obturator fenestra nor the ventral aspect of the pelvic basin is preserved. The bony acetabulum is relatively shallow — a little more than 5 mm deep. The continuous, sharp rim around the acetabulum de- fines a more circular socket than that known in African Triassic genera. In all other details, the pelvis of Massetognathus pasciiali appears to be morphologically similar to that in Pascualgnathus polanskii, galesaurids, and even larger African Triassic forms. Hindlimb The femur (Figs. 7, 8A) is a moderately slender bone, except for the expanded proximal end, which bears robust trochanters. The femoral head, bulbous and almost hemispherical as in mam- mals, is reflected medially but also somewhat dorsally by virtue of the dorsal bowing of the proximal end of the shaft (Fig. 8 A ) . The protuberant trochanter major (tr mj, Figs. 7, 8A) measures about 5 mm in thickness. A pear-shaped area of smooth bone on its apex may represent the principal site of muscle attachment or of a sub- tendinous bursa. The bone surface immediately adjacent to the apex is rugose. The trochanter minor (tr mn. Figs. 7, 8A) forms an elongate flange that arises abruptly near the intertrochanteric tr mj tr mn tr mn tr mj Figure 7. Reconstruction of a left femur in Massetognathus pasciiali in A, ventral, and B, dorsal views. X 1. Abbreviations: i f. intertrochanteric fossa; tr mj. trochanter major: tr mn. trochanter minor. 20 BREVIORA No. 352 fossa (i f. Fig. 7) and gradually disappears slightly distal to the shaft's midpoint. In cross-section, the middle of the shaft is essen- tially rectangular; its thickness from extensor (dorsal) to flexor (ventral) surfaces is about 6 mm, from the medial to lateral sur- faces 4.5 mm. The distal end of the femur expands gradually but asymmetrically, the lateral condyle being broader and farther offset from the femoral axis than the medial. The medial condyle pro- jects more ventrally than does the lateral condyle. The fibula ap- pears to have articulated on the lateral epicondylar region where a shallow groove (f f, Fig. 8A) occurs. The femur of Massetog- nathus pascuoli is morphologically comparable to the femora of Pascualgnathus pokmskii and even the larger African genera in which ossification of the extremities was well developed (Jenkins, in press). In smaller forms, such as galesaurids, the femur appears to be diiTerent because of the lack of ossification of the extremities and trochanters. The tibia is a transversely slender bone, bowed somewhat an- teriorly. The two proximal articular facets, oval in outline and shallowly concave, are separated by a low, median ridge. The B Figure 8. Elements of the hindlimb of Massetognathiis pascuoli. drawn as preserved in situ. A. Right femur, tibia and fibula in lateral view. B. Left tibia in medial view. All X 1. Abbreviations: f f. facet for articula- tion with fibula; fl, fibular flange for femoral articulation; f t, lateral tibial fossa of uncertain significance; g t, groove on medial aspect of tibia; 1 t. lineation on lateral aspect of tibia; tr mj, trochanter major; tr mn. trochanter minor. 1970 MASSETOGNATHUS POSTCRANIAL SKELETON 21 lateral margin of the lateral facet is thickened and protuberant, and it appears likely that part of the proximal fibula articulated here. On the proximal end of the lateral aspect of the shaft is a deep fossa (f t, Fig. 8A) of uncertain significance; from the posterior margin of this fossa a faint lineation (1 t. Fig. 8A) runs obliquely across the shaft to merge with the narrow anterior margin of the shaft. Also of uncertain significance is a slightly curved groove (g t. Fig. 8B) along the middle of the shaft's medial aspect. The distal extremity of the tibia has a marked lateral expansion that sufficiently widens at the terminus to accommodate a broad, slightly convex facet for the astragalus. The fibula, a very slender-shafted bone with enlarged extremi- ties, is bowed laterally. The shaft appears to have been wider transversely than anteroposteriorly. On the one complete speci- men, a shallow groove running the length of the shaft along its medial aspect can be detected. A flange on the proximal end (fl. Fig. 8A) probably contacted a groove on the lateral epicon- dylar region of the femur; the remainder of the proximal fibula articulated with the tibia. The distal end, like the proximal, broad- ens anteroposteriorly and is inflected somewhat medially. All features of both tibia and fibula found in MassetognatJius pascuali are duplicated in galesaurids and even in the larger African Triassic genera. The tibia and fibula of Exaeretodon sp., by contrast, de- part from the uniform pattern of other cynodonts by being pro- portionally more massive. Nothing can be described of the pes of Massetognathus pascuali from the available material. Discussion Massetognathus pascuali, about 50 cm in length from head to tail, was a cynodont of relatively slender build (Fig. 9). The short- ness of the limbs relative to the approximated trunk length gives the body a "low-slung" appearance. The head seems dispropor- tionately large for the body, but a relatively massive head is a common cynodont characteristic. The postcranial skeleton of Massetognathus pascuali is basically like that in galesaurids, diademodontids, cynognathids and Pas- cualgnathus polanskii. Limited morphological diversity appears to be the rule for the postcranial skeletons of Triassic cynodonts. Exaeretodon, however, is an exception; the postcranial skeleton in this genus differs in major details from the pattern characteristic of other Triassic cynodonts. Bonaparte (1963a) recognized post- cranial specializations in his original description and implied that 22 BREVIORA No. 352 "u ^ "5 C/D c ;/! >
  • > O TO TO CD-.- -r- in CO CT> CM TO .— n r~~ ro ^o en .-1 cr> in •=!■ lO in CO 1. '1— ■a. +J 10 0) «t- M • n O TO •1- JD > O ■O TO ■o s- •.- CM 00 rv. O CNO jC in r^ LD CO o o 00 CO •a- o en en TO (_) E .^ ■-< .>;r .-c . CM 0\ ..-HO • ro a\ • CM CM r^ en o OJ 1 -> E o 1 r-~ O 1 O 0) E tn CM ■^ CM • t— 1 f— 1 • CO CO 00 CO to 4-> TO • Uf) • LD .-1 1 CM CO 1 CM f-H f-H >: 01 E CO 1 CM m 1 01 in o CO CM f-H f-H S- TO TO "* "=1- rt CM in Q. 0) o n ( o in 1 o CO CM +-' E cri 1 ^ ^ 1 o in r~. CM CO CM CM f-H f-H C TO «*• CSJ .-H ^ 'S- .-H r-H CM CM CO LlJ OO ^ ro 1— ( f-H O f— ) in in f-H S- 01 Ol O «3- in CNj CM CO O CM o s- .— O "3- 00 —1 . 1— « . f-H • CM • CM 4- •1- Q. • n ^ • 00 >a- r^ 1 CO CM 1 CM t~^ 1 ^ in 1 «a- CM Ln O) +-> E r-l 1 CTi 1 CM n CM O T-H f-H rH 01 OD r^ U C TO ro en CM CO t— ( r-H r-l CM CM O CM O •r— LU OO CM Cvl f-H f-H CM CM C U1 •O CO -O 0) p^ 1 — in lO "to i. 00 e: o _j > oo o QJ -f- 1— .C O. Q. E i- TO O in B. • •D ^^ E >— 73 TO C 3 X U OJ OJ 10 Ci Z3 O c • 10 TO 'r— > •^ (/) 4-> > W S. o; C >*- o. o -I-' > OJ s- (— .c ■a- Q. o ■4-> • Q OJ OJ N •r <:|o) j: cl +J E •I- TO 3 v> o O 1- OJ U C i- OJ TO 3 > O •r- _J OJ ■5 o) en ll- c O TO s- c o - l/l 00 •I- s- S- OJ TO J3 i-i . o E q: <-j o OJ iS> • ^ i- rH "O O OJ S CO .— <; -Q ••« ci3 TO C (— TO I. O) O E •*- 1970 NOTES ON AMEIVA 5 Ameiva undulata miadis Barbour and Loveridge Ameiva f estiva miadis Barbour and Loveridge, Bull. Mus. Comp. Zool., 69: 141-142, 1929 (Holotype: MCZ 26970. Type Locality: "Great Corn Island" [=Isla del Maiz Grande], Depto. Zelaya. Nicaragua. Collector: James L. Peters). Ameiva undulata miadis, Dunn, Proc. Acad. Nat. Sci. Philadelphia, 92: 115, 1940. Villa, Rev. Biol. Trop., 15: 119, 1968. Diagnosis. Ameiva undulata miadis (Fig. IB) can be dis- tinguished from A. f estiva and other subspecies of A. undulata on the basis of its unique color pattern: No narrow, light-colored vertebral stripe; lateral pattern of blue (males) or gray-brown (females) bars on a black background extending from mid- dorsal stripe to enlarged ventral scutes; broad middorsal stripe from occiput onto tail, much disrupted by transverse black blotches that are often continuous with the black background color of the flanks; sexual dimorphism in color, but not pattern, in adults. Range. Known only from Islas del Maiz, Depto. Zelaya, Nica- ragua. Remarks. With the exceptions noted above, Ameiva undulata miadis is similar to other subspecies of undulata that have barred flanks (see Echternacht, 1970). Dunn (1940) and Villa (1968) have briefly alluded to the relationship of miadis with undulata Barbour and Loveridge (1929) considered Ameiva festiva miadis ( ^A. undulata miadis) to be intermediate between A. festiva and A. ruthveni ( =A. leptophrys) . Color pattern alone distinguishes miadis from leptophrys, but in addition, the peri- pheral gular scales of undulata are like those described for niceforoi above and differ from those of leptophrys in the same manner. Also, in leptophrys the postnasals are not in contact with the pre- frontal scales, whereas in undulata these scales are in contact. The nearest known mainland populations of Ameiva festiva are at Bluefields, Depto. Zelaya, Nicaragua, about 85 kilometers west of Isla del Maiz Grande. Ameiva festiva from Bluefields are typical of the species in having a well-defined vertebral stripe in ah but large adults and in having little barring on the flanks. Nearest known populations of leptophrys are in the southeastern lowlands of Costa Rica, and the nearest populations of undulata are in northeastern Honduras, western Nicaragua, and northwest- ern Costa Rica. Four specimens of undulata (ANSP 15438- 15441) from Huaunta Haulover, Depto. Zelaya, Nicaragua, about BREVIORA No. 354 150 kilometers north-northwest of Isla del Maiz Grande, may in- dicate an isolated population there. In color pattern, the undulata from Islas del Maiz most closely resemble those from near Cata- camas, Depto. Olancho, Honduras. In these Honduranian Ameiva, the middorsal stripe is heavily blotched, but the blotches are never continuous onto the flanks. Island populations of Middle American Ameiva (with the ex- ception of qiiadnlineata) tend to have a relatively high GAB count when compared to mainland samples of the same species (see Echternacht, 1970). Although the mean for this character does not differ statistically from those of certain mainland samples examined, it is the highest recorded for the species {A. undulata: X = 149.5, standard deviation of the mean = 16.5, range = 108-202, N = 918. A. u. miadis: x = 186.3 ± 9.3, range = 165-202, N = 42). Nothing has been reported of the ecology of Ameiva undulata miadis. I was able to make some observations on June 4-8, 1966. The lizard is very common on Isla del Maiz Grande and I found it particularly abundant in clearings within the coconut groves that cover the island (Fig. 2). It was also common in the shrub and 'i.. -^ *t,. ., ,.^5^-s. Figure 2. Habitat of Ameiva undulata miadis on Isla del Maiz Grande, Depto. Zelaya, Nicaragua. June 1966. 1970 NOTES ON AMEIVA 7 grassy vegetation along the airstrip and in clearings around build- ings. None were seen on open beaches. Individuals often were encountered basking or foraging on or around fallen palm fronds. Activity extended from dawn until dusk, but more lizards were seen early in the day than in late afternoon. The lizards were not active in rain. Hatchlings with clearly dehneated umbilical scars were observed. The chin, gular region, and anteriormost part of the chest of many males was bright reddish orange. In others, this area was pale blue. Such dichromatism is widespread among species of Ameiva and may be associated with reproductive con- dition. Evidence is lacking on this point. English-speaking resi- dents of the island refer to the Ameiva simply as "lizard," al- though they have specific names for Iguana iguana and Cteno- saura similis, both of which are common. Ameiva chaitzami Stuart Ameiva chaitzami Stuart, Proc. Biol. Soc. Washington, 55: 143, 1942 (Holotype: UMMZ 90638. Type Locality: "Along Cahabon-Languin trail about 2 km north of Finca Canihor . . . , Alta Verapaz, Guatemala." Col- lector: L. C. Stuart). Ameiva imdulata thomasi Smith and Laufe, Univ. Kansas Sci. Bull., 31: 47-50, pi. lA, 1946 (Holotype: FMNH 100006. Type Locality: "La Liber- tad, Chiapas, near Rio Cuilco where it crosses the Guatemalan border," Mexico. Collector: Henry D. Thomas). New synonymy. Diagnosis. Ameiva chaitzami (Fig. IC) can be distinguished from its congeners by the following combination of characters: Small size (maximum observed SVL 85 mm for males, 75 mm for females); central gular scales enlarged, in longitudinal series; gradual reduction in size of scales radiating outward from the central gulars; prefrontals in contact with postnasals; three parietal scales (four if median parietal divided); transverse row of abrupt- ly enlarged mesoptychial scales; usually eight transverse rows of ventral scutes at midbody; moderately narrow middorsal stripe (mean width in terms of granules 37.1, N = 61); no narrow, well-defined vertebral stripe; often a dark secondary stripe medial to the dorsolateral light stripes; dorsolateral blotches of adult males fused to the dorsolateral light stripe so that the latter has a well-defined dorsal border, an irregular ventral border. The latter character alone will distinguish chaitzami from undulata with certainty. 8 BREVIORA No. 354 Range. Valleys of the upper tributaries of the Rio Grijalva in Chiapas, Mexico, and west-central Guatemala, from the vicin- ity of Finca Canihor, Depto. Alta Verapaz, Guatemala, and near Poptiin, Depto. El Peten, Guatemala. Remarks. Stuart (1942) diagnosed Ameiva chaitzami as "An Ameiva almost identical with Ameiva nndulata stuarti Smith from which it may readily be distinguished by the fact that the median parietal is divided longitudinally to produce four instead of three parietals." The posterior scales on the dorsal surface of the head is an unstable character among Middle American species of Ameiva, and diagnoses based on scales in the area are unreliable (Echternacht, 1970). Considering nndulata only, I noted the median parietal to be divided or semidivided in 22.4 per cent of 1043 specimens obtained from throughout the range of the species. This is probably a conservative estimate of the frequency of oc- currence of this condition, because the information was not noted early in the study. There seem to be no geographic trends as- sociated with the condition of the median parietal. At least 38 of 45 samples contained individuals with divided or semidivided median parietals. Samples having a high frequency of occurrence of division were as follows: near Chinandega, Depto. Chinandega, Nicaragua (55 per cent, N = 31); Piedras Negras, Depto. El Peten, Guatemala (72 per cent, N = 32); Canihor, Depto. Alta Verapaz, Guatemala (50 per cent, N = 30); near Panajachel, Depto. Sololii, Guatemala (48 per cent, N = 29); Sabana de San Quintin, Chiapas, Mexico (100 per cent, N = 10); near Las Tazas and Florida, Chiapas, Mexico (90 per cent, N = 30). Three samples of chaitzami yielded the following frequencies: Comitan, Chiapas, Mexico, 3 per cent (N = 30); near San An- tonio Huista, Depto. Huehuetenango, Guatemala, 10 per cent (N = 31 ); near Finca Canihor, Depto. Alta Verapaz, Guatemala, and near Poptun, Depto. El Peten, Guatemala, 89 per cent (N = 9). Six of the nine in the last sample constitute the type series. At the type locality and at Poptun, chaitzami is sympatric with A . imdidata hartwegi Smith, a large subspecies quite distinct in color pattern and scutellation from chaitzami. The samples from Canihor and Piedras Negras, Guatemala, and from Sabana de San Quintin, Chiapas, are hartwegi. Elsewhere within its range, samples of hartwegi have from 13 to 30 per cent of individuals with divided or semidivided median parietals. It is noteworthy that the holo- type of hartwegi (FMNH 108600), obtained across the Rio Usu- macinta from Piedras Negras, has a divided median parietal. 1970 NOTES ON AMEIVA 9 Smith and Laufe (1946) discussed the evolution of Aineiva undidata and recognized several new subspecies, but they ap- parently did not examine specimens of chaitzami. The description of A. u. thomasi (Smith and Laufe, 1946) agrees with that of Stuart (1942) for chaitzami in most respects. I have examined the type specimens of both and consider them to be conspecific. The samples cited above from Comitan and near San Antonio Huista are from within the range of thomasi as described by Smith and Laufe (1946). LITERATURE CITED Dunn, E. R. 1940. New and noteworthy herpetological material from Panama. Proc. Acad. Nat. Sci. Philadelphia, 92: 105-122. . 1943. A new race of Ameiva f estiva from Colombia. Notu- lae Naturae no. 126, 2 pp. EcHTERNACHT, A. C. 1970. A review of Middle American lizards of the genus Ameiva (Teiidae) with emphasis on geographic variation. Un- published Ph.D. Dissertation, University of Kansas, Lawrence, Kansas. Medem, F. 1969 (1968). El desarrollo de la Herpetologia en Colombia. Rev. Acad. Colombiana Cienc. Exactas Fis. Natur., 13(50): 149-199. Barbour, T., and A. Loveridge. 1929. Vertebrates from the Corn Is- lands. Reptiles and Amphibians. Bull. Mus. Comp. Zool., 69: 138-146. Stuart, L. C. 1942. Comments on the iindiilata group of Ameiva (Sauria). Proc. Biol. Soc. Washington. 55: 143-150. Smith, H. M., and L. E. Laufe. 1946. A summary of Mexican lizards of the genus Ameiva. Univ. Kansas Sci. Bull., 31(2): 7-73. Villa, J. D. 1968. A new colubrid snake from the Corn Islands, Nic- aragua. Rev. Biol. Trop., 15: 117-121. (Received 13 June 1970.) BREVIORA Miasemim of Coimpsirsitive Zoology Cambridge, Mass. 30 November, 1970 Number 355 GENERIC RELATIONS AND SPECIATION PATTERNS IN THE CARACARAS (AVES: FALCONIDAE) Francois Vuilleumier^ Abstract. The caracaras are a group of American Falconidae occurring from temperate and subtropical North America southward to extreme south- ern South America and the Falkland Islands. The taxa of caracaras appear to be closely interrelated. It is suggested that they be classified in two genera: Daptrius (forest caracaras; two sympatric species), and Polybonis (nonforest caracaras; two species-groups: the chimachima and planciis species-groups, with two and three species, respectively). Former classifica- tion advocated the use of four genera. Speciation is long completed in Daptrius and in the Polybonis chimachima species-group. In the Polybonis planciis species-group, however, several phenomena are evidence of active species formation. Geographical isolates exist that are morphologically differentiated enough to be considered borderline cases between species and subspecies. Some of the cases of geographical isolation in the caracaras can be related to climatic and vegetational changes following glacial events of the Pleistocene. INTRODUCTION This paper constitutes the fourth of a series stemming from studies on speciation in Andean birds (see Vuilleumier, 1968, 1969, 1970). I shall discuss, first, the generic classification of the caracaras, and secondly, the patterns of distribution, geographical variation and speciation in these birds, with particular emphasis on the Andean taxa. I shall deal either with problems not, or only litde, covered by Brown and Amadon (1968) in their recent book, or 1 Biology Department, University of Massachusetts, 100 Arlington Street, Boston, Massachusetts 02116. 2 BREVIORA No. 355 with controversial issues, especially when my own conclusions differ from theirs. The data were obtained from examination of about 250 skins and some skeletons; and from field studies on the habitat prefer- ences, general behavior, and distribution of five taxa of caracaras made over a period totaling thirteen months during trips to South America in 1964, 1965, and 1967-68. THE CARACARAS The 7 to 10 or 11 species of caracaras can be distinguished from other Falconidae more by their vulture-like external morpho- logical characters and associated scavenging habits, than by clear- cut anatomical characters (Friedmann, 1950: 719). They differ from true falcons, however, in their habit of building their own nest, as pointed out by Brown and Amadon (1968: 23, 104). The problem of whether the caracaras should be accorded taxonomic rank within the Falconidae, and if so, which one (e.g., subfamily, tribe), will not be discussed here. The species of caracaras have traditionally been placed in four genera (see, e.g., Peters, 1931; Hellmayr and Conover, 1949; Friedmann, 1950; de Schauensee, 1966; Brown and Amadon, 1968): Daptrius Vieillot, 1816, with two species; Milvago Spix, 1824, with two species; Phalcoboenus d'Orbigny, 1834, with two to four species; and Poly bonis Vieillot, 1816, with one to three species. (The long controversy over the names Polyborus versus Caracara Merrem, 1826, was resolved by Amadon (1954), who showed that "Vieillot's diagnosis of Polyborus applies to the Cara- caras," so that "the diagnosis sustains the name." Caracara must, therefore, go into synonymy, and the issue can be considered closed.) As a group, the caracaras are distributed from temperate North America southward to Central and South America, ranging as far south as the Falklands and the islands off Tierra del Fuego. Sev- eral species have very broad ranges, especially Polyborus plancus, the distribution of which encompasses almost that of the entire group. Other species are, on the contrary, quite localized geo- graphically. For example, the extinct Polyborus lutosus occurred only on Guadalupe Island off Baja California, and the living Phal- coboenus australis breeds on a handful of small islands off the southern coast of Tierra del Fuego and on the Falklands. 1970 SPECIATION IN THE CARACARAS 3 Correlated with the broad distribution of the caracaras is their ecological diversity. As a group, these birds inhabit most vegeta- tion formations found in their geographical range, from desert scrub to tropical lowland wet forest, and including the highest zones of Andean vegetation. About the only major type of vegeta- tion not favored by caracaras is montane tropical wet forest (or cloud forest). All the taxa of caracaras are scavengers, but should perhaps be viewed as omnivorous, since their diet also includes live prey (ver- tebrates and invertebrates) and vegetal matter (see, e.g., Hud- son, 1920: 62-88; Wetmore, 1926: 92-96; Haverschmidt, 1962: 157-158; Friedmann, 1927: 157; Friedmann and Smith, 1950: 450-451, and 1955: 486-487; Brown and Amadon, 1968). Among the most notable food specializations of the caracaras is the habit of Polyborus planciis of attacking domestic animals the size of sheep (Johnson, 1965: 263), and the marked predilection shown by Daptrius americanus for colonial wasps (Skutch, 1959). Most caracaras are social, at least during the nonbreeding sea- son, and form intraspecific flocks that appear to be feeding associa- tions. The larger-sized species may form smaller flocks than the smaller ones. Thus the largest flock of the osprey-sized Polyborus plancus I have observed comprised about 15 birds, while I have often seen much larger flocks of the kestrel-sized Milvago chimango. The intraspecific gregariousness of some of the smaller species may also extend to the breeding season. Drury (personal communica- tion ) observed a colony of Milvago chimango where the nests were only about 30 feet (ten meters) apart. The caracaras also form associations with vertebrates other than birds. In Patagonia, Mil- vago chimango flocks are frequent near cattle and horses; I even saw one bird sitting on the flank of a lying horse, pecking from time to time at the skin, perhaps to eat ticks. Hudson (1920: 70) men- tioned that M. chimango "follows the plough," thus playing the part of gulls {Lams) elsewhere. The habit of following moving vehicles from which scraps of food can be collected has been observed in Phalcoboenus megalopterus along the Yungas Road in La Paz, Bolivia (Niethammer, 1953: 265; personal observa- tion), and in Polyborus plancus along a railroad in the Bolivian Chaco (Eisentraut, 1935: 391). The gregarious behavior exhib- ited by the caracaras may be correlated to a large extent with their scavenging habits, yet, from my own observations, I would judge their behavior to be much more plastic and diverse than that of 4 BREVIORA No. 355 other scavengers such as Old World or New World vultures (Cathartidae and Aegypiinae, respectively). ANALYSIS OF SOME CHARACTERS OF THE CARACARAS Size The caracaras vary considerably in size, as several authors (Friedmann, 1950; Brown and Amadon, 1968) have already pointed out. This variation is obvious if wing length, in the ab- sence of data on weights, is used as an indicator of overall body size (Table 1). Taxonomists have often been bothered by size differences between species, and have been reluctant to place in the same genus closely related species which differed conspicuously in size but in few, or no, other characters. This attitude might have been prevalent among the ornithologists who worked with caracaras, because the two smallest species belong in the genus Milvago and the largest species in Phalcoboenus and Polyborus. Yet the two species of Daptrius bridge the gap between these ex- tremes. I therefore believe that size should not be given undue weight in the supraspecific classification of the caracaras. In other Falconidae, notably in the genus Falco, similarly large size differ- ences between species have not prevented their inclusion in the same genus. Proportions In a number of bird taxa, a proportionately short tarsus is cor- related with arboreal habits, and a long tarsus with more terrestrial habits. Most species of caracaras are both arboreal and terrestrial, but some definite trends toward one or the other of these habits exist, especially in regard to feeding habits. Thus Polyborus and Milvago are often seen perched on trees, and they breed in trees, but they do a lot, perhaps most, of their foraging for food on the ground. The species of Phalcoboenus (with the possible exception of P. albogularis) seem to forage entirely on the ground, and breed in cliffs and rocky slopes; they do on occasion perch on buildings. The species of Daptrius seem at variance with the other three genera because they do some, perhaps even a substantial, portion of their foraging for food in trees. From this summary one would thus expect the species of Dap- trius, which are more arboreal than the other caracaras, to have proportionately shorter tarsi than other species. Figure 1 shows a 1970 SPFXIATION IN THE CARACARAS I/) -a c o •T3 o o o c o. o a a 1 "O I "O nj c ij c w -o (U C3 (U 03 — ^ j= X) x: g l-l >> J3 V— ^ ' — ' O en >< •a u 60 C ^ 01 iJ - c >^ c >< ^ cS 03 o k. u ^ ° ° « CQ < c o S cd cd ed o (/3 c .2 Cd J) -13 LU OJ - O Z a c o 03 'E o X) x; c o c o c o C o a I X) ka O 13 C o x: "c o c o k> Xi •g t: T3 c .-s g ■- c Ox: I ^ =^ E & C ^ U 03 Kj ■^ y • - c °« o 3 X X =8 «8 o CO 60 o lO 0\ O vO u-i 00 vOVOOOOOn^iOnvO — ^-H ^— ' ^ ^ *t^ >4^ IJU V.^^ IJA ' — ' W> ^k-" '— • ^^ •—< -^^ II r II ^ II — ' '7 O ^7 O m ^-i -^ ^ rn OS v^ II ^ II O "T- O "T- m 1^ m 1^ o 00 00 rj ^^ -H r-~ ^^ 00 r-~ r~- ^^ r<^ m m "— ' r^ VD 00 00 <^ II oo " r1 0 H •c: a s s s -2 ^ 1 60 6 BREVIORA No. 355 scatter diagram of tarsus length plotted against wing length (rep- resenting body size in the absence of weight data) in the caracaras. The two species of Milvago, although overlapping considerably in wing length, show little overlap in tarsus length. The southern species, M. chimango, is decidedly longer-legged than the northern one, M. chimachima. This difference, especially in regard to the zone of sympatry of the two species, is discussed again below in the section on speciation in these birds. Dapthiis ater has somewhat longer wings than either species of Milvago, but its tarsal measure- ments overlap fully with those of M. chimachima. Yet D. ater would seem more arboreal than M. chimachima. The arboreal Daptrius americanus overlaps with the more terrestrial Phalco- boenus megaloptenis, P. albogularis, and Polyborus plancus in wing length, but has a much smaller tarsus. In this case the differ- ence in tarsus size appears to correspond to a difference in habits. TARSUS LENGTH Cmm) \ -•■□ • Y-^-: I* * ,' \*_.. -■■-'■ *; / 4 * «"*> / .Q' i®. WINO LENSTM (mm) (*j^'' planeus :*;:, \*^*^ chimango /odj megalopi er u« 5JP.S-' \P^/' ater '\**y "Ibogulari* \^i-. amerieanuo Figure 1. Tarsus length plotted against wing length in seven taxa of caracaras. 1970 SPECIATION IN THE CARACARAS 7 It is worthy of note that the species of Phalcoboemis closest to D. americamis in tarsus length is P. albogularis (see Fig. 1) said by Olrog (1948: 478; 1950: 520) to occur in forests and thus to diverge from the habitat preferred by P. megalopterus. On the basis of the proportions of tarsus and wing lengths plotted on Figure 1 , Milvago chimango, M. chimachima, and Daptrius ater would form one group of species closely similar to each other, over- lapping greatly in both wing and tarsus length. A second such group would include Phalcoboemis and Polyborus, while Daptrius americanus appears clearly distinct from all other species, yet inter- mediate between the two groups. Interspecific variation in tail length is summarized in Table 2. When handling skins, Daptrius americamis appears distinctly longer tailed than the other species of caracaras. If the tail/wing ratios, instead of the absolute tail lengths, are compared among the species of caracaras, then a smooth variation is detected, from the propor- tionately short-tailed Polyborus plancus and Phalcoboemis mega- lopterus to the long-tailed Milvago chimachima and Daptrius americamis. No groups of caracaras can be established on the basis of relative tail length. TABLE 2 Variation in tail length, expressed in percent of wing length, among the caracaras. The taxa are ranked in order of increasing tail/wing ratio. Both sexes, and adult and subadult birds included. Taxon Tail/ Wing X 100 Sample Size Range Mean plancus 50.8-58.1 53.9 42 megalopterus 53.1-57.4 55.5 13 carunculatus 55.6 1 chimango 54.8-58.7 57.3 5 albogularis 54.1-64.7 58.4 5 australis 57.5-61.5 59.4 4 ater 58.4-63.5 61.1 6 lutosus 62.7 1 chimachima 60.9-70.2 66.2 25 americanus 62.2-74.6 68.4 36 8 BREVIORA No. 355 Plumage Color and Pattern Good descriptions of both adult and immature plumages of caracaras have been given by Friedmann (1950) and Brown and Amadon (1968), who should be consulted for full details. The main colors of aduh and immature plumages are summarized in Table 1. Adult caracaras vary from a brown, rather unpatterned plumage (such as is found in a variety of birds of prey) to a highly pat- terned plumage. Thus Milvago chimango has a brown plumage, variegated somewhat with barrings, reminiscent of the plumage of some kites of the genus Milvus. The plumages of both species of Daptrius, and of the three Andean species of Phalcoboenus (carun- culatus, megalopterus, and albogularis) are very different from those of the other caracaras. These birds are strikingly marked with black and white patterns, and remind one of gallinaceous birds such as Crax. Immature caracaras are less varied in their plumages. Most of them are brownish with patterns of barring or streaking, with the exception of the immature Daptrius americanus, which is like the adult. The dimorphism between immature and adult can be very well marked, as in Phalcoboenus carunculatus, P. megalopterus, and P. albogularis; or poorly marked, as in Daptrius americanus or Milvago chimango; or intermediate, as in Polyborus plancus and P. lutosus. In spite of the variation in plumage color and pattern exhibited by the different species of caracaras, certain basic resemblances are evident in the group as a whole when adult and immature plumages are compared together. For example, the immatures of Milvago chimachima, Phalcoboenus carunculatus, P. megalopterus, P. albo- gularis, Polyborus lutosus and P. plancus are all very similar to one another, as they are to the adult of Milvago chimango. Thus, regardless of the intraspecific variabihty between immatures and adults on the one hand, and the interspecific variability among adults on the other, Milvago, Phalcoboenus, and Polyborus all appear to have similarities in plumage color and pattern. In every genus except Polyborus, there is a dichotomy between one or more species that have different immature and adult plum- ages, and one or more species that have quite similar immature and adult plumages. This dichotomy is obvious in Milvago, with 1970 SPECIATION IN THE CARACARAS 9 chimanfio plumages much alike, and chimachima unlike; in Phal- coboenus, with australis plumages much alike, and with canincu- latus-megalopterus-albogularis plumages very different. The dicho- tomy is still present, but much less obvious, in Daptrius, with americanus plumages alike and ater plumages slightly different. Correlation between this dichotomy and patterns of geographical distribution among closely related species is not readily apparent. In Milvago and Phalcoboemis , the species which differ least in their adult and immature plumages are the two southern ones; but in Milvago alone, there is an important geographical overlap between chimango and chimachima, whereas there is no or only a very nar- row overlap between the southern australis and the more northern carunculatus-megalopterus-albogularis complex. In Daptrius (di- chotomy poorly marked), there is almost complete overlap between the ranges of the two species, since the range of ater is almost entirely contained within that of americanus. In Polyborus (no, or almost no dichotomy ) , lutosus and plancus are entirely allopat- ric. (The possible ecological significance of the immature and adult dimorphism is discussed below under habitat preferences.) If one were to consider the immature-hke plumage of some sex- ually mature birds as being a primitive condition, and the strikingly different plumage in other mature birds as an advanced one (i.e., secondarily developed during the evolutionary history of the cara- caras), then Polyborus and Daptrius might be thought to be more primitive than the other two genera. If a brownish, barred and streaked immature plumage were considered a more primitive con- dition, then Daptrius is more advanced in this character than the three other genera, because in that genus, immatures are hardly different from adults, being also conspicuously patterned in black and white. Naked Facial Skin Every species in the genera Polyborus, Phalcoboenus, and Dap- trius has brightly colored naked skin between the bill and the eye, and often around the eye or even the throat. In the genus Milvago, only M. chimachima has naked facial skin; M. chimango has a fully feathered head. The area of unfeathered skin varies from species to species among those that have naked facial skin. In the species of Phalcoboenus, this variation is geographical. In Phalcoboenus carunculatus (Colombia and Ecuador), the surface of facial skin is extensive, and the throat is even adorned by fleshy wattles. In 10 BREVIORA No. 355 p. megalopterus (Peru, Bolivia, and northern Chile-Argentina), only the lores are unfeathered, while the throat is largely feathered. In P. albogularis (Patagonia), the extent of bare loral skin is smaller than in megalopterus. Finally, in P. australis (southernmost islands of South America), the facial skin is almost entirely feath- ered, but there are wattles on the breast instead. The difference between P. australis and P. carunculatus-megalopterus-albogularis may be related to the fact that australis is sympatric with Polyborus plancus, a species having extensive naked facial skin, whereas the other three species of Phalcoboenus are the only caracaras in their respective ranges. The naked facial skin in the caracaras varies from salmon pink to rose-red and from yellowish to bright red (see Table 1). This variation is both intra- and interspecific. I have seen the facial skin of one individual of Polyborus plancus change, in a few sec- onds, from pale yellow to salmon-pink and finally to vivid red. Such a rapid change cannot easily be ascribed to hormonal influ- ences, but is more likely to be due to a sudden flush of blood to the superficial vessels of the skin. Brown and Amadon (1968: 738) mention a reverse change in P. plancus: "bare facial skin carmine red, changing to yellow when excited." The variation in facial skin color in the caracaras (Table 1) seems too extensive within species, and too restricted between species (differences between, say, yellow and orange, or orange and red seem relatively slight) to be important as a species-specific means of recognition. Presence or absence, together with color and area of unfeathered skin might, however, play such a role among true sympatric species, e.g., Phalcoboenus australis and Polyborus plancus already cited, or Milvago chimango and M. chimachima. Nostril Shape I have examined skulls of Polyborus plancus (several speci- mens), Phalcoboenus australis (4), Daptrius sp. (1 ), and Milvago sp. (3). Polyborus plancus has slanted, elongated nostrils (bean- shaped), while Phalcoboenus australis, Daptrius sp., and Milvago sp. all have rounded nostrils. This difference is also visible on the cere of study skins and seems to be confined to Polyborus (Table 1 ) , since only P. lutosus and P. plancus have bean-shaped nostrils. Swann (1925: 66) and Friedmann (1950: 545-546) used this 1970 SPECIATION IN THE CARACARAS 11 difference in nostril shape, together with other characters, as diag- nostic features to key out the genera of caracaras. I do not know what the biological significance of this difference may be. Habitat Preferences The preferred habitats of Milvago, Phalcoboenus, and Polyborus include a variety of nonforest types: open scrub, grassy pampas, tussock-grassland, pastures, treeless cultivated farmland, open thorny chaco, savanna woodland, and open plantations. Phalco- boenus albogularis may be an exception, since Olrog (1948: 478; 1950: 520) reports it as a forest bird. Both species of Daptrius are inhabitants of tropical forest and contrast markedly with the other caracaras in this feature. The distribution of ater and americamis corresponds to the distribution of wet lowland rain forest in Central and South America, but the actual preference of these species seems to be less for the forest interior than for more open situations within or along the forest: river banks, small clearings, secondary growth, mangroves, and tree-tops. The plumage pattern of both species of Daptrius, espe- cially D. americamis, is most similar to that of adults of Phalco- boenus carunculatus and P. megalopterus, which inhabit high Andean steppes, an environment that differs in every respect from that of Daptrius. This basic morphological similarity among taxa of widely distinct ecological preferences seems to be more easily understandable on an hypothesis of close relationship than on one of convergent evolution. The fact that the species of Phalcoboenus mentioned above have a very different immature plumage, while those of Daptrius are far less dimorphic between adult and imma- ture, might be a correlate of environmental differences between the two genera. In the open habitats of the high Andes there is never more than one species of caracara at any one locality (diversity = 1 ) . Because congeneric competitors are absent, the sharp dimorphism between adults and immatures of Phalcoboenus may consequently reflect the results of slightly relaxed selective pres- sures. In the tropical lowland forests, however, there are two sympatric species of caracaras (diversity = 2), so that selection through interspecific competition might possibly limit the range of intraspecific variability, thus resulting in the evolution of similarity (monomorphism) between immatures and adults of Daptrius. The difference in size (character divergence) between the two species 12 BREVIORA No. 355 of Daptrius, which are sympatric, may be relevant here. Being so different from each other (no overlap in range of wing lengths), they should compete very little for food (see Schoener, 1 965 ) . The interspecific difference in size, together with the lack of dimorphism between age categories in Daptrius, may be roughly equivalent ecologically to the striking dimorphism within high Andean Phal- coboenus, which do not have any sympatric congeners. In Milvago, where the two species are largely allopatric, one observes a con- siderable overlap in wing lengths between the two, but one of the species is conspicuously dimorphic (chimachima) , whereas the other is not. Therefore, the situation in Milvago appears inter- mediate to that in both Daptrius and Phalcoboenus. I beheve this intermediacy is also found in the habitats occupied by Milvago (such as open woodland, savannas) that are more or less inter- mediate between lowland wet forest (occupied by Daptrius) and barren high Andean steppes (occupied by Phalcoboenus). Summary of Character A nalysis If the characters discussed above are examined separately, the variation among some of them shows the following possible group- ing within the caracaras. (a) Polyborus is distinct in nostril shape from the other genera, (b) Daptrius and Phalcoboenus are closer to each other than to other genera since both have a strikingly patterned black and white adult plumage, (c) The immature plum- ages of Milvago, Phalcoboenus and Polyborus appear extremely similar to one another, and differ, as a group, from those of Dap- trius. If both adult and immature plumages are used, together with habitat preferences, the two species of Daptrius appear to stand out against most other species. First, the immature plumage of Daptrius, when distinct from that of the adult (as in D. ater), is quite different from the brownish, streaked or barred immature plumage of all other species except Phalcoboenus australis. Sec- ondly, the forest habitat of the species of Daptrius differs from the nonforest habitats of the species in the other genera, with the pos- sible exception of Phalcoboenus albogularis. (The remaining char- acters [naked facial skin and size] seem of httle or no use in estabUshing groups within the caracaras.) I can only conclude from this analysis that all caracaras appear to be closely interrelated, but that Daptrius is less similar to the other three genera than these are to each other. 1970 SPECIATION IN THE CARACARAS 13 CLASSIFICATION The splitting of the caracaras into four genera, endorsed by most taxonomists, does not seem to reflect properly the close relation- ships of these birds. Since the lumping of all caracaras into a single genus may be going somewhat too far in the opposite direction, I suggest here a third possibiUty, which is to put the caracaras in- habiting nonforest habitats in a single genus (Polyborus, including Milvago and Phalcoboemis) and to keep the forest caracaras in a second genus (Daptrius). Further subdivisions within the non- forest caracaras can be made by using species-groups and super- species, which do not burden the nomenclature with additional names (as pointed out by Cain, 1954), yet permit a finer hierarchy between the genus and species levels. I present below a classification outline of the caracaras, includ- ing species-groups and superspecies. Each species-group represents a former genus. The grouping of some species in a superspecies (included in braces) is given here in anticipation of the discussion on speciation in the next section of this paper. The sequence of taxa in this list is arbitrary, and does not pretend to suggest that some taxa are more primitive than others, since such decisions would be guess work. Genus Dap /m/5 Vieillot, 1816 (forest caracaras) D. ater Vieillot, 1816 D. americanus (Boddaert, 1783) Genus Polyborus Vieillot, 1816 (nonforest caracaras) 1 . chimachima species-group P. chimachima Vieillot, 1816 P. chimango Vieillot, 1816 2. plancus species-group P. plancus (Miller, Mil) (includes lutosus Ridgway, 1876, considered by some authors as a separate species) australis superspecies iP. australis (Gmelin, 1788) \p. megalopterus (Meyen, 1834) (includes carunculatus (Des Murs, 1853), and albogularis Gould, 1837, con- sidered by some authors as two separate species) 14 BREVIORA No. 355 VARIATION AND SPECIATION IN DAPTRIUS Of the two species of Daptrius, only D. americanus shows geo- graphical variation. This variation seems to be a cline of diminish- ing size, as measured by wing length, from Guatemala southward through Central America to South America. According to Brown and Amadon (1968), the populations from "southern Brazil" may show an increase in wing length over those from farther north in South America. This increase is in fact quite sharp, and is illus- trated by Figure 2. The disjunction may reflect absence of gene flow between birds from the Amazon Valley and those from the uplands of Brazil. The birds Hving in the coastal forests of south- ern Brazil and in the gallery forests of the Parana-Paraguay drain- age system of south-central Brazil may be ecologically isolated from birds living in forests of the Amazon Valley by parts of the central Brazilian plateau, which are covered by extensive open savannas and campos. Any inferences that might be drawn about the possible evolu- tionary history of this genus are prevented by the considerable sympatry between ater and americanus (see map 76 in Brown and Amadon, 1968). 320 330 340 350 360 370 360 390 400 4 10 I \ \ \ \ 1 \ 1 \ 1 Guatem,, Hond.Nicar. Cos, Rica Ecuador. Peru Brazil (Goias, Mato Grosso) Figure 2. Geographical variation of wing length in Daptrius americanus. Measurements are in miHimeters. Horizontal bars: range of measurements; vertical bars: means. 1970 SPECIATION IN THE CARACARAS 15 VARIATION AND SPECIATION IN POLYBORUS 1. The chimachima Species-Group The two species of this group, which constitute the former genus Milvago, are largely allopatric. Polyborus chimachima occurs in southern Central America (Costa Rica and Panama) and in South America from Colombia and Venezuela in the north to northern Argentina in the south, and P. chimango from northern Argentina southward to southernmost South America. They are sympatric, however, over a relatively broad zone, including Rio Grande do Sul in southern Brazil, parts of northern Argentina, Uruguay, and Paraguay, and southern Bolivia westward to the foothills of the Andes. Polyborus chimachima shows color and size variation. Tail and culmen length appear to vary clinally, and to increase from south to north, whereas the reverse seems to be true of wing length (see Fig. 3). Polyborus chimango shows geographical variation that seems to conform to eco-geographical rules, since the southernmost birds are the largest (Bergmann's rule), and the birds occurring along the wet, forested Andean slopes are darker than those living in the drier, grassy and shrubby plains of central Argentina (Gloger's rule). Geographical variation in P. chimango has been recognized tax- onomically by the naming of three subspecies. Two of them, temu- coensis (Andean slopes birds) and chimango (open plains birds) intergrade broadly. The third subspecies, juegiensis, seems re- stricted to the island of Tierra del Fuego, but is probably merely the southward end of a north to south cline of increasing size. Interestingly, the southern populations, from Tierra del Fuego and the mainland of southern Patagonia, appear to leave their breeding grounds to migrate northward as far as northern Argentina (Olrog, 1962: 112-113) in the southern hemisphere winter. Although P. chimachima and P. chimango are undoubtedly very close relatives, they have diverged morphologically rather consider- ably. One of the differences has been mentioned earlier: the im- mature plumage of chimango is very similar to that of the adults, whereas the immature chimachima is quite unlike the adult. An- other difference between the two species may be ecological. As I 16 BREVIORA No. 355 N n (H I 1- (9 Z S 111 _j o z w u S 01 -J D 0 00 0 z o 5 o o CM o 0) - 10 o in J o _ o . o t + -■ 5 o E E 0 0> - (M 0 O ■ 01 0 1^ - o s. A \ 4- ft t X u 0 O) c ig E £ i z o o Co s Q s ■"• .« VI 0^ •o c cfl /■^ N^ s ^ o Q P» ,-*. -S u 69 a v. O O PK •"-J- o a. c •"- M x: U c 4J a 4> B "3 u (A •o c c a C3 -* S" k. ^ cS a X) ■? 3 > a . », ■4-» CO ■♦ij <«-l c o (U W5 E c u _o b. 3 •c a: > s "cH o o Q J= S "^ "II I I + + I I i ^ ^ I +1111 + 1^ en to Q *5 Oi <3 5 s 5 is o O u (A 4-* c C z o ?J *>M o kSt u o [« a J3 0^ !£ C •o Ic a c X m CS u ta ?3 >^ £. 03 to ^•- E -^ £t tj ^5 a a ^ .jj .1 s: ■S .s c o Ci. •» a 1^ tn ■2 'c -Q ^ ex ^ ^ 53 "o o Q a. a. 1— ( OJ M 1970 SPECIATION IN THE CARACARAS 25 to have treated them as species. There is some evidence that (sec- ondary? ) hybridization takes place near the mouth of the Amazon where they come into contact. The taxa of the Polyborus australis superspecies present another "stage" of the speciation process, in that one of the members of the superspecies, australis, is sufficiently distinct to be considered unhesitatingly as a species. The three remaining members, how- ever, present interesting situations. In one instance, differentiation seems to have taken place across an ecological barrier (the low area of the northern Peruvian Andes), yet hybridization appears to occur in spite of this gap. In the second instance, no barrier is evident today, and some hybridization seems to occur. The south- ernmost taxon of this complex {albogularis) apparendy meets australis in Tierra del Fuego. In the Polyborus chimachima species-group, speciation is com- pleted, and the two species overlap now over a considerable area, although they are allopatric over the major portions of their respec- tive ranges. Finally, the two species of the genus Daptrius are so different morphologically and show so much sympatry that recon- struction of their history is impossible. Although six of the seven species I recognize in the caracaras have extensive geographical distributions, the existing patterns of speciation, or incipient speciation, seem to indicate that multiphca- tion of species has occurred mostly through the formation, and subsequent differentiation, of small or relatively small peripheral isolates. The present isolates of Daptrius americanus and Polyborus plancus are restricted to small areas around the periphery of the range of the species, and the geographical location of Polyborus australis relative to Polyborus megalopterus seems to suggest for- mer peripheral isolation of the first named species. Differentiation within P. megalopterus does not seem to correspond as clearly to a pattern of isolation in peripheral areas, although the central populations {megalopterus) do have a much broader distribution than either the northern (carunculatus) or southern (albogularis) ones. The previous summary of speciation in the caracaras shows that this process is most actively taking place along the Andes, where the various members of the Polyborus australis superspecies live. In the other, lowland, taxa, the speciation process is either com- pleted (as in Daptrius or the Polyborus chimachima species-group) or is not as pronounced {Polyborus plancus, Daptrius americanus) . 26 BREVIORA No. 355 It is tempting to relate the apparently greater evolutionary activity in the Andean caracaras to the recent geological history of this Cordillera. The high Andean grasslands and scrub habitats (paramo and puna) where Polyborus megalopterus now lives are undoubt- edly the most recent environments of the Andes, and cannot be older than the latest phases of uplift, which brought the mountains to their present tremendous altitudes during the Pho-Pleistocene (see Childs and Beebe, 1963; Steinmann, 1930; Ahlfeld and Branisa, 1960; and Briiggen, 1950; for summaries of the geological development of the Andes). The differentiation within P. megalop- terus most probably occurred during the Pleistocene glaciations, although to attempt the dating of such processes is almost complete guess work. However, if we recall that during glacial episodes, the temperature depression lowered the altitude of the upper vegeta- tion zones, where P. megalopterus lives, then it becomes possible to envision the separation of a northern isolate (proto-carunculatus) in Ecuador at interglacial time, when the altitudinal raising of this treeless zone occurred, thus increasing the effectiveness of a natural barrier such as the Upper Maranon Valley and northern Peruvian low for birds living on either side of it. The isolation of caruncu- latus from megalopterus, or, rather, of proto-carunculatus from proto-me galopterus, might, then, have happened during an inter- glacial. It is, of course, not possible to suggest which of the several interglacial periods was responsible for such an event. Similar glacial-interglacial oscillations may have permitted the separation of proto-megalopterus and proto-australis in extreme southern South America. During the maximum glacial, extreme southern South America was covered with an ice-sheet (Caldenius, 1932; Polanski, 1965) which probably forced Andean biota to "retreat" considerably northward. At the same time, however, the Falkland Islands were left unglaciated, and, furthermore, were of greater area than today because of a concurrent lowering in sea- level. It seems therefore possible that during the maximum glacia- tion the southernmost populations of the stock common to mega- lopterus and australis remained on a Falkland refuge, where they were geographically isolated from mainland populations by the ice barrier, added to the sea barrier. If such a separation did indeed take place during the maximum glaciation, which is attributed to the late Pleistocene (Wiirm or Wisconsin) (see Polanski, 1965), then the splitting of an ancestral stock into the modern australis (having evolved from a population in a southern insular refuge) 1970 SPECIATION IN THE CARACARAS 27 and megalopterus (having remained in Andean Patagonia, but con- siderably farther north than its present-day southernmost limit) may have taken place as recently as 50,000 to 80,000 years ago. Of course, it is also possible that the original separation took place during an earlier, somewhat less extensive, glacial episode, but since the maximum glaciation apparently obliterated earlier re- mains, it is futile to speculate any further about the possible course of this event. ACKNOWLEDGMENTS I thank Ernst Mayr for his continued advice throughout my studies of speciation in Andean birds. He and Dean Amadon criticized an earlier version of this paper. W. John Smith and William H. Drury, Jr. kindly allowed me to use their field notes on several species. Field work was financed by the Frank M. Chapman Memorial Fund of the American Museum of Natural History, the National Science Foundation (grants G-19729 and GB-3167 to the Com- mittee on Evolutionary Biology of Harvard University), the Society of the Sigma Xi, and the Bourse federate de voyages of the Societe helvetique des sciences naturelles (Switzerland). I am grateful to the following persons, who helped me in various ways during my examination of specimens: Jean Dorst (Museum national d'histoire naturelle, Paris), I. C. J. Galbraith (British Museum, Natural History), Father Antonio Olivares (Instituto de Ciencias Naturales, Bogota), Raymond A. Paynter, Jr. (Museum of Comparative Zoology), and the late R. A. Philippi (Museo Nacional, Santiago). Finally, I acknowledge the assistance of Warren Hubley during the course of this study. BIBLIOGRAPHY Abbott, C. G. 1933. Closing history of the Guadalupe Caracara. Condor, 35: 10-14. Ahlfeld, F., and L. Branisa. 1960. Geologia de Bolivia. La Paz, Insti- tuto Boliviano del Petroleo, Editorial Don Bosco. 245 pp. Amadon, D. 1954. On the correct names for the caracaras and for the Long-winged Harrier. Auk, 71: 203-204. 1964. Taxonomic notes on birds of prey. Amer. Mus. Novi- tates, No. 2166: 1-24. Brown, L., and D. Amadon. 1968. Eagles, Hawks, and Falcons of the World. 2 Vols. New York, McGraw-Hill Co. 945 pp. 28 BREVIORA No. 355 Bruggen, J. 1950. Fundamentos de la geologia de Chile. Santiago, Insti- tuto Geografico Militar. 374 pp. Cain, A. J. 1954. Subdivisions of the genus Ptilinopus (Aves, Columbae). Bull. British Mus. (Nat. Hist.), Zool. 2 (8): 265-284. Caldenius, C. C. 1932. Las glaciaciones cuaternarias en la Patagonia y Tierra del Fuego. Geogr. Annaler, 14: 1-164. Cawkell, E. M., and J. E. Hamilton. 1961. The birds of the Falkland Islands. Ibis, 103a: 1-27. Childs, O. E., and B. W. Beebe (Eds.). 1963. Backbone of the Americas. Amer. Assoc. Petroleum Geol., Memoir 2. 320 pp. EiSENTRAUT, M. 1935. Biologische Studien im bolivianischen Chaco. V. Beitrag zur Biologie der Vogelfauna. Mitt. Zool. Mus. Berlin, 20: 367-443. Friedmann, H. 1927. Notes on some Argentina birds. Bull. Mus. Comp. Zool., 68 (4): 139-236. . 1950. The birds of North and Middle America. Part XI. Bull. U.S. Nat. Mus., 50: 1-793. Friedmann, H., and F. D. Smith, Jr. 1950. A contribution to the orni- thology of northeastern Venezuela. Proc. U.S. Nat. Mus., 100: 411-538. . 1955. A further contribution to the ornithology of north- eastern Venezuela. Proc. U.S. Nat. Mus., 104: 463-524. Grant, P. R. 1965. A systematic study of the terrestrial birds of the Tres Marias Islands, Mexico. Postilla, No. 90: 1-106. Greenway, J. C, Jr. 1958. Extinct and vanishing birds of the world. New York, Special Publ. No. 13, Amer. Comm. Internat. Wildlife Protection. 518 pp. Haverschmidt, F. 1962. Notes on the feeding habits and food of some hawks of Surinam. Condor, 64: 154-158. Hellmayr,, C. E., and B. Conover. 1949. Catalogue of birds of the Americas and the adjacent islands. Part I, No. 4. Field Mus. Nat. Hist.. Publ. 634, Zool. Ser., 13: 1-358. Hudson, W. H. 1920. Birds of La Plata. Vol. 2. New York, E. P. Dutton Co. 240 pp. Johnson, A. W. 1965. The Birds of Chile and Adjacent Regions of Argen- tina, Bolivia, and Peru. Vol. I. Buenos Aires, Piatt Establecimientos Graficos S.A. 398 pp. KoEPCKE, M. 1964. Las Aves del Departamento de Lima. Lima. Morsom S.A. 128 pp. Niethammer, G. 1953. Zur Vogelwelt Boliviens. Bonner Zool. Bietr., 4: 195-303. Olrog, C. C. 1948. Observaciones sobre la avifauna de Tierra del Fuego y Chile. Acta Zool. Lilloana, 5: 437-531. . 1950. Notas sobre mamiferos y aves del archipielago de Cabo de Hornos. Acta Zool. Lilloana, 9: 505-532. 1970 SPECIATION IN THE CARACARAS 29 . 1962. Notas ornitologicas sobre la coleccion del Institute Miguel Lillo (Tucuman). VI. Acta Zool. Lilloana, 28: 111-120. . 1963. Lista y distribucion de las aves argentinas. Opera Lilloana, 9: 1-377. Peters, J. L. 1931. Check-list of Birds of the World, Vol. I. Cambridge, Mass., Harvard Univ. Press. 345 pp. Philippi. B., R. a., a. W. Johnson, J. D. Goodall, and F. Behn. 1954. Notas sobre aves de Magallanes y Tierra del Fuego. Bol. Mus. Nac. Hist. Nat., Santiago, 26: 1-55. PoLANSKi, J. 1965. The maximum glaciation in the Argentine cordillera. Geol. Soc. Amer. Special Paper 84: 453-472. Reynolds, P. W. 1932. Notes on the birds of Snipe, and the Woodcock Islands, in the Beagle Channel. Ibis, Ser. 13, 2: 34-39. Schauensee, R. M. de. 1966. The Species of Birds of South America and Their Distribution. Narberth, Pennsylvania, Livingston Publ. Co. 577 pp. ScHOENER, T. W. 1965. The evolution of bill size differences among sym- patric congeneric species of birds. Evolution, 19: 189-213. Scott, W. E. D. 1910. An apparently new species of carrion hawk of the genus Ibycter. Auk, 27: 152-153. Skutch, a. F. 1959. Red-throated Caracara — the scourge of the wasps. Animal Kingdom, 62: 8-13. Steinmann, G. 1930. Geologia del Perii. Heidelberg, Carl Winters Uni- versitatsbuchhandlung. 448 pp. Swann, H. K. 1925. A Monograph of the Birds of Prey (Order Accipitres). London, Wheldon & Wesley, Ltd., Part II: 53-124. Vuilleumier, F. 1968. Population structure of the Asthenes flammulata superspecies (Aves: Furnariidae). Breviora, No. 297: 1-21. . 1969. Systematics and evolution in Diglossa (Aves, Coere- bidae). Amer. Mus. Novitates, No. 2381: 1-44. . 1970. Chapter 1. Generic relationships and speciation pat- terns in Ochthoeca, Myiotheretes, Xolinis, Neoxolmis, Agriornis, and Miiscisaxicola, In Smith, W. J., and F. Vuilleumier, Evolutionary rela- tionships of some South American ground tyrants. Bull. Mus. Comp. Zool. (in press). Wetmore, a. 1926. Observations on the birds of Argentina, Paraguay, Uruguay, and Chile. Bull. U.S. Nat. Mus., 133: 1-448. Zimmer, J. T. 1930. Birds of the Marshall Field Peruvian expedition. 1922-1923. Field Mus. Nat. Hist., Publ. 282, Zool. Ser., 17: 233-480. BREVIORA Mnaseiiamn of Coimparative Zoology Cambridge, Mass, 30 November, 1970 Number 356 On new species in a new earthworm genus from Puerto Rico^ G. E. Gates2 Abstract. A new genus of earthworms, Estherella (family Glossosco- lecidae), with new species montana (type species) and nemoralis, is described from Puerto Rico, and its remarkable structural modification is discussed. A seemingly impossible evolutionary modification was shown by several worms received 25-30 years ago. Immaturity, amputa- tion, maceration, and paucity of specimens prevented completion of species descriptions to desired standards. Unfortunately subse- quent material never became available. Various attempts to secure it were futile. This contribution now is proferred in hope of awakening interest in: 1) An evolutionary development that some zoologists categorically maintain is impossible. 2 ) The mostly unknown earthworm faunas of Caribbean Islands. Glossoscolecidae Estherella gen. nov. Definition. Digestive system, with a gizzard in iii, three pairs of calciferous glands in v-vii, each gland sausage-shaped, vertically placed alongside gut, with a short duct from dorsal end opening into esophagus just lateral to the supra-esophageal vessel, in- testinal origin in region of xix-xx, with a lamelliform typhlosole, without caeca and supra-intestinal glands. Vascular system, with a single dorsal trunk aborted in front of hearts of iv, complete ventral and subneural trunks, the latter adherent to parietes, a supra-esophageal trunk in v-xiv, paired latero-esophageal trunks 1 From research financed by the National Science Foundation. 2 Zoology Department, University of Maine, Orono. 2 BREVIORA No. 356 in iii-ix with connectives to supra-esophageal in v-vii. Hearts, in iv-vii lateral, in viii latero-esophageal. Nephridia, holoic and vesiculate. Nephropores, obvious, in a regular longitudinal rank on each side in region of CD. Pigment, none. Septa, present from 3/4. Prostomium, none, replaced by a protrusible proboscis. Setae, eight per segment, in regular longitudinal ranks. Quadrithecal, spermathecae adiverticulate, pores in region of CD, at 5/6-6/7. Type species, E. montana n. sp. Distribution. Puerto Rico. Estherella montana sp. nov. Puerto Rico, El Yunque Mountain. Wet cloud forest at ± 2500 feet. May 1938. 3-0-0. P. J. Darlington per G. E. Pickford. (Mus. Comp. Zool.) About 2200 feet, an anterior fragment, C. W. Richmond & L. Stejneger. (U. S. Nad. Mus.) External characteristics. Length, 50-60 mm (juveniles), 175 mm (posterior amputee of 148 segments). Diameter, 4 mm (juveniles), 12 mm (amputee). Segments of 54 mm juvenile, 188. Color, white (long alcoholic preservation). Peristomium, much shorter than ii but of about the same appearance externally as subsequent segments. Intersegmental furrows, distinct. Seg- mental length, gradually increasing posteriorly to region of vii-ix. Secondary annulation, a presetal and a postsetal secondary furrow in each of ix-xxvi, postsetal secondaries unrecognized behind xxvi. Nephropores, obvious, present from ii, well behind inter- segmental furrows but usually less than half way toward segmental equators, at or near C. Setae, paired, ventral couples first certainly recognizable in v, lateral couples in region of xv, in region of xxx CD slightly < AB much < AA < BC, posteriorly A A ca. - BC, still further back AA ^ BC. Dorsal pores, none. Quadrithecal, spermathecal pores, minute, superficial, in CD, at 5/6-6/7. Other genital apertures, unrecognizable. Genital tumescences, transversely and shortly elliptical, indistinctly de- limited, each with two circular areas of epidermal translucence at center of which is a follicle aperture, a,b/xw-\xi\. Region ot AA, rather deeply depressed through xv-xxiv. Internal anatomy. Septa, 4/5-13/14 thickly muscular to muscular, funnel-shaped, large, apices well posteriorly, 14/15 and following septa slightly strengthened by muscular fibers. Septum 3/4, a delicate transparent membrane bearing on its posterior face 1970 NEW EARTHWORM GENUS 3 one pair of nephridia and on its anterior face two other pairs of tubules, with insertion on gut immediately behind gizzard. Pig- ment, if once present in body wall, completely leached by pre- servative. A large, empty canal, ellipitical in cross section, apparently completely circumferential, in anterior portion of body wall in each of i-x. Canal size, decreasing posteriorly. Brain, in ii. Nerve cord sheath, massively muscularized (Fig. 1) anteriorly but so as to leave a greyish translucent line visible at mD and mV in the cord between segmental ganglia (Fig. 2). Buccal cavity, in i dorsally of small juveniles, seemingly pro- vided, though only temporarily, with a suckerlike pad somewhat resembling the withdrawn and depressed condition of the prosto- mium in various megadriles, in ii dorsally with a circular aperture into a tunnel containing a presumably protrusible proboscis 1-2 mm long. Gut from level of intersegmental furrow 2/3 to septum 4/5, ca. 30 mm long, sigmoid, comprising a pharynx (4 mm long), a bulb (5 mm thick dorsoventrally and with a glandular chamber anteriorly), a slender esophagus (14+ mm long) with closely crowded, low longitudinal ridges on its inner wall, a sort of conical crop (5 mm long) with circular ridges on its inner wail, and a powerful gizzard (6 mm long) referable to iii. Calciferous glands, in contact with each other mesially under the gut, in vii ducts longer but concealed by adherence of 7/8 (near apex of its funnel) to the gut. Typhlosole, present from region of xxiv-xxvi, 10 mm high (3 mm, small juvenile), rolled up on itself like a scroll, ending in region of 125th segment (132d of 188). Lateral typhlosoles, not lamelliform, rounded and pro- tuberant ridges in first one or two typhlosolar segments. Ventral blood vessel, high up in coelom and near gut in v-xi at least. Extra-esophageals. interconnected by a transverse vessel just in front of 4/5 and just under the ventral trunk, anteriorly passing up and branching among nephridia associated with 3/4. Supra-esophageal, with a large branch on each side in v and vi that bifurcates, one branch to a calciferous gland near the duct, the other passing down along anterolateral aspect of the gland to an extra-esophageal trunk, connected also with extra-esophageals by a pair of vessels, seemingly on posterior face of 7/8 but mostly within the septum. Hearts, of iv-vii slender and lateral, of viii apparently latero-esophageal — posterior bifurcations to dorsal trunk slender and empty, anterior branches filled with blood and obviously joining the supra-esophageal. 4 BREVIORA No. 356 Nephridial ducts (of anterior segments), passing down through longitudinal muscle layer and then turning forward to cross the circumferential intra-parietal canal, thence anteriorly widened and with more opaque (muscularized?) wall. Spermathecae, rudimentary, adiverticulate, within the longi- tudinal muscle layer. , Remarks. Small juveniles are assumed to be of the same species as the large worm from the same mountain. Rudimentary state of the spermathecae and absence of macro- scopically recognizable gonads, funnels, and seminal vesicles, in- dicate that even the large worm was juvenile though maximum diameter for the species may have been attained. The parietal insertion of the delicate septum 3/4 had become unrecognizable presumably as a result of pinning out the specimen after a longitudinal incision had been made near the mid-dorsal line. Any connection between nephridial ducts and the circumferential intraparietal canals would have been too small to recognize in dissection, and microtome sections were unsatisfactory. The canals were crossed diagonally by delicate fibers (or septa?). Similar canals had been observed at least once before, but records were destroyed during World War II. Abortion of dorsal trunk in front of hearts of iv and posteriorly in that segment was found in each dissected specimen considered herein. That and other characters already mentioned in the generic definition do not need mention again in species descriptions. Photographs of nerve cord sections were provided by Prof. E. Carpenter. E. nemoralis sp. nov. Puerto Rico. Luquillo Forest (Caribbean National Forest), La Mina Recreational Area, at 1800 feet, February 22, 1947., 2 macerated specimens (several younger specimens possibly of the same species, also macerated.) R. Kenk. (U. S. Natl. Mus.) External characteristics. Length, 250 mm. Diameter, 9 mm. Segments, 220 (at 195/196 a tail regenerate with terminal anus). Color, possibly red originally, even after long alcoholic preserva- tion dorsum with a slight reddish tinge, except in regenerate. Nephropores, obvious, present from ii, in CD. Setae, closely paired throughout, AB = CD, AA > BC, DD ca. = VzC, ventral 1970 NEW EARTHWORM GENUS 5 couples of some segments modified (? but genital tumescences not recognized). Clitellum, perhaps represented by a dark brown coloration in xv-xxii, xxiii/eq, which is conspicuously lacking in a small area around each nephropore, no epidermal tumescence recognizable. Ouadrithecal, pores minute, superficial, each at center of a small tubercle at C and at or immediately in front of 5/6-6/7. Female pores, postsetal in AB of xii(?). Internal anatomy. Septa, 4/5-6/7 very thickly muscular, a transparent, funnel-shaped membrane bearing three pairs of nephridia inserted on the gut just behind the gizzard almost im- mediately in front of 4/5 presumably being 3/4, 7/8 lacking or else inserted on parietes over intersegmental furrow 8/9. Calciferous glands, without a central lumen but with a honey- comb appearance in cross sections, each with a small distal appendage. Supra-esophageal, bifurcating posteriorly in xiv, giving off two pairs of vessels, one immediately behind the other and both just in front of the septum in each of v-vii, the posterior of each pair giving ofl" branches to the calciferous gland and ventrally joining the extra-esophageal trunk of it^ side, the anterior vessels passing onto stalks of calciferous glands and down through the glands into the terminal appendages. Extra-esophageal, first visible in region of 3/4 as a result of union of several large vessels, with several branches to each calciferous gland of its side. Subneural trunk, large, zigzag-looped, closed ends of loops visible beyond both sides of the nerve cord, bifurcating just in front of subpharyngeal gang- lion (one specimen) or in region of xvii (one), each branch passing anteriorly in a zigzagged course lateral to the cord but connected with its twin on the opposite side by numerous trans- verse vessels. Ventral trunk, high up in coelom as in E. montana. Hearts, large, two pairs, possibly latero-esophageal and attributable to viii-ix (?). Testis sac (or sacs?) filled with coagulum, surrounding or including hearts belonging in viii (?). Spermathecae, sessile, ducts confined to body wall, ampullae small, protruding only slightly into coelomic cavities of vi and vii. Remarks. Intersegmental furrows, in spite of the maceration, are distinct, and septa 4/5-6/7 are inserted on the parietes directly over intersegmental furrows 4/5-6/7. Thickness of the subneural trunk is greater than that of the nerve cord even in regions of segmental ganglia. 6 BREVIORA No. 356 The cuticle was loose and setae had been pulled out of their follicles. Repetition in the description of characters shared identically with E. montana seems unnecessary. Mention should be made of the fact that a proboscis and circumferential parietal canals were not seen. This species is distinguished from E. montana by absence of the marked muscularity in the nerve cord sheath. Estherella sp. Puerto Rico. Luquillo Forest (Caribbean National Forest), La Mina Recreational Area, 1800 feet, February 22, 1947, 1 macerated specimen. R. Kenk. (U. S. Natl. Mus.) External characteristics. Size, 150 by 7 mm. Nephropores, obvious, present from ii, in CD. Setae, paired throughout (ventral couples modified in some clitellar segments?). Qitellum, xv-xxii (and xxiii?). Tubercula pubertatis, longi- tudinal bands of translucence, just lateral to B, each demarcated laterally by a deep furrow. Female pores, postsetal in /4B of xii(?). Internal anatomy. Male funnels, one pair, iridescent, imbedded in coagulum apparently also containing hearts of viii and possibly in sacs (or a testis sac?). Spermathecal ampullae, spheroidal, without spermatozoal iridescence, slightly protuberant into coelomic cavities from the angles of septal insertions and parietes. Remarks. Setae had been pulled out of their follicles as in the types of nemoralis. Gonads and female funnels were not found. Structure, so far as could be determined, is the same as is shared by the two preceding species. The reason for anticipating a third species is the maturity at a size smaller than is expected for the other two. SYSTEMATICS Cephalization has had little attention from oligochaetologists and perhaps least of all in connection with a family in which one manner of evolutionary modification seemingly had its most ex- tensive, as well as perhaps least appreciated development. Metamerism in oligochaetes sometimes has been said to be homonomous, i.e., similar throughout the body. Typically, the soma is in anteroposterior segments, each of which, at least in earlier stages of evolution, has four pairs of setae, a pair of 1970 NEW EARTHWORM GENUS 7 nephridia. and a section of the gut. Such a segment is demarcated externally from each of its two contiguous neighbors by inter- segmental furrows, circumferential lines where the epidermis is thinnest. Internally, a segment is deUmited by transverse parti- lions, the intersegmental septa. The latter, typically, are in exactly the same anteroposterior levels as the intersegmental furrows. Peristomium and periproct, according to such definitions, are not segments, though for practical purposes are counted as such. Differentiation in a relatively short anterior portion of the soma of special digestive organs such as gizzards, calciferous glands, etc., and localization of gonads are aspects of oligochaete cephali- zation so universal as not even to have been thought to be involved. More usually considered were abortions, as of follicle and nephri- dial anlage, and disappearance of septa and intersegmental fur- rows, i.e., those structures existence of which enables recognition of segments. Slight (or at least seeming) diplacements of septal insertions on the parietes have been known for some time. Recog- nized but recently (Gates, 1943: 92) was a seemingly posterior dislocation of the parietal insertion of septum 9/10, in Pontoscolex corethrurus (Miiller, 1857), to a position over site of inter- segmental furrow 10/11. Failure to recognize that dislocation was responsible in part for assignment of certain organs in systematic descriptions to wrong segments. How many other errors of that sort were made in past characterizations of glossoscolecids remains to be learned. Appearance of deep, secondary, and even tertiary furrows hardly distinguishable from the primary intersegmentals, in association with abortions of setae and nephridia, also has been responsible for errors in determination of organ locations. Such mistakes are unfortunate, as knowledge of exact organ position along the ante- roposterior axis is of first importance for megadrile systematics and phylogeny. Absence of data as to segmental location of the gonads in the Puerto Rico species is not so regrettable as would formerly have been thought since it is now known that "andry" often, and sometimes even "gyny," does vary intragenerically. Fortunately, the worms now under consideration, in spite of the poor condition, are free of external modifications that might lead to wrong determinations of organ locations. Sctal follicles, to be sure, have been aborted in some of the anteriormost seg- ments, but compensation is provided by the nephropores which not only are all present but also are obvious. Secondary and 8 BREVIORA No. 356 tertiary furrowing is lacking, and the anterior segments are clearly demarcated by unmistakable intersegmental furrows. The first two segments have the normal appearance of externally exposed epidermis. No evidence was found for the existence of an actual pre-oral vestibule comprising one or two rudimentary segments no longer exposed regularly to the external environment. Further- more, the unusual condition next to be discussed characterizes every individual of two, or possibly even three, species. Individual abnormality or defective anterior regeneration accordingly cannot be invoked as was contended when a single Panama specimen was described (Gates, 1968) with a similar condition. The powerful gizzard, being in front of a membrane that, be- cause of nephridial relationships, must be regarded as septum 3/4, is unusually anterior. With the single exception of the above- mentioned Panama species, a megadrile gizzard has not hitherto been found in front of segment v and often is further back in the esophagus. Other organs also are too far forward. The last pair of hearts is in viii, whereas the first pair of real hearts usually is in ix. Calciferous glands are present in v only in the Panamanian Thamnodriloides yimkeri Gates, 1968. Lastly, testes are at least two segments in front of where they would normally be expected. Accordingly, much more is involved than forward displacement (homoeosis) of a single organ but rather a condition in which all organs of the cephalic region from gizzard posteriorly are three segments in front of their expected positions (regional homoeosis). An initial or very early stage in an evolutionary development that may have reached its climax in the Puerto Rico and Panama worms is provided by P. corethrurus. In that species an inter- segmental furrow between the first and second segments has di- appeared along with the prostomium. The now rather flaccid fusion metamere is small. Proof of what happened is provided in many specimens by retention of the setae belonging to ii which are now near the first intersegmental furrow that morphologically is 2/3. With loss of those setae, as in some individuals of the species, the gizzard would have to be referred to v instead of vi. Organs behind the gizzard also would be one segment anterior to their usual position. Abortion of two further segments by the same process under way in P. corethrurus would provide the regional homoeoses of the Puerto Rico and Panama genera. Although body wall and associated nephridia were markedly reduced or deleted, the digestive system was not correspondingly shortened. 1970 NEW EARTHWORM GENUS 9 On the contrary, there has been so much elongation as to require very considerable enlargement of associated septa into posteriorly directed funnel-shapes. What happened in the nervous system may prove to be interesting also. Although homoesoses are identical, the digestive systems and especially structure of the calciferous glands show that the two fore-shortened genera are not closely related. Puerto Rican worms may have evolved from a stock with calciferous glands in viii-x. Genera so characterized are unknown. Related forms should be sought to the south as the glossoscolecids evolved in tropical South America. Puerto Rico now appears to be the northern limit of generic endemism. Completion of a development somewhat hke that now under way in P. corethrurus may be responsible for attribution of testes in Thamnodriliis matapi Righi, 1969, to segments ix and x instead of the expected x and xi. REFERENCES Gates, G. E. 1943. On some American and Oriental earthworms. Ohio Jour. Sci., 43: 97-116. . 1968. On a glossoscolecid earthworm from Panama and its genus. Megadrilogica, 1: 1-15. 10 BREVIORA No. 356 1970 NEW EARTHWORM GENUS 11 LIST OF ILLUSTRATIONS Figure 1. E. montana. Transverse section of nerve cord anteriorly to show muscularization of the sheath. Figure 2. E. montana. Transverse section of nerve cord anteriorly to show the condition responsible for the appearance of a greyish translucent line at mD and mV. Figure 3. E. montana. Transverse section of nerve cord anteriorly to show giant cells ventrally. BREVIORA MiLiiseiiinn of Coimpsirsitive Zoology Cambridge, Mass. 30 November, 1970 Number 357 A review of the fossil Pelomedusidae (Testudines, Pleurodira) of Asia Roger Conant Wood Abstract. The taxonomic status of the three Asiatic chelonian genera that have been described as pelomedusids is reviewed. Of these, "Podoc- neinis" imiica, although possibly a member of the family, is so poorly known that familial assignment is not presently possible. Carteremys leithii and Shwehoemys pilgrimi both appear to be valid palomedusid species. On the basis of new material S. pilgrimi is redescribed and, in addition, a new species of this genus from the Miocene of Baluchistan, S. gaffneyi, is proposed. INTRODUCTION Living pelomedusid turtles are restricted to sub-Saharan Africa, Madagascar, and South America. But paleontological evidence indicates that pelomedusids formerly had a much more cosmo- politan distribution; fossil representatives of this family occur in North and South America, Europe, Africa, and Asia. The purpose of the present paper is to review the extinct Asiatic chelonian genera that have been described as pelomedusids as well as to put on record two new fossil pelomedusid skulls from Asia, one referable to Shwehoemys pilgrimi and the other repre- senting a new species of the same genus. Abbreviations used in this paper are: ANSP — Academy of Natural Sciences, Philadelphia BMNH — British Museum (Natural History) GSI — Geological Survey of India MCZ — Museum of Comparative Zoology, Harvard University I have not seen the material described by Lydekker, Swinton, and Williams that is contained in the collections of the Geological Survey of India, but it is for the most part well figured in the literature. Photographs of the type of Shwehoemys pilgrimi Swinton have been available. 2 BREVIORA No. 357 PREVIOUSLY DESCRIBED ASIATIC PELOMEDUSIDS Very few fossil pelomedusids have been described from any- where in Asia, and none of these is particularly well known on the basis of published material. "Podocnemis*"' indica Lydekker 1887. This species was de- scribed on the basis of a fairly complete carapace and on some miscellaneous plastral fragments found at Nila in the Salt Range of West Pakistan. Lydekker was uncertain about the age of "P". indica but concluded (1887: 59) that it was probably of ". . . low- est eocene . . . and may . . . correspond to the Cemaysian stage of Reims, and the Puerco group of the United States." Since it is now generally recognized that the Cernay beds are of late Paleocene age and the Puerco beds of early Paleocene age, "P." indica may actually be of Paleocene rather than Eocene age. Both Paleocene and early Eocene deposits occur in the Tertiary sequence of the Salt Range (Krishnan, 1960: 494), and too much uncer- tainty exists as to the exact stratigraphic horizon from which the only known specimen of "P." indica was obtained to resolve the question of its age beyond all doubt. Invertebrates found in association with the two chelonians {"Podocnemis" and Hemi- chelys) described from this locality ". . . indicate that the bed in which they were found is either of marine or estuarine origin; and this is confirmed by the chelonians, one of which is covered with the 'spaf of oysters, while sharks' teeth are embedded in the matrix of the other" (Lydekker, 1887; 59-60). Preservation of the carapace of "P." indica is rather poor; Lydekker noted (1887: 60) that ". . . both specimens [from Nila] . . . were in a much broken condition, and . . . required all [the preparator's] skill to render them fit for description." Lydekker's restoration of the carapace (1887, plate 13) indicates that its most distinctive feature is a peak along the midline of the neurals, with the pleurals on either side apparently sloping away flatly rather than on a curve. The nuchal bone appears to be unusual in that its postero-lateral sides are considerably longer than its antero-lateral ones. The nuchal is rather small and does not transgress the lateral boundaries of the first vertebral scute. There are seven neurals; the first five are all longer than broad, whereas the last two are broader than long. Two pleurals (the seventh and eighth) meet in the midline between the last neural and the suprapygal. No indentation occurs at the midline along 1970 ASIATIC FOSSIL PELOMEDUSIDAE 3 the anterior margin of the carapace and a cervical scute^ is lacking. Whereas the first vertebral is nearly twice as wide as it is long, the second and third vertebrals are both longer than broad. Lydekker estimated (1887: 63) that the overall length of the carapace must have been approximately 35 inches (87 centi- meters), exceptionally large for a fossil pelomedusid. Critical taxonomic evidence, such as whether or not the pelvis was fused to the shell, the presence or absence of mesoplastra, and the scute pattern on the anterior plastral lobe, is not preserved. Conse- quently, there is no assurance that "P." indica is actually a pelo- medusid, let alone a member of the genus Podocnemis. Neverthe- less, its midline ridge and straight loping sides are somewhat remi- niscent of the carapace structure of a recently discovered shell of Shweboemys from Egypt (see p. 00), and the number, shape, and arrangement of its neurals is typical of many pelomedusids, as is its lack of a cervical scute. Therefore, it does not seem altogether unlikely that "P." indica may represent the remains of some kind of pelomedusid, although it probably is not a member of the genus Podocnemis. Until more complete specimens of this taxon are discovered, I do not think that a better identification of this speci- men is possible than Pelomedusidae? incertae sedis. Cartereniys leithii (Carter 1852). A second Asiatic pelome- dusid taxon, "Hydraspis" leithii, was recovered from Intertrappean beds near Bombay, India. Some uncertainty exists as to the age of these sediments. Lydekker (1887: 60) regarded them as "lower eocene." Referring to the age of the volcanics within which the Intertrappean beds occur, Wadia (1953: 302) stated that "... it is quite apparent that the Deccan Traps cannot be older than the Danian stage of the uppermost Cretaceous [now Paleocene], while . . . they cannot be much younger than the Eocene." According to Krishnan (1960: 483-486), paleontological evidence afforded by fossils contained within the Intertrappean beds is not par- ticularly helpful for dating, but on other grounds he concluded that the Deccan Traps ranged in age from late Cretaceous to, perhaps, Ohgocene. Robinson (1970: 245) has expressed similar views, stating that the Intertrappeans are ". . . probably mainly early ij have here adopted the term suggested by Zangerl (1969: 315) for this particular scute to avoid the confusing redundancy arising from the conventional procedure of referring to both it and the underlying bone as the nuchal. 4 BREVIORA No. 357 Tertiary in age. . ." Since the Intertrappeans of the Bombay region are confined to the upper part of the Deccan Traps (Pascoe, 1964: 1385), their attribution to the Eocene would not be unrea- sonable. Krishnan (1960: 482), Pascoe (1964: 1385) and Robinson (1970: 245) all agree that the Bombay Intertrappeans were laid down in fresh water, Pascoe further suggesting (1964: 1386) that the depositional environment was a shallow marsh. Carteremys leithii was a small species; the only two complete carapaces for which measurements have been recorded are IVk and 8 inches (18-20 centimeters) in length (Carter, 1852: 187; Williams, 1953: 6). Carter's original description was based on material that is now unfortunately lost (Williams, 1953: 2). Only three other specimens that can be referred to this species with reasonable confidence have subsequently been discovered (Williams, 1953: 6, and plate 3). The total number of neurals is uncertain, although there are at least five. Evidently no cervical scute was present, and the first vertebral was much smaller than the second. The outer surface of the shell is covered with fine sculpturing. In several respects the plastron is quite distinctive: it is relatively narrow, with a semicircular anterior lobe projecting well forward of the front of the carapace; between the broadly rounded xiphiplastral tips is a very shallow anal notch, not com- parable in its shape to those of any pelomedusid with which I am familiar; and the outlines of the pelvic scars on the xiphiplastron are also unusual. In spite of Williams' suggestion (1953: 4) that small, laterally placed mesoplastra, a characteristic pelomedusid feature, may have been present, there is no conclusive evidence bearing on this point. None of the three existing specimens (all belonging to the collections of the Geological Survey of India) is sufficiently complete to demonstrate the presence or absence of these structures (WiUiams, 1953: 6). Trapeziform gulars were widely separated by an extremely broad intergular scute that extended posteriorly to the humero-pectoral sulcus and thus pre- vented the humerals from meeting in the midline also. The most notable feature of the skull is the extensive emargination of the roof from behind, a condition typical of most pelomedusids. A well-developed jugal-quadratojugal bar is present. Although the extremity of the mandibular rostrum was broken off in the material Carter described, it is clear from what was preserved that there must have been a broad, robust symphysis at the mid- line of the lower jaw. 1970 ASIATIC FOSSIL PELOMEDUSIDAE 5 Long regarded as a chelid, "Hydraspis" leithii was eventually redescribed and designated as the type of a new pelomedusid genus by Williams (1953: 3-4). A combination of several characters — the absence of a cervical scute, the relative proportions of the first and second vertebrals, the remarkably large intergular, and the lack of a parieto-squamosal arch combined with the presence of a jugal-quadratojugal bar — led Williams to conclude that this species could not be a cheHd. These characters, together with the inferred existence of small, laterally placed mesoplastra, per- suaded him that Carteremys was in fact a pelomedusid. On the basis of the data presented in Table 1, I agree with Williams' conclusion. If we disregard for a moment the question of whether or not mesoplastra were present in Carteremys, then we see that for all the characters tabulated, this genus and pelomedusids agree. In contrast, none of the other families share more than two characters with Carteremys. This comparison strongly suggests that Carteremys is a pelomedusid. Should it eventually be possible to determine that Carteremys had mesoplastra, the evidence would overwhelmingly favor this determination. Lydekker (1890: 22-23, fig. 2) referred an epiplastral frag- ment, probably from Intertrappean beds and therefore possibly of Eocene age, to "Hydraspis" leithii. Differing from Carteremys in size and in gular-intergular proportions, this specimen may well represent an unknown taxon, but too little is known of it to permit useful discussion. Shweboemys pilgrimi Swinton 1939. The type, and until now only known specimen, of this species is a partial skull of Pliocene or Pleistocene age from Burma (Swinton, 1939). Swinton be- lieved that nasal bones, although not preserved on the specimen he described, must have been present originally. Were this sup- position true, his placement of the genus in the Pelomedusidae would be suspect because one of the diagnostic characters of the family is the absence of nasals (cf. Romer, 1956: 515). Evidently Swinton was either unaware of the significance of this character or else he did not consider it to be of great importance; at any rate, he did not discuss its bearing on taxonomy. Nevertheless, other observations led him to conclude (1939: 551) that Slnveboetnys is a pelomedusid: 'in brief, the interest of the speci- men is that in superior and lateral aspect there is little to dis- tinguish it from the genus Podocnemis, while in palatal view it has much similarity to Stereogenys. There is no doubt that it BREVIORA No. 357 CD JS 2 _>. ^ ^ "3 o 13 o tn 1> "3 S3 to c 3 3 c >. 3 3 '% ** a 4> Vi U .52 JO CA 6 :§ ."2 a "3 v> "iJ M-4 1 *-• <1> J= § >> o c Vi u O c o a a ^ ■^ B o 4-* c o Cu , >. cd ^ i-i «> c o >> 3 o c o c & O c O e 3 en CO C/2 o o "3 o n J=! c O <-> .^^ g J2 -4— ► 2^ o 6 5 9 1 13 k. ■^ cn O c v> «5 (« 4> o «u U u u 1) _4> li 5 >. >. >» >> 'C a 1) cs a, c o< a o u o c/^ J3 o 1) ^.» •** ss -s: J3 .c ^ GO •** 1^ •... _a Q •5 o a '3 09 U >> >> 1 1 tj o c 3 •< ^ *« tn b 3 <3 e U eo 5 "a o O a. B II * •> u t3 w at) c a u o k- a II O :0 •c & o a O 5 >. eu k. 11 6 o _2 II •T3 77^ CL, • Vk C X "rt OJ ca ^ u e ' ^ _« u cd 'C >. II '4—* cd > Xi ;s. a 13 (/5 s II o n3 o oj a. TJ _o t« u ^ _C ■3 c c o a 3 C/5 "3 •^ Xi o (N c« t-l II 1 ro l-i o o • '> C3 Z a II s XI >-i 03 < — E ,*- ."s 13 X II -2? o § B t4-( 'S. 3 C o 60 II 3 s tC o 0 O k. X a. ■2 _d ^ .IT 'u ^ S 03 -4-t en C C/3 o __; E |o C/5 a >. E 3 "3 II 1^ k> u cu X Si 0 3 "u < 4-* 13 kH ca 3 _o 3 O 0 i-i 3 C a o 4-( o X 3 00 C/3 -o ^-1 X « E "o II .52 'a II k. C/5 Uh o C^' 1970 ASIATIC FOSSIL PELOMEDUSIDAE 15 Scute sulci can be detected on the roof of the British Museum skull (Fig. 3). The anterior border of the interparietal overlapped slightly onto the postero-lateral corners of the frontal bones and was bow-shaped, more like that of Podocnemis sextiiberculata {cj. Siebenrock, 1902, fig. 9) than any other pelomedusid known to me. Behind the orbit a scute intervened between the maxillary and frontal scutes. Whether this was an enlarged masseteric scute, as in Podocnemis diimeriliana, or a subocular scute, as in most of the other species of Podocnemis (Williams, 1954), is uncertain. Shweboemys gaffneyi sp. nov.^ (Plates MB, IIIB, IVB) Type. BMNH — R. 8570, a partial skull lacking the premaxillae, most of the prefrontals, the bones of the cheek region, the quad- rates and squamosals, the supraoccipital crest, and whatever supratemporal roofing there may have been. Hypodigm. The type, only. Horizon and locality. Early Miocene, Bugti Hills, Baluchistan, West Pakistan. No locahty or stratigraphic data are associated with this speci- men, but apparently it was obtained by C. Forster-Cooper during one of his two expeditions to the area around Dera Bugti during the years 1910 and 1911 and would therefore be of the age and from the general region cited above. It was subsequently given to Professor D.M.S. Watson for description, who in turn passed it on to Dr. E. E. Williams for the same purpose. Other commit- ments having prevented either from formally describing this interesting skull, the task has now fallen to me. Pilgrim (1908: 144) briefly mentioned that chelonian remains are not uncommon in the Miocene sediments of the Bugti region. Unfortunately, none of these have ever been described. Knowl- edge of their existence nevertheless reinforces the probabihty that the skull under consideration is of the age of and from the locality given above. Diagnosis. Differing from S. pilgrimi in : interorbital width less than diameter of orbits; tomial margins of upper jaw (so far as preserved) horizontal, not notched; medial borders of palatine 1 The species is named for Dr. Eugene Gaffney in recognition of his work on pelomedusid turtles. 16 BREVIORA No. 357 flanges curving away from midline; breadth between postero- lateral corners of palatines equal to 50 per cent of skull length from snout to occipital condyle; broad contact between pterygoids at midline; precondylar fossa semicircular; foramen for trigeminal nerve situated relatively higher on ascending wall of brain case and directed laterally. Differing from S. antiqua in: lack of fore- head groove; orbits round rather than oval; no median notch in upper jaw; broad contact between pterygoids at midline; size much greater. In total length, this skull is slightly longer than the smaller of the two specimens of Shweboemys pilgrimi (10.5 versus 10.2 centimeters for the distance from the snout to the occipital condyle). The width at the postero-lateral corners of the palatines is significantly greater, however, indicating that the skull of S. gaffneyi was proportionately wider. Compared to S. pilgrimi, the orbits of the new species are also proportionately much larger. The increase has been achieved primarily by dorsal emargination of the external face of the maxilla. The diameter of the orbits in S. pilgrimi is essentially the same as the distance across the maxilla from the base of the orbit to the tomium. In contrast, the diameter cf the orbits in S. gaffneyi is nearly twice as great. Because the premaxillae are missing, it is impossible to reconstruct the shape of the upper jaw at the midline. There may have been some kind of median notch, but if so it must have been relatively small and thus quite unlike that of S. pilgrimi (see Fig. 2B). The two Asiatic species differ somewhat in the structure of their secondary palates. In S. gaffneyi, the secondary palate is broader than it is long, whereas in S. pilgrimi this region is slightly longer than wide (cf. Figs. lA and B). For analyzing proportional dif- ferences, direct comparisons may be made between comparable measurements of the two British Museum skulls of Shweboemys, each representing one of the Asiatic species, since they are both of nearly the same length and are equally well preserved in the region under consideration. At the point where the sutures be- tween the maxillae and the palatines reach the outermost extent of the secondary palates, the distance across the secondary palate of the type of S. gaffneyi is 7.2 centimeters, whereas this same distance in BMNH — R. 8432 is 6.1 centimeters. But the length of this structure is 5.3 centimeters in the former and 6.7 in the latter. Another difference, of unknown biological significance, involves the pitting on the ventral surface of the secondary palate. 1970 ASIATIC FOSSIL PELOMEDUSIDAE 17 In both species, these pits appear to be more abundant on the maxillae than on the palatines. The maxillary pits, however, seem to be deeper, larger in diameter, and fewer in number in 5. gaffneyi (see Plate II). No scute sulci are clearly discernible on the preserved part of the skull roof of 5. gaffneyi. Linear depressions, which might be interpreted as scute furrows, follow the courses of the fronto- parietal and fronto-postorbital bone sutures. But a similar groove running antero-posteriorly along the midline does not have a homologue in any other pelomedusid. Thus I am dubious that any of these indentations necessarily corresponds in position to the actual boundaries between scutes. The position of the trigeminal nerve foramina and the extent to which the pterygoids meet on the ventral surface of the skull in 5. gaffneyi are typically pelomedusid and clearly serve to differenti- ate this species from 5. pilgrimi, which is specialized in these characters. The skull of S. gaffneyi exhibits no features that would bar the species from the ancestry of S. pilgrimi. The skull of S. gaffneyi differs from that of 5. antiqua in a num- ber of respects, of which a few may be mentioned here, pending detailed description of the Fayum species. Size is the most obvious (although not necessarily the most significant taxonomically ) dif- ference between the two: the distance from the snout to the occipital condyle in S. gaffneyi is more than one and one-half times longer than in 5. antiqua ( 10.5 versus 6.6 centimeters). The orbits of S. gaffneyi are round and directed forward, those of S. antiqua are oval and face laterally. There is a forehead groove between the orbits of the latter but not of the former. As in S. pilgrimi, and in strong contrast to 5. gaffneyi, there is little or no contact between the pterygoids at the midline in 5. antiqua. DISCUSSION Of the small number of Asiatic fossil pelomedusids hitherto described, one, "Podocnemis" indica, is not well enough known at the present time to merit formal taxonomic assignment. How- ever, Carteremys probably was a pelomedusid, and Shwebuemys certainly was. The relationships of Carteremys within the Pelomedusidae are uncertain. On the basis of skull structure, Shweboemys appears to have been more closely related to Stereogenys than to any other 18 BREVIORA No. 357 pelomedusid. Within the genus, Shweboemys pilgrimi may well have been derived more or less directly from 5. gaffneyi, and there is no reason why this latter species could not in turn have been directly descended from S. antiqua. Occurrences of this genus are separated by such great distances and represented by such a paucity of material, however, that future discoveries may reveal that this interpretation is overly simplistic. Shweboemys is the only pelomedusid genus so far known to have established a successful, enduring lineage outside of Africa or South America. The Shweboemys lineage appears to have been restricted to southern Asia and Africa; no pelomedusid has thus far been reported, even questionably, from central or eastern Asia. The localities where the two Asiatic species of this genus occur are very widely separated geographically and no representative has been recorded from the intervening (and considerably younger) Siwalik deposits of India. Nothing, however, has been published on Siwalik fossil turtles for more than three-quarters of a century. Those described by Lydekker (1885), apart from the large testudinids, show relationships to the recent turtle fauna of India, but available collections need to be examined thoroughly with an eye to the possible presence of pelomedusid remains. Some inferences are possible concerning the ecology of the two Asiatic species of Shweboemys: the Irrawaddy beds are fluviatile in origin (Krishnan, 1960: 498), so that this species was evidently not a marine form.^ In view of the fact that all pelomedusids (except for one or possibly two undescribed fossil genera from Africa) are aquatic, S. pilgrimi was probably a freshwater rather than a terrestrial turtle. Pilgrim (1908: 159) referred to the Bugti beds from which S. gaffneyi was presumably recovered as a '^freshwater formation" and Krishnan (1960: 492) regarded them as being fluviatile, so that this species was in all likelihood also a freshwater rather than a marine form. Moreover, the specialized palatal structures of these two species have definite im- plications regarding their feeding habits. Like some living triony- chids with enlarged secondary palates, their diet may have con- sisted largely or perhaps even exclusively of molluscs of one sort or another. 1 1 am suggesting elsewhere that the pelomedusids were of marine origin. 1970 ASIATIC FOSSIL PELOMEDUSIDAE 19 ACKNOWLEDGMENTS I am particularly grateful to Mr. C. A. Walker of the British Museum (Natural History) for bringing to my attention the skull of Shweboemys pilgrimi belonging to that institution. To the authorities of the British Museum (Natural History) I am indebted for permission to describe this specimen as well as the type of S. gaffneyi. I am also much obliged to Professor Bryan Patterson and Dr. E. E. Williams for critically reading this manuscript, to Professor B. Kummel for help concerning the stratigraphy of West Pakistan and Burma, and to Mr. A. D. Lewis for his skillful preparation work on both of the British Museum skulls. The photographs of the type of 5. pilgrimi reproduced in Plate 1 were sent to Dr. E. E. Williams by the authorities of the Geological Survey of India. The figures were drawn by Mr. Laszlo Meszoly. Finally, I would like to express my appreciation to the National Geographic Society for their generous financial support of my research on pelomedusid turtles. LITERATURE CITED Andrews, C. W. 1901. Preliminary note on some recently discovered extinct vertebrates from Egypt. (Part II.) Geol. Mag., 8: 436-444. . 1906. A descriptive catalog of the Tertiary Vertebrata of the Fayum, Egypt. Brit. Mus. (Nat. Hist.), London. 324 pp. Carter, H. J. 1852. Geology of the Island of Bombay. Jour. Bombay Branch Roy. Asiatic Soc, 4, 16: 161-215. Colbert, E. H. 1938. Fossil mammals from Burma in the American Museum of Natural History. Bull. American Mus. Nat. Hist. 74, art. 6: 255-436. 1943. Research on early man in Burma, Part III: Pleistocene vertebrates collected in Burma by the American Southeast Asiatic expedition. Trans. American Phil. Soc, n.s., 32, part 3: 395-429. Gaffney, E. S.. and R. Zangerl. 1968. A revision of the chelonian genus Bothremys (Pleurodira: Pelomedusidae), Fieldiana (Geol.), 16 (7): 193-239. Krishnan, M. S. 1960. Geology of India and Burma (4th ed.). Madras. Higgenbothams (Private) Ltd. 604 pp. Lydekker, R. 1885. Indian Tertiary and post-Tertiary Vertebrata. Siwalik and Narbada Chelonia. Mem. Geol. Surv. India, Palaeont. Indica, sen 10, 3, part 6: 155-208, pis. 18-27. 1887. Indian Tertiary and post-Tertiary Vertebrata. Eocene Chelonia from the Salt Range. Mem. Geol. Surv. India, Palaeont. Indica, ser. 10, 4, part 3: 59-65, pis. 12-13. 20 BREVIORA No. 357 1890. Note on certain vertebrate remains from the Nagpur District. Records Geol. Surv. India, 23: 20-24. Pascoe, E. H. 1964. A manual of the geology of India and Burma, 3 (3rd ed.). Geol. Surv. India: XXIV and pp. 1345-2130. Pilgrim, G. E. 1908. The Tertiary and post-Tertiary freshwater de- posits of Baluchistan and Sind, with notices of new vertebrates. Records Geol. Surv. India, 37, part 2: 139-166, pis. 2-4. Robinson, P. L. 1970. The Indian Gondwana formations — a review. l.U.G.S. First Symposium on Gondwana Stratigraphy: 201-268. RoMER, A. S. 1956. Osteology of the Reptiles. Chicago, The University of Chicago Press. 772 pp. SiEBENROCK, F. 1902. Zur systematik der schildkroten-gattung Podoc- nemis Wagl. Sitz. der kaiserl. Akad. der Wiss. in Wien, Math.-Naturw. Classe, 111, abt. 1: 1-14. Stamp, L. D. 1922. An outline of the Tertiary geology of Burma. Geol. Mag., 59 (11): 481-501. SwiNTON, W. E. 1939. A new fossil fresh-water tortoise from Burma. Rec. Geol. Surv. India, 74, pt. 4: 548-551. Wadia, D. N. 1953. Geology of India (3rd ed.). London, Macmillan and Co. 531 pp. Williams, E. E. 1953. Fossils and the distribution of chelyid turtles, 1. "Hydraspis leithii" (Carter) in the Eocene of India is a pelomedusid. Breviora, no. 13: 1-8, pis. 1-3. ^__ 1954. A key and description of the living species of the genus Podocnemis {sensii Boulenger) (Testudines, Pelomedusidae). Bull. Mus. Comp. Zool., Ill, (8): 279-295. Zangerl, R. 1969. The turtle shell. In Biology of the Reptilia, /. Morphology A (C. Gans, ed.) London and New York, Academic Press. 311-339. 970 ASIATIC FOSSIL PELOMEDUSIDAE 21 -'-r^np^ .^i*^^*^"^' »>• B I. Type specimen oi SJnveboemys pilgrimi (GSI 17255): A — palatal view of skull; B — dorsal view. Approximately Va natural size. 22 BREVIORA No. 357 ! ■ 1 11 Hj CM ■ ■1 ^^^^1 • •'\^' " A *^"^^^^§L^^- ^'' -«<■•■ >^^o|^^3H|^t^ k b "** "*** ^^BSiBt ilraMI^T'^ ^^^^s »^«w ■J^ ■J^^^Pk '^^^^^^m^B^^-, «-v» --t^^t "•» '^^H ' II. Palatal views of: A — Shweboemys pilgrimi (BMNH- R. 8432); "R— Shweboemys gafjneyi (BMNH— R. 8570). 1970 ASIATIC FOSSIL PELOMEDUSIDAE 23 III. Lateral views of: A — Shweboemys pilgrimi (BMNH- R. 8432); B— Shweboemys gaffneyi (BMNH— R. 8570). tgf trieeminal nerve foramen. 24 BREVIORA No. 357 CM IV. Dorsal views of: A — Siiweboemys pilgiimi (BMNH — R. 8432); B—Shweboemys gaffneyi (BMNH— R. 8570). BREVIORA Miaseium of Comparative Zoology Cambridge, Mass. 30 November, 1970 Number 358 South American anoles: Anolis apollinoris Boulenger 1919, a relative of A. biporcatus Wiegmann (Sauria, Iguanidae) Ernest E. Williams Abstract. Anolis apollinaris is a central Andean derivative of A. bipor- catus, probably from an earlier invasion of South America than that which has provided the present Colombian, Ecuadorian, and western Venezuelan populations of the latter species. Anolis apollinaris Boulenger 1919 was described from a unique type, a female, said to come from "near Bogota." The description made no mention of relationships. The next mention of the species was made by Burt and Burt (1931: 255), who referred numerous Colombian specimens in the American Museum to this species. They suggested that the species belonged to the "chrysolepis stock" but also said that their speci- mens closely resembled A. gemmosus of Ecuador with which they believed A . apollinaris "may prove to be identical or subspecifically allied." An examination of the type of apollinaris in the British Museum shows that these statements of relationships are entirely mistaken and that the specimens referred to the species by Burt and Burt — one specimen received in exchange from the American Museum by the Museum of Comparative Zoology — are mis- identified.^ 1 The type of A. gemmosus O'Shaughnessy has also been examined. It is not of chrysolepis stock nor related at all closely to the two species mis- identified by Burt and Burt as A. apollinaris. The affinities of A. gemmosus are with A. fasciatus Boulenger and A. andianus Boulenger. 2 BREVIORA No. 358 A correct judgment on the affinities of A. apollinaris was made by E. R. Dunn in 1944 (p. 25), who at that time reported: "The Instituto de La Salle has a specimen of this lizard (de- scribed from 'near Bogota') from Paime, Cundinamarca, 1038 meters. A number of students have overlooked the statement that this is a large Anolis (type head-body length 106 mm) and mis- applied the name. Thus the "Anolis apollinaris" of Burt and Burt (1921 I sic J, p. 255) is not Boulenger's species but a composite of two smaller species, incomperliis Barbour from Villavicencio and mariaruin Barbour from Medellin^ True apollinaris is allied to solijer of Santa Marta and copei of Central America." The two latter names are now regarded as synonyms of bipor- catus (see Williams, 1966) and it is with this species, which ranges from Mexico to Ecuador, that apollinaris requires comparison. Brother Niceforo Maria of the Instituto La Salle tells me that Dunn's specimen of A. apollinaris was one of many specimens destroyed in a fire at the Institute in 1948. Fortunately, a number of previously unreported specimens have been discovered, one in the Institut Royale (Brussels), a series in the Zoologische Staat- sammlung (Munich) and three, indeed, in more recent collections of the Instituto La Salle (ILS), and two more in the American Museum of Natural History (AMNH). On the basis of these new specimens and the type specimen at the British Museum (BM), I present a revised standard descrip- tion of the species: Anolis apollinaris Boulenger Type. BMNH 1919.3.6.7 (1946.8-13.22), from near Bogota, Cundinamarca, Colombia. Referred specimens. (All Colombia.) Antioquia (all Cauca Valley): AMNH 38725, Sabanalarga; ILS 81, Puerto Antioquia. Caldas: ILS 101, Pueblo Rico. Cundinamarca: Brussels 3580, La Esperanza, 1250 m; ILS 65, Paime; ILS 106, Quipile; Munich 427-432, San Pablo, west side of cordillera between Bogota and La Dorada. "Western Colombia": AMNH 4844. 1 A. incompertus Barbour is a composite species: specimens from Villavicencio are A. chrysolepis scypheus Cope and, from near Bogota. A. tropidogaster Hallowell. A. mariaruin Barbour is a synonym of A. antonii Boulenger. All types have been examined. 1970 ANOLIS APOLLINARIS u d Z, B Xi o c 3 2 BREVIORA No. 358 Diagnosis. Allied to biporcatus Wiegmann and its subspecies parvauritus Williams but differing in color, in one or no scales separating nasal from rostral, and in a modally higher number of lamellae under phalanges ii and iii of the fourth toe. Head. Head scales small, sharply uni- or tricarinate. Ten to thirteen scales across snout between second canthals. A distinct frontal depression, scales within it not smaller than surrounding scales. Five to nine scales border rostral posteriorly. Circumnasal scale separated from rostral by one small scale or in contact. Six to seven scales between circumnasals dorsally. Supraorbital semicircles separated from each other by 2-4 scales, from the supraocular disk by one row of smaller scales. Supra- ocular disk not very distinct, of 4-12 keeled scales grading laterally into granules. One to three overlapping elongate supraciliary scales, continued posteriorly by granules. Anterior corner of supraocular filled by larger subgranular scales. Canthus sharp, of 6-7 over- lapping scales, the first and second or second and third the largest. Five to seven loreal rows, subequal or the uppermost largest. Tem- poral scales granular. A distinct double line of enlarged inter- temporal scales. Supratemporals granular, slightly smaller than Figure 2. Anolis apollinaris Munich No. 422. Dorsal view of head. 1970 ANOLIS APOLLINARIS 5 temporals. Scales surrounding interparietal moderately to abruptly enlarged, swollen, largest anteriorly and laterally. Interparietal less than or greater than ear, separated from semicircles by 3-4 scales on each side. Suboculars separated from supralabials by one row of scales (or narrowly in contact), anteriorly separated from canthal ridge by one scale, posteriorly continued by an indistinct double row of smaller scales. Seven to eight supralabials to center of eye. Mental slightly wider than long, in contact with 4-8 scales be- tween supralabials posteriorly. Sublabials not well differentiated. Central throat scales quadrangular, swollen, gradually increasing in size laterally. Dewlap. Dewlap in male large with close-packed scales. A gular fold only in female, moderate, scales rather closely packed. Trunk. Middorsals slightly enlarged, swollen, keeled. Dorsal and flank scales keeled, subequal. Ventrals larger, weakly keeled, imbricate, not mucronate. Limbs. Largest fore and hind limb scales strongly unicarinate, except at knee and elbow, smaller than largest ventrals. Supra- digital scales multicarinate. Twenty-four to twenty-seven lamellae under phalanges ii and iii of fourth toe. Tail. Slightly compressed, almost evenly scaled all round. Ver- ticils indistinct. All scales keeled. Enlarged postanals in male. Size. Type: 106 mm snout-vent length. Comparison. Table 1 lists the major features differentiating A. apollinaris and A. biporcatus. I comment on each of these features below: 1. Scales in narial area. The exact pattern of the scales sur- rounding the naris and their relation to the rostral have been repeatedly used in lizards generally (e.g., geckos), and this pattern has also proved empirically very useful at the species level in Anclis. Published examples of the utility of this character at the species level are Ruibal and Williams (1961 ) and Lazell (1964). Although, like all squamation characters in Anolis, these patterns are subject to some intraspecies variation, they are rather surpris- ingly constant. Frequently a single pattern is consistently main- tained; this is usually a simple one. More complex patterns tend to greater variation but the variations are readily derivable from the modal condition (Fig. 3). The pattern of apollinaris is always simpler than that of biporcatus and may be more primitive. (The judgment that this pattern may be more primitive is based not upon BREVIORA No. 358 Figure 3. Nasal rostral relationships. Upper left: Anolis biporcatus biporcatiis, MCZ 15426. Upper right: A. b. parvaiiritus, MCZ 78942. Lower left: A. apollinaris, Munich 422. Lower right: A. fraseri, MCZ 43772. its simplicity but upon its association with other characters re- garded as primitive and on its occurrence in species regarded on other grounds as primitive.) The variation in apollinaris is in the direction of the pattern in biporcatus, but there is no overlap. This is a sharp and clear distinction. 2. Supraciliary scales (Fig. 4). Again the pattern of this area tends to be species specific and again the condition in apollinaris tends to be simpler and perhaps more primitive than that of bipor- catus. The common pattern in biporcatus is indeed unusual (though not unique). Two patterns are common for the supracihary region of anoles: (1) one or more elongate supraciliaries followed by un- differentiated granules; (2) one or more elongate supraciliaries followed by a double series of enlarged scales rather even in size. 970 ANOLIS APOLLINARIS Figure 4. Supraciliary area. Top: Anolis apoUinaris, Munich 422. Lower left: A. b. biporcatiis, MCZ 15426. Lower right: A. fraseri, MCZ 43772. A. biporcatiis, exhibiting several rather short supraciharies grading into large scales that tend to grade again into granules, presents a condition hardly more frequent than that of A . fraseri with its series of quadrate scales along the whole supraciliary margin, the first of these sometimes elongate. 3. Scales around interparietal. In general, the degree of enlargement of scales around the interparietal is a good specific character in Anolis. Particularly valuable may be the degree of enlargement of the scales posterior to the interparietal as compared with the adjacent dorsal or supratemporal scales. There may be rapid intergradation of enlarged scales lateral to the interparietal into much smaller dorsal and supratemporal scales, or the scales behind the interparietal may be sharply and conspicuously larger than dorsal or supratemporal scales (as in some apoUinaris). The two subspecies of A. biporcatus difl'er in this regard. A. apoUinaris is variable; perhaps the variation is geographic, but there is not enough material to say. 4. Ear shape and position. The ear of apoUinaris is quite dif- ferent from that of either southern or northern biporcatus. It is 8 BREVIORA No. 358 closer in size to that of southern biporcatus but quite distinct in its obliquity, a rather unusual feature. 5. Ventral keeling. This character is minor. Keeled ventrals may be a good species character, but there are many instances of intraspecies variability, both geographic and (typically qualitatively less extreme) at a single locality. A. apollinaris has the ventrals more weakly keeled than either subspecies of biporcatus. 6. Toe lamellae. The number of toe lamellae is an extremely useful character in Anolis and very characteristic of species. It is, however, subject to variability (a range of 6 or 7 is quite usual) and overlap is, as in the present case, frequent. A. apollinaris tends to a higher number of toe lamellae than either subspecies of biporcatus. 7. Color. I have no descriptions of color in life of apollinaris and the varying colors of biporcatus as preserved (it is uniform green in life) do not make comparison very easy. Boulenger de- scribed the type female as "Dark olive above and on the sides, with a fine blackish network, head and a vertebral band pale, the latter with narrow transverse processes; small round light spots on the sides and tail; forearm, tibia and lower parts pale green." The Brussels specimen has preserved its pattern rather well. Description follows: Head greyish. A dark streak from back of eye to shoulder, there merging with dark flanks. Below this, labials and nape lighter, their color continuous with the smudged grey of the throat. A hght brown middorsal zone, irregularly darker lat- erally, bordered on each side by a narrow grey line. Flanks dark brown with indications of white spots or broken narrow vertical white bars. Forelimbs obscurely annulate, hind limbs boldly so. Tail above with longitudinally oval light spots with irregularly dark centers. Belly lighter than any part of dorsum but still heavily infuscated. Tail below lighter still. The new specimens resemble the Brussels specimen in head coloration, as the British Museum type now does; presumably the latter differs from Boulenger's description as a result of change during preservation. A. apollinaris, when compared with A. biporcatus, differs in few and superficially trivial ways. The ventrals are less strongly keeled. There are fewer scales between nostril and rostral: one or more rather than two or three. The color is quite unlike anything 1 have seen in biporcatus. I am neither confident that this form is a full 1970 ANOLIS APOLLINARIS 9 species nor convinced that it is not. It is certainly not to be con- fused with Mexican or Central American biporcatus, nor with the south Colombian-Ecuadorian population, nor does it resemble the two Venezuelan specimens of that species. Whether it is distinct from, or intergrades with, some of the other Colombian or the Ecuadorian populations is a matter for future discovery. Provi- sionally, since it was described as a species, it may remain so allocated. A. biporcatus has the distribution of a recent immigrant into South America, extending from Panama through the Choco region of Colombia down into Ecuador west of the Andes and extending eastward through the Santa Marta Mountains into western Vene- zuela. A. apollinaris, now recorded from Antioquia, Caldas, and Cundinamarca, may be supposed to have arisen from A. bipor- catus by isolation and subsequent minor modification in the central Andean regions of Colombia. Since it has some primitive features (e.g., nasal-rostral relationship) and since it has reached species status, it may represent an earlier invasion of South America than that which resulted m A. b. parvauritus Williams and the Vene- zuelan specimens oi A. b. biporcatus. REFERENCES BouLENGER, G. A. 1919. Descriptions of two new lizards and a new frog from the Andes of Colombia. Proc. Zool. Soc. London, 1919: 79-80. Burt, C. E., and M. D. Burt. 1931. South American lizards in the col- lection of the American Museum of Natural History. Bull. American Mus. Nat. Hist. 61: 227-395. Dunn, E. R. 1944. Herpetology of the Bogota area. Revista Acad. Co- lombiana Cienc. 6: 68-81. Lazell. J. D., Jr. 1964. The anoles (Sauria, Iguanidae) of the Guade- loupeen Archipelago. Bull. Mus. Comp. Zool. 131: 359-401. RuiBAL. R., AND E. E. Williams. 1961. The taxonomy of the Anolis honwlechis complex of Cuba. Bull. Mus. Comp. Zool. 125: 209-246. Williams, E. E. 1966. South American anoles: Anolis biporcatus and Anolis fraseri (Sauria, Iguanidae) compared. Mus. Comp. Zool., Breviora No. 239: 1-14. (Received 9 June 1970.) 10 BREVIORA No. 358 ■10 O' -:^/0' Figure 5. Map of the distribution of Anolis apollinaris in central Colombia. 1970 ANOLIS APOLLINARIS 11 TABLE 1 apollinaris b. biporcatus scales across snout 9-13 7-11 b. parvauritus 8-13 scales between nasal and rostral 0-1 2-3 2-3 scales between supra- orbital semicircles 2-4 1-4 0-3 supraciliaries 1-2 elongate plus usually 3-4 short as in biporcatus series of small supraciliaries scales of rather plus a series of uniform size small scales of variable size scales behind interparietal variable, slightly abruptly larger to abruptly larger than dorsals than dorsals grading gradually into dorsals scales separating interparietal from semicircles 3-5 3-6 3-7 loreal rows 5-8 5-10 6-9 supralabials to center of eye ear 7-8 8-11 small to moderate moderate to large, vertical 7-12 small ventrals weakly keeled strongly keeled, mucronate strongly keeled, mucronate lamaellae under phalanges ii and iii of fourth toe 25-27 22-26 22-26 BREVIORA MiaseiLaitii of Comparative Zoology Cambridge, Mass. 30 November, 1970 Number 359 The Swimbladder as a Juvenile Organ in Stromateoid Fishes Michael H. Horn^ Abstract. The swimbladder regresses in 14 of the 15 genera of strom- ateoid fishes and apparently in each case before maturity is reached. In one genus the swimbladder is absent. The organ appears to be completely functional and is present in juvenile fishes that inhabit the surface layers of the ocean, often in association with jellyfish medusae or floating objects. In the transition from the juvenile to the adult habitat and mode of life, the swimbladder regresses to a nonfunctional state, and other morphological changes occur. A detailed study in progress of the state and structure of the swimbladder in stromateoid fishes in relation to their evolution and ecology shows that the organ regresses with age in 14 of the 15 genera comprising this suborder of perciform teleosts. The regression of the swimbladder occurs at different ages, depending upon the species in question and seems to be correlated with other morphological changes and with certain changes in mode of life. Thus, the state of the swimbladder may be useful in predicting or determining at what age or size a particular species undergoes a shift in habitat or way of life. The suborder Stromateoidei consists of six families (Haedrich & Horn, 1969), the members of which range in maximum size from about 30 to about 120 centimeters. It is a fairly diverse group of temperate and tropical marine fishes, which, as adults, variously occupy a wide range of depths in coastal and oceanic waters. The Stromateidae occur at all depths over the continental shelf. The Ariommidae are benthic or benthopelagic on the shelf. 1 Present address: Department of Biological Science, California State College, Fullerton, California 92631. 2 BREVIORA No. 359 The Centrolophidae are either coastal, as in Hyperoglyphe and Seriolella, or oceanic at various depths, as in Centrolophiis and Icichthys. The Nomeidae, Tetragonuridae, and Amarsipidae are oceanic usually in epi- or mesopelagic layers. However, as occurs in many other marine fishes, the larvae and juveniles of stroma- teoids are pelagic in the surface layers, mostly in the upper 100 meters. Stromateoid fishes commonly undergo marked changes when approaching maturity, and these changes are often associated with the migration from the surface layers to the deeper layers where the fishes live as adults (Haedrich, 1969). Along with certain changes in body proportions (see Haedrich, 1967; and Horn, 1970), including, frequently, changes in the length of paired fins, the swimbladder regresses. This regression and the significance of the swimbladder in the life of young fishes are discussed below. MATERIALS AND METHODS Specimens examined are from the collections of the British Museum (Natural History); the Zoological Museum, Copen- hagen; and, the Woods Hole Oceanographic Institution. The fishes from the last institution will ultimately be deposited in the Museum of Comparative Zoology, Harvard University. Specimen data will be included in a forthcoming comprehensive paper on stromateoid swimbladders. Swimbladder volumes were calculated treating the bladder as an ellipsoid. Body volumes were determined by displacement in water. Ten per cent was allowed for shrinkage of the preserved material, and volume determinations were made from swimblad- ders that were in most cases well expanded. The number of retial capillaries was estimated from thin sections of swimbladders. RESULTS AND DISCUSSION Very little mention has been made of the swimbladder in stromateoids. Jordan & Evermann (1896) stated that the organ was "usually absent" in the Stromateidae. Fowler (1936) indi- cated that it was "present or absent" in his treatment of several stromateoid genera. Grey (1955) reported it to be absent in Tetragonurus. Haedrich (1967) cited its presence in Ariomma. While I have not yet examined every species, I have found a swimbladder to be present in all genera except Pampus. 1970 SWIMBLADDER OF JUVENILE STROMATEOIDS The swimbladder in stromateoids (Fig. 1) is euphysoclystous, i.e., with a distinct gas-resorbing area — a posterior chamber separated from the anterior by a diaphragm — and has unipolar retia that are either medial or posterior in position. It is generally of a relatively simple type found in a number of perciform fishes (N. B. Marshall, personal communication). Figure 1. Ventral (slightly flattened) view of the swimbladder of Nomeus gronovii, 26.4 mm SL; rm = rete mirabile; rv = retial vein; ra = retial artery; gg = gas gland; ac = anterior chamber; dm = dia- phragm; pc = posterior chamber. (X 33) The swimbladder in these fishes is relatively small and fre- quently below the 3.1-5.7 range of percentage volumes calculated by Alexander (1966) to be necessary for neutral buoyancy in sea water (Table 1). Measurements of swimbladder percentage volume for Hyperoglyphe, Cubiceps, and Ariomma show the organ to be within the above range, but in other genera, notably Schedophiliis and Nomeus, the percentage volume may be as low as 0.5 or 0.6 per cent. Some calculated values may be lower than normal because of the swimbladder being in a slightly con- tracted state. The organ does appear to be completely developed and almost certainly functional. The blood supply is complete, and the surface area of the gas gland and the length of the retia in relation to swimbladder size (Table 1 ) compare closely with the range of values calculated by Marshall (1960: tables 3, 4, & 5) for deep-sea fishes with well-developed swimbladders. The organ is fully formed in the smallest individuals (9 mm SL) that have been examined. Regression to a nonfunctional state occurs, apparently in all species, before maturity is reached. BREVIORA No. 359 ■S ^ q; ^ ^ a 5: * '/^ O «o o \D VO ^H r- »— < (S (M 00 < I- CO O u ■»-> O CJ ^ *::? c/i -J O X 1^ 13 2 ■= 2? -^ ?. ^ -2 •? 2,^ l>3 X C3 O ^ 1/2 > O ■J C 1-1 (U C3 Q E c S §! .5 ;^ ^ W T3 OS o to .~- a a S 5 a o J3 to * f^J — Vi -^ rJ (N o ^ Q tn 05 .Is* ^ i: CO o_ ^Oi r^ r-- r<^ 0 rn a\ rn 00 ■"t •X- q OS -< u-i — <* -- r<-i rn <^ Tf ri ^ ^ ri ^ O d * ^ -^^ '^ d d d d 00 r-- f^ r-; 00 q ON q >r) d (N ^ eN fN — < rn (N ro "=> ^r^ t^ 4- ^ ^ ^ _H ''T d rK ^ d d o< z: - ^ >/-, 0 v£i 10 q >A q ~ 'O ^ 60 c^ !^ u 0 - ^ u ol ■■• 1> u ^ s: ^ CS c^ i> _-o l^ rt "2 5 s ^ -5 «^ 1:! ■^^ B 0 u «v S3 I" 3 C •2i T3 1 Centrolop dophilus n thys lockii eroglyphe , 5 .2 E 0 0 leus grono 'ceps gracii 0 00 2 to 5 E < D. tn >3 "i: *! 1 a. 1 la c .0 1970 SWIMBLADDER OF JUVENILE STROMATEOIDS 5 Being fully developed and certainly capable of hydrostatic ad- justment, the stromateoid swimbladder seems very unlikely to be merely an evolutionary remnant. The clue to its significance appears to lie in the behavior of the juveniles. The young of all or nearly all of the species live in the surface layers and fre- quently in association with jellyfish medusae or other animate or inanimate floating objects. Protection is presumably provided by the medusae or floating objects, and the fishes feed upon the small invertebrates concentrated around the objects or upon the jellyfishes themselves. This existence seems to require both con- siderable maneuverability and the ability to hover and remain motionless in midwater. I have observed young Peprilus triacan- thus hovering beneath and near the tentacles of the sea nettle, Chrysaora qiiinquecinha, and they show little locomotor activity except for backing of water by the pectoral fins. Mansueti (1963) has made similar observations on Peprilus alepidotus {= P. paru.). Few, if any, of the stromateoid genera which consort with medusae are completely immune to jellyfish toxins but merely avoid the tentacles, according to observations made by Mansueti (1963) on P. alepidotus, by Maul (1964) on Mupus {-Schedo- philus), and myself on P. triacanthus. Lane (1960) reports that Nomeus gronovii can survive doses of Physalia toxin as much as ten times that which would kill other fishes of the same general size and type; however, Nomeus still exhibits considerable agility in avoiding the tentacles of Physalia. The swimbladder, even if smaU, would provide a degree of buoyancy, although not necessarily complete neutral buoyancy. During the period of life in which the fishes have a gas bladder, the skeleton is not well ossified and the musculature may not be completely developed; thus, the juveniles probably have a lower specific gravity than the adults. In those adults with soft muscula- ture and light ossification, such as Schedophilus, the swimbladder probably becomes unimportant and uneconomical because of a change in mode of life, even though, as in juveniles, a small volume of gas would provide nearly neutral buoyancy. Using Alexan- der's (1966) formula for calculating the percentage swimbladder volume necessary to achieve hydrostatic equilibrium, I find that only a 1 per cent reduction in specific gravity of the fish lowers the required percentage volume from 3.1 per cent, the lower figure in Alexander's calculated range, to 2.2 per cent. The latter figure is within or near the range of volumes for most of the stromateoid 6 BREVIORA No. 359 genera (Table 1). Also, the more firmly muscled and more heavily ossified fishes, such as Hyperoglyphe, have greater swim- bladder volumes than those with softer muscles and lighter bones, such as Schedophiliis (Table 1 ). An exception is Nomeus, which has relatively firm musculature but a small swimbladder. The regression of the swimbladder is a gradual process, with the sac diminishing and the gas gland becoming a small yellowish mass before being completely resorbed. The mass representing the regressed gas gland may persist in the mesentery beneath the kidney for a considerable period of time after the swimbladder becomes nonfunctional. In none of the stromateoids does the swimbladder appear to become fat-filled upon regression as it does in some deep-sea fishes (Marshall, 1960). It is meaningful to consider the duration that the swimbladder remains functional in the different stromateoid genera in rela- tion to the time of change in habitat and mode of life. Nomeus among the stromateoids appears to have the most intimate and enduring association with jellyfishes, usually with the siphono- phore, Physalia. The fish may remain with Physalia throughout its life, although this is uncertain. Significantly, Nomeus retains what seems to be a functional swimbladder longer than any other stromateoid that has been examined. The largest specimen I have studied (142.7 mm SL) had a relatively large sac with a some- what contracted gas gland and was captured at the surface with Physalia. It has not been possible to determine whether or not the large specimens that have been found with Physalia were mature. The swimbladder of Peprilus triacanthus is usually completely regressed by the time the fish reaches a length of 100 mm SL, and this is about the size at which it has completely abandoned jellyfish medusae. Large individuals (> 100 mm SL) of P. tria- canthus do not hover as do the juveniles but swim continuously. The pecioral fins increase in relative length with age and are used more extensively for propulsion in adults than in juveniles. The angle through which the pectorals are adducted apparently pro- vides lift. (The locomotion and buoyancy of P. triacanthus are being considered in a separate study.) Continuous swimming with some degree of pectoral propulsion is probably the rule in adult stromateoids. In Stromateus fiatola, the swimbladder is greatly regressed at a fish length of 75 mm SL, a size at which the fish seems to have 1970 SWIMBLADDER OF JUVENILE STROMATEOIDS 7 ended its association with medusae (Mansueti, 1963: 60). In the size interval of 75 to 100 mm SL, this species loses the pelvic fins, and its coloration changes from a vertically-banded pattern to a more uniform one in which there are often dorsal spots. Tetragoniirus is a strictly oceanic genus (Grey, 1955), and its swimbladder is considerably regressed when the fish reaches a size of 50 mm SL. The young have been found associated with medusae in the surface waters (Mansueti, 1963: 60). According to Haedrich ( 1967), the adults are probably members of the meso- or bathypelagic fauna. Finally, Pampus, a coastal genus and the one considered by Haedrich (1967) to be the most advanced of the stromateids, evidently has no swimbladder. Whether fishes of this genus as- sociate with medusae as frequently as other stromateids is not known. There are indications that they do not. According to Suyehiro (1942), P. argenteus does feed to a certain extent on jellyfishes in Japanese waters. However, studies by Kuthalingam (1963) and Nath (1966) show that both juveniles and adults of this species off the Indian coast are macroplankton feeders whose diet largely depends upon seasonal changes in abundance of crus- taceans and polychaetes; this may indicate that the young do not regularly associate with medusae. Also, the pectoral fins of Pampus become quite long early in life and do not greatly increase in relative length with age as they do in Peprilus, another stro- mateid, and certain other genera. At a fish size of 30 mm SL, the pectoral length of Pampus argenteus is about 40 per cent of the standard length compared to only about 30 per cent in Pep- rilus paru, which has a very similar body shape. The pectoral length in Peprilus triacanthus of the same size is about 25 per cent of standard length. The relative length of the pectorals has increased to 40 per cent in P. paru and to about 35 per cent in P. triacanthus at a size of 80 mm SL. Thus, assuming that hover- ing beneath objects and possession of a swimbladder are related and that increased pectoral length is important in continuous swim- ming, it seems that members of the genus Pampus acquire the adult mode of locomotion and habit at an earlier stage than most other stromateoids and, in so doing, completely dispense with the swim- bladder. A number of other fishes have regressed, age-dependent swim- bladders. These include such shallow-water marine fishes as cer- tain gobies, blennies, flatfishes, and most muraenid eels (N. B. 8 BREViORA No. 359 Marshall, personal communication ) and also some deep-sea fishes, such as certain species of Cyclothone and Stomias, whose swim- bladders, upon regressing, become invested with fat (Marshall, 1960). As in stromateoids, the regression of the organ in these fishes is probably associated with habitat and/or internal changes. Swimbladder regression is eventually to be the subject of a general review. Yet to be examined is the state of development of the swim- bladder in newly-hatched stromateoid larvae. Since it is fully formed in fishes as small as 9 mm SL, the organ must develop quite early. The swimbladder may be of considerable importance to larvae in orientation and in positioning the body for food- capturing, although it is not known when the larvae begin to feed. Indeed, the swimbladder, as I judge from its particularly early regression in some stromateoids, may have its greatest functional significance in the larvae and smallest juveniles. ACKNOWLEDGMENTS 1 sincerely thank N. B. Marshall and Richard L. Haedrich for reading and ofi'ering suggestions on the manuscript. I am also grateful to Dr. Marshall for valuable advice on swimbladders and for providing space and facilities in the British Museum (Na- tural History). This work was supported by a NATO Postdoctoral Fellowship awarded by the National Science Foundation. LITERATURE CITED Alexander, R. McN. 1966. Physical aspects of swimbladder function. Biol. Rev., 41 (1): 141-176. Fowler, H. W. 1936. The marine fishes of West Africa, based on the collection of the American Museum Congo Expedition, 1909-1915. Part 11. Bull. American Mus. Natur. Hist., 70(2): 607-1493. Grey, M. 1955. The fishes of the genus Tetragoniirus Risso. DANA- Report No. 41: 1-75. Haedrich, R. L. 1967. The stromateoid fishes: systematics and a classi- fication. Bull. Mus. Comp. Zool., 135(2): 31-139. 1969. A new family of aberrant stromateoid fishes from the equatorial Indo-Pacific. DANA-Report No. 76: 1-14. Haedrich, R. L., and M. H. Horn. 1969. A key to the stromateoid fishes. WHOl Tech. Rept. No. 69-70, 46 pp. UNPUBLISHED MANUSCRIPT. 1970 SWIMBLADDER OF JUVENILE STROMATEOIDS 9 Horn, M. H. 1970. Systematics and biology of the stromateid fishes of the genus Peprilii.s. Bull. Mus. Comp. Zool., 140(5): 165-262. Jordan. D. S., and B. W. Evermann. 1896. The fishes of North and Middle America: a descriptive catalogue of the species of fish-like vertebrates found in the waters of Ncrlh America north of the Isthmus of Panama. Part 1. Bull. U. S. Nat. Mus., No. 47: 1-1234. KuTHALiNGAM, M. D. K. 1963. Observations on the fishery and biology of the silver pomfret, Pampiis argenteiis (Euphrasen), from the Bay of Bengal. Indian J. Fish.. 10(1): 59-74. Lane, C. E. 1960. The Portuguese man-of-war. Sci. Amer., 202(3): 158-168. Mansueti, R. 1963. Symbiotic behavior between small fishes and jelly- fishes, with new data on that between the stromateid, Pepriliis alepi- dotus, and the scyphomedusa, Chrysaora qiiinqiiecirrha. Copeia, 1963 (1): 40-80. Marshall, N. B. 1960. Swimbladder structure of deep-sea fishes in rela- tion to their systematics and biology. Discovery Rept., 31: 1-122. Maul, G. E. 1964. Observations on young live Miipus tnaculatus (Giinther) and Miipiis oralis (Valenciennes). Copeia, 1964 (1): 93-97. Nath, p. R. 1966. Biology and seasonal distribution of the pelagic food fishes of Travancore coast. Kerala Univ. Pub., India, 1-140. SuYEHiRO, Y. 1942. A study on the digestive system and feeding habits of fish. Japan. J. Zool., 10 (1): 1-303. BREVIORA Muisenajnini of Comparative Zoology Cambridge, Mass. 30 November, 1970 Number 360 MAMMALS FROM THE EARLY CENOZOIC OF CHUBUT, ARGENTINA George Gaylord Simpson Abstract. Angelocabrenis daptes, new genus and species (Mammalia, Marsupialia, Borhyaenidae), Coelostylodon florentinoameghinoi, new genus and species (Mammalia, Notoungulata, ?Isotemnidae) and Coelostylodon caroloameghinoi, new species, are described from the Casamayor forma- tion, probably early Eocene. Knowledge of upper premolars of Didolodiis (Mammalia, Condylarthra. Didolodontidae) is increased and the status of Acoelodiis (Mammalia, Notoungulata, Acoelodidae) is discussed. The latter genus and the family based on it are essentially indeterminate, and previous usage of the names is unjustified. INTRODUCTION A visit to Mar del Plata, Provincia de Buenos Aires, Argentina, early in 1970 enabled me for the second time to examine parts of the important collections of fossil mammals in the Museo Munici- pal de Ciencias Naturales of that municipality. In collections from the Casamayoran Stage of Chubut, three specimens were found to be of particular interest and to make especially important con- tributions to knowledge. The Director of the Museo, Sr. Galileo J. Scaglia, very kindly permitted me to study those specimens and to publish descriptions and discussions of them, presented here- with. I am again and increasingly indebted to Sr. Scaglia and to the whole staff of the Museo for their courtesy and cooperation. The accompanying illustrations were prepared by RaVae Marsh. In the following, MMP precedes catalogue numbers of the Museo Municipal de Ciencias Naturales de Mar del Plata and MACN those of the Museo Argentino de Ciencias Naturales "Ber- nardino Rivadavia," Buenos Aires. While carrying out the research for this paper I was employed jointly by the Museum of Comparative Zoology and the Univer- sity of Arizona. 2 BREVIORA No. 360 Order Marsupialia lUiger Family Borhyaenidae Ameghino Angelocabrerus, new genus Etymology. For the late Angel Cabrera, a great mammalogist, who, among many other things, wrote an important study of bor- hyaenids. This kind of nomenclature is Ameghinian and is con- sonant with the related Arminiheringia. I have ventured to give the compound an appropriately mascuUne ending. Type-species. Angelocabrerus daptes, new species, infra. Known distribution. Casamayoran, Argentine Patagonia. Diagnosis. Specialized borhyaenines. Canines with closed, rapidly tapering roots; short, heavy, fully enameled crowns. P., one-cusped, heavy, with small distinct talonid. M 1.4 essentially two-cusped, with paraconid anterior and only slightly lingual to the larger protoconid. No trace of metaconid. Talonids un- basined, reduced to very slight, simple ledges. Protoconids and paraconids truncated with wear on M^.g, becoming sharp points on M4 with protoconid a high slender needle. Discussion. As far as known, the dentition is similar to that of the much later (Santacrucian) Borhyaena and by the same token is also similar to the contemporary Casamayoran Arminiheringia. The talonid reduction seems to have gone even further in Angelo- cabrerus than in the other genera. The lower canine is unlike that of Arminiheringia, with a shorter closed root and more fully enameled crown. The way in which M^ wears, quite distinctive from either Arminiheringia or Borhyaena, would seem to imply different occlusion and hence different structure in the unknown upper teeth, M^.^ , with which M^ occluded. This wear is much as in Plesiofelis, considered by Cabrera (1927: 274-278) Des- eadan in age and synonymous with Pharsophorus but almost cer- tainly Mustersan and probably distinct from Pharsophorus. How- ever, in Plesiofelis the molar talonids are considerably less re- duced than in Angelocabrerus. In Arminiheringia auceta, the only adequately known species of its genus, there is a rapid increase in size of the molars pos- teriorly, M4 being about twice as large as M^. The figures (some- how omitted in Simpson, 1948) are here given in Table 1. It is there shown that the increase is much less in Angelocabrerus dap- tes, with M4 only about half again as large as M^. The increase 1970 EARLY CENOZOIC CHUBUT 3 is even less in Borhyaena. As far as the evidence goes, Angeloca- hrerus could be ancestral to Borhyaena, and in that case the rate of evolution in known parts must have been extremely slow. Knowledge of the present genus is too incomplete, however, to warrant a firm conclusion. As in Arminiheringia and Borhyaena but to even more marked degree, Pg is a large and heavy tooth. It here approximates M , in dimensions. It has a single main cusp, with a long anterior and short, nearly vertical posterior slope. There is a minute cuspule at the anterior base. There is a distinct but small, shelflike talonid with a single cuspule. This is absent in Borhyaena and also in the only known specimen of Arminiheringia that might show it, but the latter is so worn that a small talonid could have been present originally. Except for the points already mentioned, the lower molars are like Arminiheringia and Borhyaena in structure. Angelocabrerus daptes, new species. Etymology. Greek daptes, eater, gnawer, from the inferred carnivorous, possibly ossifragous habits of the animals. Holotype. MMP 967M, part of right mandibular ramus with M2.4, left Pg, Ml, and M. probably of the same individual, two lower canines somewhat broken, and small caniniform tooth and tip of another doubtfully associated. Hypodigm. Holotype only. Horizon and locality. Casamayoran, south of Lago Colhue- Huapi, Chubut, Argentina. The specimen was a surface find high in the beds, and derivation from an overlying formation is pos- sible but quite improbable. Diagnosis. Only known species of the genus as diagnosed above. Discussion. The loose left M. has somewhat darker enamel and is slightly less worn than the right M. in the mandibular frag- ment. Its color and wear are more consonant with those of the loose teeth identified as left Pg and M^. However, there can be little serious doubt that those and the two loose lower canines do in fact belong to the same individual as the mandibular fragment. All were found together, they are congruent in size and structure, and they add up to a unique specimen of a group extremely rare in these beds. The two slender caniniform teeth are dubious and I do question whether they belong to the same animal. BREVIORA No. 360 The diagnosis and discussion of the genus and the illustrations make further description unnecessary. Figure 1. Angelocabrerus daptes, new genus and species. Holotype, MMP 967M. Right M2-4. A, buccal view. B, occlusal view. C, lingual view. X 1. Order Condylarthra Cope Family Didolodontidae Scott Didolodus sp. indet. Specimen. MMP 696M, fragment of right maxilla with 2-4 Horizon and locality. Casamayoran of Caiiadon Vaca, tribu- tary to the left (northwest) bank of the Rio Chico, Chubut, Ar- gentina. 1970 EARLY CENOZOIC CHUBUT Figure 2. Angelocabrerus daptes, new genus and species. Holotype, MMP 967M. Left P^ - M,. A, lingual view. B, occlusal view. C, buccal view. XI. Figure 3. Angelocabrerus daptes, new genus and species. Holotype, MMP 967M. Lower canine. X 1. BREVIORA No. 360 Discussion. This specimen is interesting because it shows the coronal structure of P^'^ of Didolodus in relatively little worn condition and because the proportions of these teeth are distinctive. Comparison is mainly with MACN 10690, holotype of Didolo- dus multicuspis Ameghino, the only other specimen of Didolodus known to me that includes P^^. It is figured in Simpson, 1948, text figures 25 and 26 and plate 10, figures 1 and 2. MACN 10738 includes P-, which has not been figured but was included in my description of D. multicuspis (Simpson, 1948: 101). AMNH 2847 is a P'^ referred to D. minor by me (Simpson, 1948: 103) but not separately described or figured. As shown in Table 2, P- and P^ are each shorter than in the holotype by 7 per cent, which does not in itself suggest specific dis- tinction, and P^ has almost the same length in the two specimens. However, all three teeth are more notably narrower in MMP 696M, by 16 per cent, 20 per cent, and 19 per cent for P-, P^, and P^ respectively. A result is that all three teeth are longer relative to their widths in MMP 696M. This is especially notice- Figure 4. Didolodus sp. MMP 696M. Right PS-i, occlusal view. X 3. able in P"^, which is distinctly transverse in MACN 10690 but equidimensional in MMP 696M. P=^ of AMNH 28471, referred to D. minor, is even more transverse than in the holotype of D. multicuspis. Its width slightly exceeds that of P^ in MMP 696M, although its length is decidedly (25 per cent) shorter. The in- dividual represented by MMP 696M was probably closer to D. multicuspis in over-all size, but the differences in some dimensions and in proportions make reference doubtful either to that species or to D. minor. At the same time, they do not warrant definition of a new species, which in any case should preferably not be based on upper premolars, for which there is so little comparative material. 1970 EARLY CENOZOIC CHUBUT 7 Specimens of known origin referable to D. midticiispis without much doubt are all from south of Lago Colhue-Huapi, and those similarly referable to D. minor are from Canadon Vaca. MMP 696M is from Canadon Vaca but probably does not belong to D. minor. Specimens and field data for Didolodus are still far from sufficient for identification of populations within the genus and for determination of their distribution. P- of MMP 696M, unlike the less well-preserved specimens previously known, is seen to have two distinct external cusps, con- nate above the tips but still separated by grooves. These are at least descriptively paracone and metacone, and the metacone is only slightly lower and smaller than the paracone. The tooth is completely surrounded by a cingulum, but this is feeble on the middle of the labial face. The lingual slope from paracone plus metacone to the labial cingulum is sUghtly uneven, but is without really distinct cusps. P^ and P^ also have distinct paracone and metacone but, unusually, the metacone is lower and smaller rela- tive to the paracone progressively from P- to P^. P^ and P^ have well-developed protocones, and the cingula do not cross their lin- gual faces. A distinct cingular cusp or style is present on each tooth anterior and slightly labial to the paracone and there is a similar but smaller and less distinct cuspule posterior to the meta- cone. Each tooth has a distinct protoconule but no metaconule. There is no hypocone. It is also unusual that P^ is distinctly shorter than P-^ although wider. The structure of P- in this specimen seems to be rather different from that in the holotype of D. multicuspis and more molariform. Except for dimensions and proportions, apparent differences in structure of P ^"^ are possibly due only to the more worn condition of the holotype of D. multicuspis. Although far from identical, there is considerable resemblance between P^-^ of MMP 696M and the homologous teeth of North American Phenacodus. A fairly close ancestral relationship is con- firmed to that extent. However, considerable independent evolu- tion is also suggested. For example, P^-^^ especially, of MMP 696M, are more distinctly molariform than their homologues in Phenacodus. 8 BREVIORA No. 360 Order Notoungulata Roth Family Isotemnidae Ameghino Coelostylodon, new genus Etymology. Greek koilos, hollow, stylos, pillar, odon, tooth. The name is meant to recall former reference to Acoelodiis and resemblance to Pleurostylodon. It is also consonant with much Ameghinian nomenclature. Type-species. Coelostylodon florentinoameghinoi, new species, injra. Known distribution. Casamayoran, Argentine Patagonia. Diagnosis. Primitive notoungulates with complete, nearly closed dentition. Upper canine small and fully incisiform. Cheek teeth brachydont, P--M^ soon wearing so that crown presents a single fossa, without complex folds or anterior opening. M^-^ with flattened, slightly bifid lingual faces. M^^ with slight para- style and paracone folds and very feeble metacone swelling on ectolophs, no mesostyle. M^ subtriangular, with short but distinct metaloph, longer than M' or M- and almost as long as broad. Figure 5. Coelostylodon florentinoameghinoi, new genus and species. Holotype, MMP 1723M. Right C and pa - M^. X W2. Discussion. This genus is essentially that called Acoelodus by Ameghino (1901: 467) and discussed by me (Simpson, 1967: 57) under that name, but that apphcation of the name can no longer be sustained. The type-species of Acoelodus is A. oppositus Ameghino, 1897 (p. 454). The holotype of that species is MACN 10770, a fragment of a left mandibular ramus with much worn P 2.3 and part of Pj. That specimen is essentially indeterminate. In 1901 Ameghino referred to the species a poorly preserved skull, MACN 10753, and redefined the genus essentially on the basis of 1970 EARLY CENOZOIC — CHUBUT 9 that specimen. Desiring to validate as much as possible of Ame- ghino's nomenclature, I accepted the reference and redefinition, taking MACN 10753 as essentially a neotype in Ameghino's usage and designating it as such (Simpson, 1967: 58). However, that action is invalid under the present code of nomenclature (Stoll et al., 1964, Article 75) because in fact the holotype, MACN 10770, has not been lost or destroyed. If now there were a reasonable probability that the referred skull and the holotype belong to the same genus and species, the type designation would have no practical importance and one could continue to use the name Acoelodus for the genus and A. oppositus for the species represented by the referred skull. In fact, however, as I already mentioned in previous discussion (1967), there is no good evidence that the two specimens are of the same genus and species, and there is some contrary evidence. The contrary evi- dence is weightier than I previously indicated. If the comparative sizes of P 2.3 and P^-^ jn Acoelodus were approximately as in Pleurostylodon, a reasonable assumption, then P ^-s of the skull MACN 10753 are some 22 to 44 per cent larger in various dimen- sions than would be expected from the holotype of Acoelodus oppositus and are also different in relative sizes and proportions. It is thus highly improbable that the two specimens are conspecific, and if they are not conspecific, there is no reason to consider them congeneric. The genus represented by the holotype of Acoelodus oppositus is indeterminate, a notoungulate incertae sedis as far as my knowl- edge and judgment go. The genus represented by MACN 10753 is determinate and is distinct from any other for which there is comparable material known to me. This conclusion is reinforced by discovery of another specimen, MMP 723M, that can be re- ferred to the same genus as MACN 10753 but is specifically dis- tinct. In order to avoid possible further confusion with Acoelodus, the species represented by MMP 723M is made type-species for the new generic name. In order to avoid still another confusion, it must be noted that none of Ameghino's figures labeled "Acoelodus oppositus" in sev- eral of his publications are conspecific or congeneric either with the holotype of that species or with MACN 10753 (for details and discussion see Simpson, 1967: 58-59). Ameghino considered Acoelodus as closely related to Oldfield- thomasia and so placed the latter in his family Acoelodidae, but 10 BREVIORA No. 360 that was based on specimens dubiously or incorrectly referred to Acoelodus, including at least one that in fact belongs in Oldfield- thomasia. Since the holotype of the type-species of Acoelodus is not identifiable as to family, the name Acoelodidae has no estab- lished significance. For that reason, I named a family Oldfield- thomasiidae for Oldfieldthomasia and its probable relatives (Simp- son, 1945: 126). In revision of the Casamayoran fauna, I put MACN 10753, under the incorrect reference name Acoelodus oppositus, in the Oldfieldthomasiidae. The genus Coelo stylo don, to which that specimen is now referred as holotype of C. caro- loameghinoi, is of uncertain family position. It differs from all adequately characterized previously named genera of both the Oldfieldthomasiidae and the Isotemnidae, but has resemblances to both families. Present reference to the Isotemnidae is very tentative. The upper molar structure is most nearly similar to that of Pleu- rostylodon, an isotemnid, among adequately known genera, but Coelostylodon differs from Pleurostylodon and other isotemnids in its small, incisiform canine and various other details. The canine is more like that of Oldfieldthomasia, but the molars are quite different. Coelostylodon florentinoameghinoi , new species Etymology. For Florentino Ameghino, famed describer of most of the Casamayoran fauna. Combination of given and family names is a nomenclatural device that he often used. Holotype. MMP 723M, nearly complete but badly crushed skull. Hypodigm. The holotype only. Horizon and locality. Lowest Casamayoran beds in the bar- ranca south of Lago Colhue-Huapi, Chubut. Diagnosis. Significantly larger than C. caroloameghinoi (see Table 3 ) . Posterolabial angle of M^ less projecting. Discussion. The teeth are deeply worn in both of the holo- types now referred to this genus. As preserved, there is no marked, discernible difference in structure of the cheek teeth. MMP 723M clearly has the small canine considered characteristic of the genus. P^^^are almost completely fragmented, but seem to have been quite small, perhaps more so, relatively, than in MACN 10753. The posterolabial corner of M^ does not project so dis- tinctly as in MACN 10753. 1970 EARLY CENOZOIC CHUBUT 1 1 The skull is too badly crushed to make out much significant de- tail. It seems to have been a primitive, unspecialized notoungulatc skull generally similar to that of Pleurostylodon. Coelostylodon caroloameghinoi , new species. Acoelodus oppositus, in error, Ameghino, 1901: 365; Simpson, 1967: 58 and plate 11, fig. 1. Etymology. For Carlos Ameghino, who discovered the Casa- mayoran fauna and found the holotype of this species. Holotype. MACN 10753, poorly preserved anterior part of skull. Hypodigm. The holotype only. Horizon and locality. Casamayoran, barranca south of Lago Colhue-Huapi. Diagnosis. Significantly smaller than C. florentinoameghinoi (see Table 3). Posterolabial angle of M^ sharply produced posteriorly (or distally). Discussion. In 1901 and thereafter Ameghino based his con- cept of Acoelodus oppositus mainly on this specimen, but he did not figure it, and none of the specimens figured by him as Acoe- lodus oppositus belong to the present genus or species. The holo- type is adequately figured in my previous memoir, as cited above. REFERENCES Ameghino, F. 1897. Mammiferes cretaces de I'Argentine. Deuxieme contribution a la connaissance de la fauna mammalogique des couches a Pyrotherium. Inst. Geog. Argentina, 18: 406-429, 431-521. . 1901. Notices preliminaires sur des ongules des terrains cretaces de Patagonia. Bol. Acad. Nac. Cien. Cordoba, 16: 350-426. Cabrera, A. 1927. Datos para el conocimiento de los dasiuroideos fosiles argentinos. Rev. Mus. La Plata, 30: 271-315. Simpson, G. G. 1945. The principles of classification and a classification of mammals. Bull. American Mus. Nat. Hist., 85: i-xvi, 1-350. . 1948. The beginning of the Age of Mammals in South America. Part 1. Bull. American Mus. Nat. Hist., 91: 1-232. . 1967. The beginning of the Age of Mammals in South America. Part 2. Bull. American Mus. Nat. Hist., 137: 1-259. Sinclair, W. J. 1906. Marsupialia of the Santa Cruz beds. Repts. Prince- ton Univ. Exped. Patagonia, 1896-1899, vol. IV, part III: 333-460. Stole, N. R., et al. 1964. International code of zoological nomenclature adopted by the XV International Congress of Zoology. International Trust for Zoological Nomenclature, London. 12 BREVIORA No. 360 1^ vo J f*-l s On o k. cu 1/2 S D S O ^ a s "S ■^ u 00 a. c >, s as .rs \J ^j^**- »^iiw i-«»- t'j*- ^^« w^ (11 ^^ j^ v^ I— • ■■ri -^ -^ = j= 00 lO ■_S X) C tsr-^osvoor-^ II II ^« UJ ^ CO 'q ? 2 <^ < _ § 3 *- •a ^ ^ «5 C « ja c D. '-' Oftoii i|«otsoo<7\TtTj- a--' C S'-v II lldr-^rit-^-^td 2w |« c c'o c'O d"^ 0*0 0*0 cco cJi^ ^^ H>J^ »^^ ^^ C^ 'c3 C ""So 4) >. C s (o '<->jS'>->.C'>->x:'Mf'W.c t/i3 ^ W)*-'60-«->&0-«-i0O-«-'6O*-' ^cn C/0 S o c /1 o in 10 o\ ^ r<-i 00 o o I I o o I in o r^ ON ^ VO 00 <-o m o \0 vo o VO ■* ^ m in ON >n m -Si O m ON >n Tj- o -"t o VD ^ 00 O O '-I NO ^H Tj- ON f<-) 00 Tt -H rt in 00 n o m in ri r- I o o ^ fs| 00 r4 r-- in I I O 00 in in a ■ij ri fN in in o VO I ND 00 O <1J 1970 ECTOPARASITIC ODOSTOMIA 7 REFERENCES CITED Ankel, F., and a. M. Christensen. 1963. Non-specificity in host selec- tion by Odostomia scalaris MacGillivray. Vidensk. Medd. fra Dansk naturh. Foren., 125: 321-325. Boss, K., AND A. Merrill. 1965. Degree of host specificity in two species of Odostomia (Pyramidellidae: Gastropoda). Proc. Malac. Soc. Lon- don, 36: 349-355. Dehlinger, R. 1969. Host specificity and the feeding ecology of the ectoparasitic family Pyramidellidae. Unpublished manuscript, 55 pp. Emlen, J. M. 1966. The role of time and energy in food preference. Amer. Nat., 100: 611-617. Fretter, v., and a. Graham. 1949. Feeding and reproduction in pyra- midellids. Nature, 163: 361-362. . 1962. British Prosobranch Molluscs. London, Ray Society, 755 pp. Hopkins, S. H. 1956. Odostomia impressa parasitizing southern oysters. Science, 124: 628-629. Ivlev, V. S. 1961. Experimental Ecology of the Feeding of Fishes. New Haven, Yale Univ. Press, 302 pp. Loosanoff, V. L. 1956. Two obscure oyster enemies in New England waters. Science, 123: 1119-1120. Robertson, R. 1967. The life history of Odostomia bisiitiualis, and Odostomia spermatophores (Gastropoda: Pyramidellidae). Year Book Amer. Phil. Soc, 1966: 368-370. Robertson, R., and V. Orr. 1961. Review of pyramidellid hosts, with notes on an Odostomia parasitic on a chiton. Nautilus, 74: 85-91. Scheltema, a. H. 1965. Two gastropod hosts of the pyramidellid gas- tropod Odostomia bisiitiiralis. Nautilus, 79: 7-10. Wells, H. W. 1958. Predation of pelecypods and gastropods by Fas- ciolaria hunteria (Perry). Bull. Mar. Sci. Gulf and Caribbean, 8: 152-166. Wells, H. W., and M. J. Wells. 1961. Three species of Odostomia from North Carolina, with description of new species. Nautilus, 74: 149-157. BREVIORA MiaseitairM of Comparative Zoology Cambridge, Mass. 8 January, 1971 Number 364 A NEW SCINCID LIZARD FROM BOUGAINVILLE, SOLOMON ISLANDS Allen E. Greer and Fred Parker^ Abstract. The relationships of Sphenomorphus transversus, n. sp., from Bougainville, Solomon Islands, are obscure, but in squamation it is most similar to inaciilatiis, boiilengcri, forinosensis, Uneopiinctiilatiis. and indiciis from eastern Asia; nielanochlorus from New Guinea; and sanctiis from Sumatra and Java. S. transversus differs most noticeably from these species and from other Bougainville skinks in its dorsal pattern of trans- verse dark brown bands on a light olive ground color. During investigations in 1960-1963 by Parker on Bougain- ville, Solomon Islands, a single individual of a previously unde- scribed species of skink was collected. Since one subsequent trip (1966) has failed to uncover other specimens of the species, and as the possibilities of a second return trip to Bougainville in the near future are slim, it seems best to describe the new species from the single specimen at hand. On the basis of current generic concepts, the species is assigned to the genus Sphenomorphus and may be known as Sphenomorphus transversus^ new species Holotype. Museum of Comparative Zoology 76485; collected by a native for Fred Parker at about 2000 feet above sea level in an area approximately five miles east of Kunua, northeastern Bougainville (Fig. 1), on 9 September 1962. 1 P. O. Box 52, Daru, Western District, Territory of Papua and New Guinea. - The species name calls attention to the dark transverse bars on the dorsum. BREVIORA No. 364 155° E 56° E BUKA I 30 miles ■6°S •7°S 6°S-^ I55°E SHORTLAND MONO I (^ FAURO 7°S- 156° E Figure 1. Map of Bougainville showing the location approximately 5 miles east of Kunua where the type and only known specimen of Spheno- morphiis transversus was collected. Diagnosis. Similar in squamation to those skinks of the genus Sphenomorphus (Table 1 ) that have a single anterior loreal, the frontal in contact with 3 or more of the 5 or more supraoculars, frontoparietals and interparietal distinct, no nuchals or trans- versely enlarged scales in the two vertebral rows (Figs. 2 and 3), I97' A NEW SOLOMON ISLANDS SKINK and the digits and limbs well developed and overlapping when adpressed to the body, but differing from other skinks with this diagnosis in having the following combination of characters: pre- frontals separated medially, 36 rows of smooth scales around mid- body, 28-29 smooth subdigital lamellae on the 4th toe, and a color pattern of brown transverse bands on a light olive-green ground color (Figs. 4 and 5) — a color pattern most similar to those of the distantly related Sphenomorphus flavipes, Scincella prehensicaiida, and Leiolopisma semoni of New Guinea. Description. Body form relatively slender; well-developed pen- tadactyl digits and Umbs that overlap when adpressed to the body (tip of 4th toe reaches middle of forearm); snout-vent length 68 mm, tail 92 mm. Figure 2. Dorsal view of the head of the holotype of Sphenomorphus transversus (MCZ 76485). BREVIORA No. 364 Head not depressed, snout somewhat pointed; rostral slightly wider than deep, projecting slightly onto dorsal surface of snout between nasals; external naris in single nasal; no supranasals; single anterior and posterior loreals; frontonasal slightly wider than long, forming a short suture with the rostral and a very short suture with the frontal; prefrontals large, barely separated at their inner angles; frontal 1% times as long as wide, in contact with the three anteriormost supraoculars; 5 supraoculars, the first smallest, but in no way to be confused with the anterior super- ciliaries; lower eyelid scaly; 6th supralabial most directly below eye; a complete row of subocular scales separates scales of lower eyelid from supralabial series; frontoparietals paired and subequal in size with the single interparietal that is sharply pointed pos- teriorly; parietals meeting behind interparietal and bordered posteriorly by a single large temporal on either side and 5 large dorsal scales between the temporals; no symmetrical series of nuchals. External ear opening vertically elliptic, without auricular lobes; tympanum sunk slightly below level of skin; 36 smooth scales around midbody, the scales of the two vertebral rows not larger than those of the immediately adjacent rows; a pair of enlarged preanals; scales of three median subcaudal rows subequal in size. Digits rather long and slender; subdigital lamellae smooth and undilated throughout length of digit; 28-29 lamellae beneath 4th (longest) toe; upper surface of 4th toe covered by one or two Figure 3. Lateral view of head of Sphenoworpliiis tnnisversus (holo- type). 1971 A NEW SOLOMON ISLANDS SKINK 5 single scales at distal end, 3 rows of scales throughout center part and 4 rows near base (see Brongersma, 1942). Dorsal ground color light olive with a series of complete and incomplete transverse dark brown bands from nape to base of tail, the brown bands terminating in slightly expanded blotches on sides (Fig. 5), an effect especially pronounced at midbody; a horizontal brown stripe from anterior loreal through eye to tem- poral region; brown blotches on anterior and upper surfaces of limbs as well as on upper surface of tail; venter immaculate except for a few faint brown spots on throat and underside of tail. In life the undersides of the limbs, body, and tail were bright yellow. Field Notes. The only known specimen of S. transversus was taken by a native collector under a decaying log on the steep side of a montane river valley covered with tall primary forest. The natives did not recognize it as being distinct from S. concinnatus, a species common at the type locality of 5. transversus. Morphological Comparisons with Other Bougainville Skinks. S. transversus is immediately distinguishable from S. concinnatus by its more sharply tapered, longer snout; the absence of a dark blotch between the ear opening and the forellmb; the smaller ex- ternal ear opening; and, the regular transverse barring. Only two other Bougainville skinks, S. taylori and S. cranei — both very different from S. transversus in squamation — have transverse bands on the dorsum. In both these species, however, the dorsal pattern consists of very light transverse bands on a dark ground color, whereas in S. transversus the transverse bands are darker than the ground color. S. transversus also has a more noticeably pointed snout than either S. taylori or S. cranei. Skull Characters. It is extremely difficult, if not impossible, to remove the skull of most skinks without severely damaging the skin of the head. For this reason we have not attempted to pre- pare a skull from the type and only known specimen of S. trans- versus. We have, however, had a palatal view of the skull, and the salient features are as follows: there are 9 premaxillary teeth; the palatine and pterygoid bones meet along the midhne to form a fairly extensive secondary palate; there is no ectopterygoid pro- cess; and, there are no pterygoid teeth. Unfortunately, these characteristics are not particularly diagnostic, for they would not exclude S. transversus from close relationship with any number 6 BREVIORA No. 364 of other lygosomines, including those discussed below that are most like S. transversiis on the basis of external morphology. Comparison with Morphologically Similar Species. In squama- tion S. transversus is most similar to those species of Sphenomor- phus (Table 1) that have a single anterior loreal, the frontal in contact with 3 or more of the 5 or more supraoculars, the fronto- parietals and interparietal distinct, no nuchals or transversely en- larged scales in the two vertebral rows, and the digits and limbs well developed and generally overlapping when adpressed to the body. This assemblage, which is almost surely not monophyletic, is distributed from southern Asia through the Indo-Australian archipelago and Philippines to New Guinea, but not Australia. Seven species in this group have the prefrontals separated me- dially (in all or some individuals), as is the case in the single specimen of S. transversus. Six of these seven species (maculatus, boulengeri, jormosensis, lineopimctulatus , and indiciis from southern Asia and melanochlorus from New Guinea) have sub- stantially fewer subdigital lamellae on the 4th toe (16-22) than does transversus (28-29), and the seventh {sanctus from Suma- tra and Java) has finely striated body scales to distinguish it from the smooth-scaled transversus. Furthermore, none of these seven species have a dorsal body pattern consisting of well-defined dark crossbars as does transversus. This very characteristic dorsal body pattern of dark crossbars on a light ground color is most similar to the patterns of the cross- banded color morph of Sphenomorphus fiavipes, the females of Scincella prehensicauda, and of all Leiolopisma semoni. These three species are endemic to neighboring New Guinea and might, therefore, seem to be likely relatives of Sphenomorphus trans- versus. Current work on scale and palatal characters, however, indicates that while fiavipes, prehensicauda and semoni are them- selves closely related (in spite of current generic allocations), they are only distantly related to SphenomorpJius transversus. ACKNOWLEDGEMENTS Dr. Ernest E. WiUiams of the Museum of Comparative Zoology read the manuscript in several drafts and offered several helpful suggestions. Mr. Laszlo Meszoly did the drawings for Figures 2 and 3, and Mr. Tan T. Riddell took the photographs for Figures 1971 A NEW SOLOMON ISLANDS SKINK 7 4 and 5. Part of Greer's work on this paper was done while he was a postdoctoral fellow of the National Science Foundation. Partial support was provided by National Science Foundation grant GB 6944 to Ernest E. Williams. LITERATURE CITED Brongersma, L. D. 1942. On the arrangement of the scales on the dorsal surface of the digits in Lygosoma and allied genera. Zoologische Mededeelingen. 24 (1-2): 153-158. BREVIORA No. 364 (D <3 <>i r3 C3 .« :>. 1 1 1 1 1 1 1 1 1 ■a J= 0 a '5 1 1 1 1 1 1 1 1 1 (U (/: ^ c ;-! S -4— • o ^ 2 ^ £? •i-HKt ,_^ ^ _ ^ c3 a ;^ 5 •^ a ^ "■"s /^ /—v fo -— V 1 <^ m X i^ -Q o U c N— ^ ^.^ ^^ r4 **— ' fN| fS s 5 5 2 1 r^ r~- vo ^^ t-~- in ^"^ ^"^ o 1 5. *»• s~. \o >/^ 1 ■^ ^ cd o ^ -C *-4-^ ^ D • ^H Q 1 1 ^ 0) *.» 5 1 1 s + + + + 1 + + u. 0 + J3 o a. -a c > 5 6<5 + + + + + + + + + Dh 4) ^ "Ci on 3 li; ^ s cd cd X • ^H c o S-i <4J ^ 0 .1— < '4—* O ■4—* ^ 'v.* a\ (N >n 0 0 03 .& ID ■^ (N -* 5 0 > LU .S IS to ■Q c 0 (N 1 OJ 3 cd a ■| C _c ^ > 03 ^ Q > > 'S >SJ _5^ ■4—* o ^ X) _cd 2 * S cd e '3 > c 2 * 0 -s: * i 0 * -2 1 * 0 s: C/5 x> ^ -<: ^ t: ^ >? ~5 -c '^ 1971 A NEW SOLOMON ISLANDS SKINK '^ f^ ^ '^ ^ ^ ^-v f*^ I r*^ f^ f*% f*^ f*^ so O + + I + + + + + + + + rj OS (N O t~~ rj /^ V-> — rj tt C3 .2 - - ? c - "= i I o -I .^ S o i § I o ^ i I H tZ) Z CQ V5 00 ^ ■a F = S c (N ON r-4 00 o\ ^^ uT :i3 ■5) a L. 0 H 0 > 0\ OS o 00 r^ OS _ 00 — u a. c 1- u c > 3 E ffl =3 ■** ;^ * C:: ^ =2 >^ * ^ 0 C ^ 5 * s. a c i. C ^ 0 ^ 0 •S S i .£3 s 1 *5 10 BREVIORA No. 364 Figure 4. Dorsal view of the holotype of Splicnomorphiis transvcrsus (MCZ 76485) showing the dark transverse bands which give the species its name. 1971 A NEW SOLOMON ISLANDS SKINK 11 Figure 5. Lateral view of Sphenomorphiis transversiis (holotype). BREVIORA Me seem of Comparative Zoology Cambridge, Mass. 15 January, 1971 Number 365 CHARACTERS AND SYNONYMIES AMONG THE GENERA OF ANTS. PART IV. SOME GENERA OF SUBFAMILY MYRMICINAE (HYMENOPTERA: FORMICIDAE) William L. Brown, Jr.^ Abstract. Archaeomyrmex is a new junior synonym of Myrmecina, and tribe Archaeomyrmicini accordingly is a synonym of Myrmecinini. Dodoiis is a new junior synonym of Pristomyrmex, and the synonymy of Hylidris under Pristomyrmex is reaffirmed. Limnomyrmex is a new junior synonym of Leptothorax subgenus Nesomyrmex. The genera considered in this part all belong to subfamily Myr- micinae. My main purpose here is to explain briefly why some new synonymy should be proposed. The taxa concerned have all been studied during the course of the project "a reclassification of the Formicidae," supported by National Science Foundation Grants G-23680, GB-2175, and GB-5574. The taxonomic conclusions will eventually be embodied in a synopsis and illustrated keys to the ant genera of the world. I feel that justification for revisionary changes should be published as the need for the changes becomes clear. In this way, important findings of the study are made avail- able for use by all ant taxonomists without undue delay, and the synoptic parts can be freed from the clutter of many detailed taxonomic arguments. Myrmecina Myrmecina Curtis, 1829, Brit. Entom. 6: 226, pi. 265, male. Type species by original designation Myrmecina latreillii := Formica graminicola. Arclweomyrmex Mann, 1921: 448-451. Type species: Archaeomyrmex cacabau, by original designation. NEW SYNONYMY. 1 Department of Entomolo{!;y, Cornell l'ni\ ersity, Ithaca, New York, 14850. 2 BREVIORA No. 365 The unique type of A. cacabau has been searched for in vain in the U. S. National Museum and Museum of Comparative Zoology ant collections, and must be considered lost. Fortunately, Mann's description and figures are reasonably detailed. From them, it is clear that the species is essentially a Myrmecina, a fact acknowl- edged by Mann when he wrote: "The epinotal and petiolar struc- ture are not unlike certain species of Myrmecina." Mann empha- sized the ventrolateral carina on each side of the head, probably without realizing that this is an invariable character of Myrmecina. He also cited the partly smooth, partly costate sculpture of the body, especially the trunk, which does set this species off from the "average" Myrmecina. But Myrmecina forms with more or less of the head and trunk smooth have not been completely unknown, and they even occur as phenetic variants in species that are usually heavily sculptured in these regions. Several Myrmecina species (at least three of them still unde- scribed) with predominantly smooth and shining head and trunk are in the MCZ collection. Two of the undescribed species are from the New Hebrides; in each of these, the posterior pair of propodeal teeth is spiniform, but the coarse costate sculpture of the lower pleural areas is preserved, as it is in cacabau. Another smooth species has come to me from the Philippines. The geo- graphical and morphological gaps are thus closed between the Fijian and Indo-Australian representatives of what is obviously a single stock. A complete examination of the A. cacabau descrip- tion fails to reveal any character that will separate it from Myrme- cina at generic level. Together with the two New Hebrides species, cacabau should be taken to represent no more than a weak species- group of Myrmecina. Incidentally, an interesting characteristic of some of these species is the extreme reduction of antennomere III, just distal to the pedicel. The distinction between " 1 1 -segmented" and "12-seg- mented" antennae in these forms may prove impossible to maintain as a key character. Mann's tribe Archaeomyrmecini of course falls as a new synonym of Myrmecinini. Pristomyrmex Pristomyrmex Mayr, 1866: 903. Type species: Pristomyrmex piingens, monobasic. 1971 SOME GENERA OF MYRMICINE ANTS 3 Hylidris Weber, 1941: 190. Type species: Hylidris myersi, by original designation. — 1952: 15-22. Synonymized by Brown, 1953: 9-10. Dodous Donisthorpe, 1946: 145; worker, male, larva. Type species: Dodous trispinosus, by original designation. NEW SYNONYMY. Dodous was based on the single species trispinosus, from Cocotte Mountain, Mauritius. Syntypes of this species deposited in the British Museum and in the Museum of Comparative Zoology are basically Pristomyrmex in form, but they have an extra pair of small teeth on the mesonotum, and the head and trunk are finely and regularly costulate (= striate). The larva, rather vaguely figured in the original description (fig. 4), has the elongate, pro- tuberant head characteristic of Myrmecina and Pristomyrmex. Donisthorpe also figured the male, which is like known Pristomyr- mex males in general habitus. The genitalia as very briefly de- scribed may be aberrant, but then the genitalia have not been studied at all in most other Pristomyrmex species. A second species of Dodous, D. bispinosus, was described from Mauritius by Donisthorp)e in 1949. I collected a small sample of strays of this species at the type locality, Le Pouce (Mountain) on 1 April 1969, the last day of a short stay on Mauritius. The speci- mens were all foraging workers taken on trees and on the main path through the scrubby native forest at about 800 m elevation on the "plateau^' near the summit. (A sudden storm prevented my finding any nests.) D. bispinosus differs from D. trispinosus in lacking the mesonotal teeth (though actually some of my bispinosus specimens have low mesonotal tubercles in place of the teeth) and in having predominantly smooth and shining sculpture. In fact, D. bispinosus is a rather ordinary, if slightly long-legged, Pristo- myrmex, and D. trispinosus goes only one step beyond. It seems absurd to put these two closely related species in different genera. Discovery of the annectant D. bispinosus makes it clear to me that Dodous is only the Mauritian complement of the widespread Old World genus Pristomyrmex. The concept of Dodous as a separate genus is, as far as 1 am concerned, as dead as its namesake. I have already (Brown, loc. cit.) placed Weber's Hylidris as a synonym of Pristomyrmex. Weber opposed this synonymy, main- taining that Hylidris is a distinct genus. But when he described Hylidris, Weber took no note of the African species of Pristomyrmex described previously by Santschi, Arnold, and Karavaiev, at least some of which are senior synonyms of his own Hylidris species and 4 BREVIORA No. 365 subspecies (Weber, 1952). Weber has never produced a charac- terization of Hylidris that will separate it as a genus from Pristo- myrmex, and particularly from the long-synonymized "subgenus" Odontomyrmex. \ have collected P. orbiceps in the Ivory Coast, and can affirm that colony behavior (lethisimulation), larval form, and general adult morphology are fully those of Asian and Aus- tralian Pristomyrmex as I have seen them in nature. Pristomyrmex is a sharply defined and compact genus, and there is no reason that I know of to set the African species apart from it. In fact, the African species are as nearly "average" for the genus Pristomyrmex as one is likely to find. Some other misconceptions expUcit or implied in Weber's dis- cussion of 1952 need correction. The petiole of Pristomyrmex has a distinct anterior peduncle, though it is short in some species. In contrast, the related genus Myrmecina has a sessile petiole of more or less prismatic shape. Pristomyrmex and Myrmecina are not very closely related to tribe Tetramoriini, though two species have been wrongly placed in tetramoriine genera in the past. The larvae, for one thing, are very different, and it seems that they may furnish the best tribal character for the Myrmecinini if we limit the tribe to Pristomyrmex, Myrmecina, Acanthomyrmex, and possibly the little-known Perissomyrmex of Guatemala, the larvae of which have not yet been found. Tribe Tetramoriini is not "worldwide" in dis- tribution, if one ignores obvious introductions by man. The tribe has no native species in South or Central America, and only a single species of Xiphomyrmex occurs in (Sonoran) North America. Leptothorax subgenus Nesomyrmex Nesomyrmex Wheeler, 1910, Bull. Amer. Mus. Nat. Hist., 28: 259. Type species Nesomyrmex clavipilis, monobasic. Leptothorax (Goniothorax) aiict., preoccupied. Leptothorax (Caulomyrma) Forel, 1914, Bull. Soc. Vaudoise Sci. Nat., 50: 233. Limnomyrmex Arnold, 1948, Occas. Pap. Nat. Mus. S. Rhodesia, 2(14): 222. — 1952, Ibid., 2(17): 460, discussion. Type species Limiioiuyrmex stramineiis, monobasic. NEW SYNONYMY. Soon after its description, the late Dr. Arnold and I engaged in correspondence on the question of the distinctness of Limnomyr- mex from the subgenus Nesomyrmex of Leptothorax. He stoutly maintained that Limnomyrmex was a good genus, and in 1952 {loc. cit.) he argued again in print for this stand. Now that I have finally 1971 SOME GENERA OF MYRMICINE ANTS 5 seen the unique worker type of L. stramineiis in the Arnold Collec- tion at Bulawayo, I can only place this species among the other known African Nesomyrmex. I have searched in vain for any characters that might set Limuomyrmex apart as a genus. In the form of the trunk and both petiolar and postpetiolar nodes, it is about "average" for a Nesomyrmex from Africa, and resembles in a subdued way some neotropical members of the group. Sculptural and a few other differences mark stramineus , but these do not seem to be more than species characters. The antennae of stramineus are 12-segmented; Nesomyrmex can have either 11 or 12 segments. The status of Nesomyrmex as a subgenus of Leptothorax is main- tained for the time being, pending the proper study of both taxa. REFERENCES Brown, W. L.. Jr. 1953. Characters and synonymies among the genera of ants. Parti. Breviora No. 11: 1-13. DoNisTHORPE, H. St. J. K. 1946. A new genus and species of Formicidae (Hym.) from Mauritius. Proc. Roy. Entomol. Soc. London, ser. B, 15: 145-147. Mann, W. M. 1921. The ants of the Fiji Islands. Bull. Mus. Comp. Zool., 64: 401-499. Mayr, G. 1866. Diagnosen neuer und wenig gekannter Formiciden. Verhandl. Zool.-bot. Ges. Wien, 16: 885-908, pi. 20. Weber, N. A. 1941. Four new genera of Ethiopian and Neotropical Formicidae. Ann. Entomol. Soc. Amer., 34: 183-194. . 1952. Studies on African Myrmicinae, I (Hymenoptera, Formicidae). Amer. Mus. Novitates 1548: 1-32. JBREVIORA Mmseuinni of Comparsitive Zoology Cambridge. Mass. 15 January, 1971 Number 366 PULSED SOUNDS OF THE PORPOISE LAGENORHYNCHUS AUSTRALIS William E. Schevil! and William A. Watkins^ Abstract. Pulsed sounds of the porpoise Lagenorhynchns aiistmlis of southern Chile were recorded and analyzed. Most were low-frequency clicks; some had a 2-kHz bandwidth centered near 1 kHz, and others had a 10- or 12-kHz bandwidth with the principal frequency in the lower 5 kHz. These porpoises also produced a rapidly pulsed tonal sound. All these sounds were very low-level and rarely audible at a distance as great as 20 m. We heard none of the whistlelike squeals characteristic of many delphinids. From 12 November to II December 1968 the research ship HERO of the National Science Foundation (Antarctic Research Program) cruised between Valparaiso and Cape Horn, searching for cetaceans and pinnipeds, mostly in the sheltered inland water- ways of southern Chile. We were concerned with listening for and recording the underwater sounds of these animals. The species most frequently seen and collected was Lcigeno- rhynchus australis (Peale, 1848 ), which has been reported from the west coast of Chile south of about S Lat. 40° around Cape Horn to the Falkland Islands. We heard them much less often than we saw them. Methods. The recordings that are analyzed here were made on 23 November in Canal Messier (at 48° 10' S) and 1, 3, 5, and 6 December west and south of Navarino Island behind Cape Horn. An Atlantic Research LC-34 hydrophone was used to pick up the sounds. An impedance-matching pre-amplifier (WHOI) was in- serted in the cable 30 cm from the hydrophone. Two cable lengths 1 Contribution No. 2562 from the Woods Hole Oceanographic Institution. 2 BREVIORA No. 366 were used, 125 m from R/V HERO and 30 m from HERO's whale- boat. The hydrophone depth varied with local conditions and ex- periments, from 2 m to nearly 125 m; it was usually suspended 6-8 m from a surface float (a rubber balloon) and allowed to drift as far from ship or boat as cable-length permitted. Tape recordings were made with either a modified Uher 4400 recorder or a WHOI-built springwound recorder, using a hydro- phone amplifier (Watkins, 1963). When the Uher was in use, the system-response was limited to a bandwidth, within 4 db, of 40 to 20,000 Hz; with the WHOI machine, system-response was 20 to 32,000 Hz (within Vz db from 30 to 30,000 Hz). Playback for analysis was on Crown 800 tape recorders. Spectrographic analy- ses were made on a Kay Electric model 7029A analyzer and ampli- tude analyses on a Tektronix 5 35 A oscilloscope. The porpoises were approached as closely and as unobtrusively as possible, but even so, often the only sounds heard from the ani- mals were within the first 5 seconds of the listening attempts. Un- fortunately, because of the disturbance of the water by arrival of the boat and the motion of the hydrophone, a longer time than this usually was required before local ambient noise could settle down enough for faint sounds to be recorded. Usually nothing was heard from the porpoises, partly because they were generally taci- turn and seemed to produce sounds only occasionally, and partly because their sounds were too faint to be audible except on close approach, within a few meters of the animals. They appeared to be silent when disturbed. Sounds. The sounds heard from Lagenorhynchus aiistralis were all pulsed. Mostly they were clicks produced in short series or slow bursts, but sometimes a rapidly pulsed sound (a buzz) that had a tonal quality was heard. The buzz was the only sound heard from L. australis at any distance, and it was produced only occasionally; consequently most attempts to listen to these porpoises were entirely unsuccessful. No squeals (whistles) were heard; this was unex- pected since we have heard squeals from other Lagenorhynchus (L. albirostris, acutus, obliquidens, and the obscurus of New Zealand). These sounds of L. australis were low level and generally in- audible beyond about 10-20 m. We estimate that the loudest chcks were no more than -20 db re 1 dyne/cm- at 1 m, from known hydrophone sensitivities and tape saturation levels, and 1971 PORPOISE SOUNDS 3 assumed supply voltages and amplifier gains. On only a few occa- sions were we convinced that we knew which individual produced the sounds that we heard, and therefore our estimates of signal strength and of distance from the hydrophone are but guesses. The click-sounds were of two types: a broadband click, and a relatively restricted-bandwidth click (narrowband) at predomi- nantly low frequencies. These two clicks never seemed to be mixed. Both types were heard, we thought, from any one individ- ual, with no obvious separation between the different kinds of clicks, and no gradual transition. Though both types of clicks were sometimes heard at slow rates (1 or 2 per sec), the broad- band click was usually produced at a more rapid repetition-rate (20 to 80 per sec.) than the narrowband click (5 to 25 per sec.). The broadband click was shorter and had less energy at low fre- quencies than the narrowband click. See the table for a comparison of these two clicks. The broadband click (Figs. lA and 2) was characterized by a sharp onset, a short duration, as well as a more or less continuous spectrum to 10 or 12 kHz, occasionally to 16 kHz. Analyses of clicks showed a general drop in intensity of 1 to 2 db per 1000 Hz above 5 or 6 kHz. This drop was greater than is consistent with normal frequency-selective absorption for these distances and fre- quencies, so we assume that this attenuation is characteristic of the click of L. australis. The duration of the broadband click was consistently a little less than 1 msec. Because of the general low level of the sounds as well as their usual reduction in intensity at higher frequencies, the clicks were easily masked by background ambient. The narrowband click (Figs. IB and 2) was restricted in fre- quency to the lower 2000 Hz and appeared to have its greatest in- tensity at or below 1000 Hz. Harmonics did exist, though at greatly reduced levels. Analysis at high gain (but still undistorted) showed some of the narrowband clicks with harmonic frequencies to 5 or 6 kHz. The narrowband click, with a duration of 1.5 to 3 msec, usually occurred at slower repetition-rates (5 to 25 per sec), and consistently had higher intensity at low frequencies than the broad- band click. Perhaps because of its lower-frequency emphasis and therefore better transmission characteristics, the narrowband click was the one most commonly heard. The third type of sound, the buzz (Fig. 3) was heard on a few 4 BREVIORA No. 366 occasions. This buzz had emphasis at discrete higher frequencies, such that both the fundamental and high frequency overtones were predominant in the aural impression of the sound. The buzzes varied in duration from 0.6 to 1.1 sec. They were composed of a pulsed fundamental near 300 Hz (Fig. 4) and strong overtones at 4 to 5 kHz. Two or three sidebands of the pulse repetition-rate (modulation. Fig. IC) may be noted grouped around the 4- to 5-kHz overtone in spectral analysis (Fig. 3) of these buzzes (see Watkins, 1967). The fundamental frequency of the buzz was more intense than the overtones, yet at greater distances only the 4- to 5-kHz tone (with its associated sideband structure) was audible. This was probably because of higher background ambient at the lower frequencies. The buzz appeared to be produced at a higher level than the clicks. Discussion. Because of both the pulsed quality of the buzz and its restricted frequency, we suppose that this sound was formed by rapidly repeated narrowband clicks. Singly, the narrowband clicks had few higher frequency components, but in a rapid series the overtones were prominent. This is somewhat similar to sounds pro- duced by Phocoena phocoena, composed of a rapid repetition of narrowband clicks to form a continuous sound with selected higher frequency emphases (Schevill, Watkins, and Ray, 1969). We did not find the variation in the overtones of the buzz of L. australis that we noted for Phocoena, but this may have been due to the limited number of the former's buzz sounds that were recorded well enough for such detailed analysis. Perhaps the buzz was used in communication and it may have been associated with stress. This could explain its relatively infre- quent occurrence. The only time that the buzz was heard when we thought we knew which porpoises were producing it (in Paso Micalvi outside of Seno Grandi, Navarino Island, 6 December), a group of three animals 15 to 20 m distant suddenly seemed to be in a scuffle, darting at and away from each other. This sudden un- usual activity coincided with the production of four buzzes, two of them concurrently (Figs. 3 and 4), and so we assume that these sounds were produced by these porpoises. We have no evidence that Lagenorhynchus australis echolocates. If the click sounds were used for echolocation as in some other species ( Tursiops truncatus, Steno bredanensis, Orcinus orca, Phocoena phocoena) , it must have been at relatively close ranges 1971 PORPOISE SOUNDS 5 because of the low level of the clicks. We did not hear any "ac- celerando" in click series such as is typical of echolocation runs during feeding by these other animals; however, we had no sugges- tion that the L. aiistralis were feeding when the clicks were heard. In fact, the clicks were not consistently associated with apparently investigatory behavior by the animals. Porpoises sometimes passed within a meter of the hydrophone and even appeared to return and examine it without our detecting any sounds. On the other hand, clicks were never heard unless a porpoise was close by. The two click-types perhaps are equivalent to the two basic click-categories noted for Tursiops by Norris, Evans, and Turner (1967). They name these clicks by their function, ''discrimination clicks" and "orientation clicks." The discrimination click of Tur- siops has a reduced bandwidth and emphasis of lower frequencies, while the orientation click has a wide bandwidth. In these respects they match the sounds heard from L. australis, though no be- havioral correlation was possible. Although two types of clicks were heard, one with a relatively restricted low frequency and other with broadband characteristics, it suggests the possibility that only one click type exists in reality and the variations noted result from changing orientation by an animal possessing a directional sound system. Other cetaceans have been shown to have such a directional sound field ( Tursiops, Nor- ris, Prescott, Asa-Dorian, and Perkins, 1961; Orcinus, Schevill and Watkins, 1966; Steno, Norris and Evans, 1967; and, Platanista, Evans in Herald et al., 1969). Our data is insufficient to rule out this possibility completely, but the evidence that we have seems to argue instead for two distinct click types: 1 . The click durations of the two types are different. High fre- quency emphasis in a low frequency click would not shorten the length of the pulse but would simply extend the bandwidth. 2. The two click-types suddenly interchange with no pause be- tween. We have no examples of a gradual shift from one type to the other and we have very few individual clicks whose char- acteristics are intermediate in form. Some of the subtle varia- tions observed in the higher frequency components of successive clicks of both types, however, may result from such direction- ality, though we did not have opportunity to observe any cor- relation of orientation with bandwith. 6 BREVIORA No. 366 Because of the difficulties we encountered in hearing the por- poises, we were impressed with the low level of their sounds. We also were acutely aware that it was not high background that obscured their sounds, since the ambient noise levels in this region were actually very low. Without carefully and recently calibrated equipment, such low sound-levels are difficult (and probably mean- ingless) to assess; however, our limitation much of the time ap- peared to be the self-noise of the equipment rather than the local ambient background. Perhaps the land barriers shielded the inland channels from the usual open sea sounds and at the same time provided enough shelter so that very little local wind and wave noise was generated. In addition, we recognized very little contri- bution of sound from other biological sources, and certainly these porpoises had but small influence on the local ambient sound. ACKNOWLEDGEMENTS We thank the Antarctic Research Program of the National Science Foundation for support and good help during the cruise on board R/V HERO. The acoustic analysis and preparation of this report were supported by the Office of Naval Research (Biol- ogy branch) under contract Nonr 4446 and Nonr 241.09, and by the National Science Foundation grant GA 1475. Experience and observation were shared with the other members of the scientific party aboard HERO; Kenneth S. Norris as chief scientist and George Harvey were particularly involved and helpful. We thank Elizabeth T. Bunce and Paul T. McElroy for criticism of the manu- script. LITERATURE CITED Herald, E. S., R. L. Brownell, Jr., F. L. Frye, E. J. Morris, W. E. Evans, AND A. B. Scott. 1969. Blind river dolphin: first side-swimming ceta- cean. Science. 166 (3911): 1408-1410. NoRRis, K. S., AND W. E. Evans. 1967. Directionality of echolocation clicks in the rough-tooth porpoise, Steno hredanensis (Lesson). In W. N. Tavolga (ed.), Marine Bio-Acoustics, vol. 2. Oxford, Pergamon Press. Pp. 305-314. Norris, K. S., W. E. Evans, and R. N. Turner. 1967. Echolocation in an Atlantic bottlenose porpoise during discrimination. //; R.-G. Busnel (ed.), Les Systemes Sonars Animaux, Biologic et Bionique, Jouy-en- Josas, France. Pp. 409-437. 1971 PORPOISE SOUNDS 7 NoRRis, K. S.. J. H. Prescott, P. V. Asa-Dorian, and P. Perkins. 1961. An experimental demonstration of echo-location behavior in the por- poise, Tursiops truncatus (Montagu). Biological Bull.. 120 (2): 163-176. SCHEVILL, W. E., AND W. A. Watkins. 1966. Sound structure and direc- tionality in Orciniis (killer whale). Zoologica (N. Y.), 51 (2): 71-76. ScHEViLL, W. E., W. A. Watkins, and C. Ray. 1969. Click structure in the porpoise Phococna phocncna. Jour. Mammalogy. 50 (4): 721-728. Watkins, W. A. 1963. Portable underwater recording system. Undersea Technology, 4 (9): 23-24. Watkins, W. A. 1967. The harmonic interval: fact or artifact in spectral analysis of pulse trains. In W. N. Tavolga (ed.). Marine Bio- Acoustics, vol. 2. Oxford. Pergamon Press. Pp. 15-42. BREVIORA No. 366 B / msec/div. Figure 1. Oscillographic pictures of (A) the broadband click, (B) the narrowband click, and (C) the pulse modulation of the buzz. Ambient noise is superimposed on these sound traces. 1971 PORPOISE SOUNDS 8000- Hz 4000- I 0 '1 05 Seconds Figure 2. Spectrographic analysis shows a burst of broadband clicks followed by narrowband clicks. Although the latter become much greater in amplitude as the animal conies closer, the frequency spectrum remains relatively restricted. The bandwidth of the analyzing filter is 300 Hz. This figure is the result of a repetitive analysis, with a small horizontal displace- ment of the paper between analyses to widen artificially the traces of these short-duration sounds for better photographic reproduction. Narrowband click Broadband click Bandwidth Principal frequency Duration Repetition rate Intensity (re 1 dyne/cm2) 2 kHz 1 kHz or less 1.5 to 3 msec 5 to 25 /sec — 20 db at principal frequency 10 or 12 kHz from less than 1 to 5 kHz 0.8 to 1 msec 20 to 80/sec — 20db spread over bandwidth Table of characteristics of the two types of click heard from Lagenorhynchiis australis. 10 BREVIORA No. 366 7000- 5000- Hz 2000- Seconds tw i \ ^ai 1.5 Figure 3. Two simultaneous buzzes have empiiasis in the 4- to 5-kHz region as well as a strong fundamental at about 300 Hz. The analyzing filter bandwidth is 300 Hz. Compare Fig. 4. 1000- Hz 500- ■ ' ' Seconds Figure 4. The fundamental frequencies of the same two buzzes shown in Fig. 3 show variation in the region of 300 Hz. The continuous low fre- quency band is ship's propulsion noise from the HERO about 5 or 6 miles away. The analyzing filter bandwidth is 45 Hz. BREVIORA Mniseeim of Cojniiparative Zoology Cambridge, Mass. 15 January, 1971 Number 367 MICROMISCHODUS SUGILLATUS, A NEW HEMIODONTID CHARACIN FISH FROM BRAZIL, AND ITS RELATIONSHIP TO THE CHILODONTIDAE Tyson R. Roberts Abstract. A new genus and species of Hemiodontidae, designated as a new subfamily, Micromischodontinae, is described from the lower Rio Negro and Middle Amazon of Brazil. Its osteology is described and figured. It appears to be the hemiodontid genus most closely related to Chilodontidae. The relationship between the functional and replacement teeth of the pharyngeals suggests the probable manner in which the peculiar pharyngeal teeth characteristic of Chilodontidae and Anostomidae evolved. A brief definition is given of the family Chilodontidae. ACKNOWLEDGEMENTS Dr. Stanley H. Weitzman, Division of Fishes, U.S. National Museum, suspected hemiodontids and chilodontids might be re- lated and mentioned this to me two or three years ago. He is now working on the relationships of these families and has graciously let me examine unpublished illustrations of chilodontid osteology. I am indebted to Sr. Heraldo Britski, curator of the fish collections of the Museu de Zoologia, Universidade de Sao Paulo, and leader of the Expediyao Permanente da Amazonia when the new fish was collected, for permitting me to prepare its description. Professor George S. Myers, Division of Systematic Biology, Stanford Uni- versity. Dr. Weitzman and Sr. Britski reviewed the manuscript. 2 BREVIORA No. 367 INTRODUCTION The hemiodontid herein described, collected by the Expedi^ao Permanente da Amazonia' in 1967, represents a new subfamily. While lacking certain morphological peculiarities of the highly specialized Chilodontidae, it nevertheless appears to be more closely related to them than is any other hemiodontid. In the light of its discovery, there can be little doubt that Chilodontidae and Hemiodontidae are indeed closely related. Although the pharyngeal teeth of this new form are single cusped, the relation- ship between functional and replacement teeth on the pharyngeals suggests the primitive condition from which the peculiar multi- cuspid pharyngeal teeth characteristic of Chilodontidae and Anostomidae presumably evolved. Its highly distinctive trophic structures indicate an unusual mode of feeding, perhaps similar to that of Bivibranchia, considered to have the most specialized trophic structures of all characoids. Nevertheless, it represents a hne distant from Bivibranchia (and the related but less specialized Argonectes) . Hemiodontidae are marvelously streamlined, swift-swimming fishes mostly six inches to a foot long. They form small groups in open water in big rivers and lagos. Hemiodus are known in Brazil as "voadores" because of their ability to jump. A group of voadores leaping away from predaceous fishes or over a seine net to safety is an impressive sight. Spawning presumably takes place in open water and the young probably grow up in aggregations staying near the bottom in fairly shallow open water. It appears to be unrecorded whether hemiodontids form huge schools or undertake major spawning migrations. Hemiodontids hitherto known readily fall into two subfamilies, Hemiodontinae and Bivibranchiinae, adults of which have multi- cuspid teeth in the upper jaw and no teeth in the lower jaw. The minute, unicuspid teeth and other features of the new fish differ so greatly from previously known forms that it represents a new sub- family. 1 The Expedicao Permanente da Amazonia, under the direction of Dr. P. E. Vanzolini, is a cooperative effort among the Museu de Zoologia, Universidade de Sao Paulo (MZUSP); Instituto Nacional de Pesquisas da Amazonia (INPA) in Manaus; and Museu Goeldi in Belem. It is financed by the Fundagao de Amparo a Pesquisa of the state of Sao Paulo. Much attention has been devoted to fishes since fieldwork began in 1967. The fish collections are housed at MZUSP. 1971 BRAZILIAN CHARACIN FISH 3 MICROMISCHODONTINAE, new subfamily Highly streamlined, fusiform fishes typically hemiodontid in habitus and osteology. Teeth pedicellate, with a single strongly recurved cusp; tooth crowns black or brownish black, stalks deep yellow or yellowish brown. Each dentary and lower pharyngeal with two, nearly coextensive rows of 50-60 teeth. Teeth on upper pharyngeal arranged in numerous, extremly regular rows (as in Anostomidae). Upper jaw with a broad frenum, nonprotractile; roof of mouth with fine, linear ridges, without valvelike structures. Gill rakers elongate and numerous, with six to eight papillae on either side of each gill raker, forming a dense carpetlike hning to gill chambers. Lower pharyngeals exceptionally long and slender, tooth-bearing for virtually their entire length. Upper limb of second gill arch with a fleshy membrane forming a sort of pocket anterior to upper pharyngeals. Posterior face of fourth gill arch with ordinary gill filaments; face of fifth gill arch smooth; no dorsal diverticulum between fourth and fifth arches. Stomach reversed, that is, with cardiac portion (entrance of esophagus) posterior and pyloric portion anterior in position; pyloric caeca about thirty, not well differentiated; length of intestine in preserved specimens about equal to standard length, forming a single loop upon leaving stomach, then passing straight to vent. Posterior chamber of swim bladder about one-half of standard length or six times length of anterior chamber, terminating in a fine taper which extends to above base of last anal fin ray. Adipose eye-fid thick and very strong, extending from immediately behind nostrils well onto gill cover, and with a narrow vertical sfit over the pupil (Fig. 1). Cranial fontanels as in chilodontids and other hemiodontids, anterior fontanel linear and narrow, posterior fontanel slightly wider (Fig. 2). Size, shape, and position of jaw bones as in Hemiodus; a peculiar fenestra in tooth-bearing portion of dentary (a similarly located fenestra present in Hemiodus, absent in fore- shortened dentary of chilodontids). Anterior end of ethmoid with small lateral knobs. Circumorbital series with simple antorbital, supraorbital, and full complement of six infraorbitals; first infra- orbital smaller than those succeeding it (enlarged in Chilodon- tidae), j^econd through fourth infraorbitals each slightly larger than the preceding one. Branchiostegal rays five (five in Hemiodus and Argonectes, four in Chilodus and Caenotropus); proximal end 4 BREVIORA No. 367 of fourth branchiostegal ray greatly expanded; hyoid bar general- ized (apparently highly specialized in Chilodontidae). Gill mem- branes free from isthmus, united to each other at a point below middle of eye (broadly united to isthmus in chilodontids); isthmus scaled anterior to cleithral symphysis (scaleless in chilodontids). Three postcleithra; third (lowermost) postcleithrum with a lamel- lar, posteriorly directed extension (as in Hemiodus). Weberian apparatus and caudal skeleton without unusual modifications. Forty vertebrae, including Weberian apparatus. IVIICROMISCHODUS, new genus Nomendatiiral type-species: M. sugillatus, new species Body fusiform and highly streamlined. Secondary sexual di- morphism unknown (specimens at hand collected in November and December, with unripe gonads). Cranial roof smooth. Sides of head largely covered by adipose eye-lids. Nares nontubular, close-set and separated only by a flap flush with surface of head. Tip of snout extends slightly beyond included lower jaw. With mouth fully opened, gape almost vertical and about as large as eye diameter. With mouth closed, dorsoposterior edge of maxillary slips under first infraorbital bone; maxillary not extending as far back as anterior margin of eye, but only to below posterior nostril. Toothless portion of lower jaw (posterior to rictal membrane) about four or five times longer than tooth-bearing portion. Articu- lation of lower jaw below middle of eye; rictal membrane below nostril, distinctly in front of anterior orbital rim. Lateral line complete, slightly decurved anteriorly, then running just below lateral midline of body to last scale row on caudal base. Origin of dorsal fin midway between snout tip and base of caudal fin. Anal fin small. Caudal fin deeply forked. Dorsal, anal, and median caudal fin rays Vv'ith well-developed, overlapping, mem- branous lappets or alae (Fig. 1 ). Similar structures occur in many fast-swimming lower teleosts, including various other characoids, some cyprinoids and clupeoids, and Chanos. Known by a single species from the Middle Amazon and lower Rio Negro in Brazil. Derivation of name. Micromischodus is from the Greek micro, small, little; mischos, stalk, petiole, peduncle; and odous (odon), odontos, masc, tooth. 1971 BRAZILIAN CHARACIN FISH 5 M. SUGILLATUS, new species Figure 1 Note. Standard lengths are used throughout this paper. Un- less stated otherwise, proportional measurements are expressed as times in standard length. Counts and measurements are given for the holotype first, followed in parentheses by the ranges for the paratypes. Holotype. MZUSP 6773, 152.0 mm, Igarape Tarumazinho, 15 km NW of Manaus, on the left bank of the Rio Negro, Estado do Amazonas, 17-18 November 1967. Paratypes. MZUSP 8870, six specimens 125.0-144.6 mm, same data as holotype; MZUSP 6691, four specimens 107.3-137.1 mm, Rio Negro, arredores de Manaus, Estado do Amazonas, 15-23 November 1967; MCZ 46718, two specimens 118.4 and 137.1 mm, Rio Canuma, Canuma (near Maues), Estado do Amazonas, 28-29 November 1967; MCZ 46719, seven specimens 79.5-108.9 mm, Igarape do Rio Jamari, Terra Santa, Estado do Para, 14 December 1967. Proportional measurements. Greatest body depth (at origin of dorsal fin) 4.1 (4.0-4.5). Greatest body width (below dorsal fin) 5.7 (5.6-6.6). Least depth of caudal peduncle 11.4 (11.3- 12.4). Distance from snout-tip to dorsal origin 2.01 (1.94-2.08). Length of caudal peduncle 7.45 (6.4-7.8). Head 3.4 (3.1-3.4). Eye 3.9 (3.4—4.1) in head. Bony interorbital space 3.8 (3.5-4.1) in head. Snout 3.5 (3.4-3.8) in head. Width of mouth (measured to outer sides of maxillaries) 4.1 (4.1-4.7). Fins. Fins, excepting caudal, scaleless. Dorsal fin rays 11 (11 in all paratypes except one with 10), first two rays simple, last ray divided to its base. Proximal two-thirds to three-quarters of each ray with well-developed membranous lappets or alae, largely overlapping when fin is depressed. Dorsal fin margin falcate. Height of dorsal fin 4.7 (4.3-4.8). Base of dorsal fin 9.2 (8.1-10.3). Anal fin small, about one-third as large as dorsal fin, with similar membranous lappets. Anal rays 10 (10), first two rays simple, last ray divided to its base. Height of anal fin 8.2 (7.4-8.3). Anal fin margin falcate. Base of anal fin 12.4 (11.8- 13.4). Caudal fin deeply forked. Lower caudal lobe slightly larger than upper caudal lobe, with its upper margin shghtly con- vex (Fig. 1). The lower caudal lobe is similarly modified in other BREVIORA No. 367 00 B 4) c a B B q CO 3 u a ■4-* 45 en a 60 s "n en a o tn s o u 3 oo 1971 BRAZILIAN CHARACIN FISH 7 fast-swimming characoids, for example in Hydrocynus goliath (Boulenger, 1899, pi. 11). Principal caudal rays 10 -f- 9. Median four to six rays with well-developed lappets. Procurrent rays moderately numerous (9 + 7 in 125-mm specimen), unspecial- ized. Adipose fin small, its base over posterior portion of anal fin base. Pectoral and pelvic fins with two or three peculiarly thickened and elongated scales forming a sheath for edge of outer- most rays. Pectoral fin rays 21 (18-22). Length of pectoral fin 6.3 (5.3-6.5). Pelvic fin rays 11 (11), outermost ray only un- branched. Length of pelvic fin 7.0 (6.1-7.2). Scales. Scales cycloid, or with laciniate (but not ctenoid) borders. Body completely scaled, 57 (56-64) scales in a lateral series; about 21 (16-22) predorsal scales (irregularly aligned); 11 (11 or 12) scales between lateral line and dorsal origin; 4 (4) scales between lateral line and pelvic insertion; 20 (17-22) scales between last dorsal ray and adipose fin; 15 (14-16) scales along midline between symphysis of cleithra and pelvic base; and 16 (16-18) scales around caudal peduncle. Prepelvic scales below lateral line gradually increasing in size ventralwards; scales on abdomen about two or three times larger in diameter than scales above lateral line. Caudal fin beyond hypural fan with two or three rows of scales. Coloration in alcohol. After two years in preservative (fixed in formalin while alive and transferred to ethyl alcohol within a few months) the specimens are straw colored, darkest above. The most notable feature is the solid black or blue-black color on the membranes between the distal halves of all of the dorsal fin rays (but not on the rays themselves), (The hemiodontid Argonectes scapularis and chilodontid Tylobronchiis maculosus have similarly marked dorsal fins. ) Pectoral, pelvic, anal, and adipose fins color- less; caudal fin dusky, lower lobe of caudal becoming darker (but not as dark as dorsal fin) near its tip. A faint longitudinal stripe running length of the body but stopping at shoulder, not running through eyes or onto caudal fin. Above lateral line fine melano- phores lie beneath the scales and in concentrations paralleling exposed scale margins. Below lateral line melanophores almost entirely absent in some specimens; in other specimens melano- phores are virtually absent beneath the scales but are present along the scale margins. The reticulated pattern of vertically oriented spots thus produced is similar to the pattern present in 8 BREVIORA No. 367 all Chilodontidae, except that since the scales are much smaller, the spots are small and not as dark. In all specimens body dis- tinctly lighter below lateral line than above. Gill cover marked by an indistinct dark oval patch caused by a concentration of melanophores in the membrane lining inside of gill cover, which shows through the opercular bones. The trivial name sugillatus (Latin: sugillo, -atus, beaten black and blue; sugiUatum, black and blue spot, bruise) refers to the coloration on the dorsal fin and opercles. Ecological notes. According to Sr. Britski, specimens of M. sugillatus were collected by seining along shore in lakes or places with slow-flowing, black or crystalline water. A very large number of species, predominantly characins, was obtained at these locali- ties. The stomach contents of specimens from the Igarape do Rio Jamari include an assortment of bottom material, including many of what appear to be droppings of small fishes. Very small insects are well represented, including larval Diptera and, most abundant, a corixid (identified by Dr. John Lawrence) about L5 mm long. These are about the largest organisms in the stomach contents. OSTEOLOGY Figures 2—15 Illustrations of the osteology of M. sugillatus are based on a 125.0-mm paratype from MZUSP 8870. Supplemental observa- tions were made on a 102.2-mm paratype from MCZ 46719. The nomenclature of bones follows Weitzman (1962) except that "vomer" and "intercalar" are used in place of "prevomer" and "opisthotic." Cranium (Figs. 2-4). Roofing bones of skull smooth and flat- tened. Cranial fontanels narrow, complete. Anterior fontanel linear and about half as wide as posterior fontanel for almost its entire length, but widening immediately in front of epiphyseal bar. Ethmoid narrow, with two lateral knobs anteriorly articulating with either premaxillary, ventro-lateral laminar projections immediately posterior to these knobs, and a median cleft in its posterior half (Fig. 2). Supraoccipital crest flat, not extending as far as end of cranium. Posttemporal fossae well developed. Dilator groove well developed, frontal participating in its formation; dorsal Umit of 1971 BRAZILIAN CHARACIN FISH antorbjtal nasa vomer ethmoid parietal epiphyseal bar lateral ethmoid frontal sphenotic epiotic exoccipital supraoccipital premaxillary infraorbital 6 Dterotic Figure 2. Cranium of Micromischodiis siigillatiis (dorsal view), osteological figures based on 125-mm specimen from MZUSP 8870. All dilator groove marked by a concavity in dorsolateral margin of frontal bone; sphenotic spine lamellar, rounded at tip. Ethmoid separated from vomer by a cartilaginous septum (in- dicated by heavy stippling in Figure 3). Vomer with peculiar anterior knoblike processes on either side. Lateral ethmoid rela- tively large, with elongate, strutlike process articulating with knob- hke process of vomer (Figs. 3 and 4) and a small median projec- tion firmly articulated to antero-ventral comer of rhinosphenoid. Rhinosphenoid large, sharing strongly interdigitating sutures with antero-ventral projection of orbitosphenoid (Fig. 3). Cartilaginous septum between dorsal edge of rhinosphenoid and roof of cranium and another between ventral margin of rhinosphenoid and para- sphenoid (cartilage indicated by heavy stippling in Figure 3). Rhinosphenoid and orbitosphenoid not directly contacting para- sphenoid. Parasphenoid deeply cleft posteriorly, the thin divisions thus formed cleft at their tips (Fig. 4; compare with Weitzman, 1962, fig. 4 of Brycon on p. 60; and Roberts, 1969, fig. 20 of Acestrorhynchus on p. 463). Intercalar well developed. Pterotic with strong posteriorly directed spine. Jaws and jaw suspension (Figs. 2, 5-7). The size and shape of the jaw bones and their relationships to each other are similar 10 BREVIORA No. 367 lateral ethmoid ^"^""^^l ^Phenotic pterotjc parietal supraocclpital ethmoid vomer rhinosphenoid orbitosphenoid pterosphenold parasphenoid Prootic '"20'; G = ground; R = rocks; N = sample size. Diameter Ht.^~~--^in.) (feet) >5 5-2 1/2 2 1/4-7/8 7/8-1/8 leaves Total N = 385 10.5-20 5-10 3-4 3/4 <3 Total N = 246 10.5-20 5-10 3-4 3/4 <3 Total N = 31 10.5-20 5-10 3-4 3/4 <3 Total N = 214 10.5-20 5-10 3-4 3/4 <3 Total male grahami H = 2.3 G = 0 R = 0 3.6 8.1 6.6 3.8 0 3.6 13.8 21.4 9.5 0.3 3.6 5.5 4.6 3.5 0 3.6 2.6 3.3 2.5 0 12.4 30.0 35.9 19.3 0.3 female- -sized grab ami H = 0 G = = 0.3 R 0.7 2.7 2.9 2.6 0 0.3 3.4 14.9 28.2 1.0 0.3 2.7 6.0 11.5 0 2.7 4.1 7.2 8.4 0 22.1 48.9 16.2 10.7 = 0 8.9 47.9 20.5 22.4 4.1 13.0 30.9 50.6 1.0 juvenile £r ah ami H = 0 G = 0 R = 3.2 0 0 0 0 0 0 0 0 25.8 0 0 0 4.8 37.1 3.2 0 0 6.5 19.4 0 0 25.8 45.1 25.9 0 0 11.3 male opalinus H = 0 1.9 4.2 2.3 4.2 21.0 21.3 2.3 7.0 6.3 1.9 5.6 3.3 10.3 37.8 33.2 82.3 ; = 0 1.4 12.4 3.5 0.5 17.8 3.2 R = 0 0 0.5 0 0 0.5 9.8 59.4 19.1 11.3 30 BREVIORA No. 368 Table 1. (cont'd) -~-~...,^^ Diameter Ht>-~~^in.) (feet)^---^ >5 5-2 1/2 2 1/4-7/8 7/8-1/8 leaves Total N = 235 female-sized opalinus H = 0 G = 0.9 ^ = 0 10.5-20 0.4 2.1 1.3 0.9 0 4,7 5-10 2.6 10.2 18.1 14.3 0 45.2 3-4 3/4 1.7 4.7 10.9 7.5 0 24.8 <3 3.4 8.1 8.3 4.9 0 24.7 Total 8.1 25.1 38.6 27.6 0 N = 15 juvenile opalinus H = 0 G = 0 R = 0 10.5-20 0 0 0 0 0 0 5-10 0 0 0 22.2 0 22.2 3-4 3/4 0 0 0 11.1 0 11.1 <3 0 11.1 22.2 33.3 0 66.6 Total 0 11.1 22.2 66.6 0 N - 838 male lineatop us H = 0.2 G = 0. 8 R = ( ) 10.5-20 0.4 1.1 1.8 0.1 0 3.4 5-10 2.6 13.4 22.6 13.0 0 51.6 3-4 3/4 2.1 7.2 14.1 9.5 0.1 33.0 <3 1.7 2.5 3.6 3.0 0 10.8 Total 6.8 24.2 42.1 25.6 0.1 K = 436 femal e-sized lineatopus H = 0 G = 4.4 R = 0 10.5-20 0 0 0 0 0 0 5-10 0.5 1.4 6.0 3.9 0.2 12.0 3-4 3/4 0.5 4.6 8.4 18.9 0 32.4 <3 1.4 8.3 16.3 24.8 0.7 51.5 Total 2.4 14.3 30.7 47.6 0.9 1971 JAMAICAN ANGLES 31 Table 1. (cont'd) Diameter Ht. -^in.) (feet) N = 24 10.5-20 5-10 3-4 3/4 <3 Total N = 15 10.5-20 5-10 3-4 3/4 <3 Total K = 16 10.5-20 5-10 3-4 3/4 <3 Total N = 3 10.5-20 5-10 3-4 3/4 <3 Total >5 5-2 1/2 2 1/4-7/8 7/8-1/8 leaves Total juvenile lineatopus H=0 G=12.5 R=0 0 0 0 0 0 4.2 0 8.3 0 0 0 2.1 0 0 8.8 56.3 0 12.5 male valencienni 0 6.7 0 6.7 0 0 0 0 2.1 64.6 H = 6.7 G = 0 6.7 13.3 6.7 33.3 0 13.3 0 6.7 66.6 6.3 25.1 12.6 juvenile valencienni 0 0 0 0 0 0 0 0 0 0 0 0 50.1 = 0 G 33.3 0 0 66.7 100.0 0 0 0 8.3 0 0 12.5 75.0 8.3 26.7 46.7 13.3 6.7 female- -sized val encienni H = 6 3 G = 0 0 6.3 6.3 12.5 0 0 12.5 6.3 6.3 0 6.3 6.3 0 12.5 0 0 0 0 18.8 0 25.1 25.1 25.1 18.8 0 33.3 0 0 66.7 32 BREVIORA No. 368 Table 2. Port Antonio Open. Percent observations in various structural habitat categories. H = >20'; G = ground; R = rocks; N = sample size. Diameter Ht>^~..,,.^(in.) (feet)^^^-,^ >5 5-2 1/2 2 1/4-7/8 7/8-1/8 leaves Total N = 146 male grahami H = 4.1 G = 1.4 R = 0 10.5-20 2.7 6.9 2.7 0 0 12.3 5-10 5.5 7.5 12.3 11.0 0 36.3 3-4 3/4 5.5 3.4 1.7 9.9 0.7 21.2 <3 4.8 2.1 6.8 8.9 2.1 24.6 Total 18.5 19.9 23.6 29.8 2.7 N = 163 female-sized < grahami H = 0.6 G = 2.5 R = 0 10.5-20 0.6 0.6 0.6 0 0 1.8 5-10 1.2 2.5 7.7 5.2 0 16.6 3-4 3/4 1.2 0 2.8 10.7 1.8 16.6 <3 3.7 2.5 4.6 34.7 16.6 62.0 Total 6.8 5.5 15.6 50.6 18.4 N = 75 juvenile grab ami H = 0 G = 0 R = 0 10.5-20 0 0 0 0 0 0 5-10 0 0 0 0 0 0 3-4 3/4 0 0 0 10.7 0 10.7 <3 0 0 1.3 57.3 30.7 89.4 Total 0 0 1.3 68.0 30.7 N = 17S male opalinus H = 1.1 G = 0 R = 0 10.5-20 0.6 1.7 10.8 1.7 0 14.8 5-10 5.7 15.9 12,8 4.3 0 38.6 3-4 3/4 6,3 11.4 5.4 6.0 0 29.0 <3 4.6 7.4 0.6 3.4 0.6 16.5 Total 17,1 36.4 29.6 15.3 0.6 1971 JAMAICAN ANGLES 33 Table 2. (cont'd) . -...,,^^ Diameter HtT~~-^-^4in . ) (feet) ^^-^-^^ >5 5-2 1/2 2 1/4-7/8 7/8-1/8 leaves Total N = 136 female -sized opa linus H = 0 G = = 0 R = 0 10.5-20 0 1.5 0 0 0 1.5 5-10 3.7 5.2 14.3 4.8 0.7 28.7 3-4 3/4 3.7 2.9 5.5 7.0 0.7 19.9 <3 2.9 6.6 11.0 27.9 1.5 50.0 Total 10.3 16.2 30.9 39.7 3.0 N = 21 juveni le opalinus H = 0 G = 0 R = 0 10.5-20 0 0 0 0 0 0 5-10 0 0 4.8 4.8 0 9.5 3-4 3/4 0 0 0 14.3 0 14.3 <3 0 0 9.5 66.7 0 76.2 Total 0 0 14.3 85.7 0 N = 183 male 1 ineatopus H = 0 G = 3.3 R = 0 10.5-20 0 0 0 0 0 0 5-10 0 1.1 1.6 3.8 0 6.6 3-4 3/4 2.2 6.6 5.2 13.4 0.6 27.9 <3 3.8 7.7 12.0 37.7 1.1 62.3 Total 6.0 15.3 18.9 54.9 1.6 N = 110 female -sized lineatopus H = 0 G = 11.8 R = 0 10.5-20 0 0 0 0 0 0 5-10 0 0 0 0 0 0 3-4 3/4 0 0 0.5 5.0 0.9 6.4 <3 0 0.9 9.1 63.6 8.2 81.8 Total 0 C.9 9.5 68.6 9.1 34 BREVIORA No. 368 Table 2. (cont'd) Htr-~-~-^n.) (feet) ^~~--.>.^ >5 5-2 1/2 2 1/4-7/8 7/8-1/8 leaves Total N = 23 juveni le lineatopus H = 0 G = 26.1 R = 0 10.5-20 0 0 0 0 0 0 5-10 0 0 0 0 0 0 3-4 3/4 0 0 0 0 0 0 <3 0 0 0 56.5 17.4 73.9 Total 0 0 0 56.5 17.4 N = 25 male valencienni H = 4. 0 G = 0 R = 0 10.5-20 8.0 12.0 16.0 0 0 36.0 5-10 8.0 12.0 12.0 8.0 0 40.0 3-4 3/4 4.0 4.0 0 0 0 8.0 <3 0 0 4.0 8.0 0 12.0 Total 20.0 28.0 32.0 16.0 0 N = 36 female -sized valencienni H = 2.8 G = 0 R = C 10.5-20 2.8 0 0 0 0 2.8 5-10 2.8 2.8 8.3 19.5 0 33.4 3-4 3/4 2.8 8.3 0 11.1 2.8 25.0 <3 2.8 0 0 33.3 0 — 36.1 Total 11.1 11.1 8.3 63.9 2.8 N = 3 juveni le valencienni H = 0 G = 0 R = 0 10.5-20 0 0 0 0 0 0 5-10 0 0 0 33.3 0 33.3 3-4 3/4 0 0 0 0 0 0 <3 0 0 0 66.7 0 — 66.7 Total 0 0 0 100.0 0 1971 JAMAICAN ANGLES 35 Table 3. Port Antonio Closed. Percent observations in various structural ha±>itat categories. H = >20'; G = ground; R = rocks; N = sample size. Diameter N = 185 male opalinus H = 1.6 G = 0. 5 R = 0 10.5-20 6.5 1.6 1.6 2.2 0 11.9 5-10 7.0 9.7 22.2 9.7 0 48.6 3-4 3/4 5.9 1.6 8.6 2.7 1.1 19.9 <3 1.6 2.7 9.2 3.2 0.5 17.2 Total 21.1 15.7 41.6 17.8 1.6 N = 206 fema] .e-sized opalinus H = 1.0 G = 2.4 R = 0 10.5-20 0.5 0.5 0.7 1.2 0 2.9 5-10 8.3 3.4 9.5 15.8 0 37.0 3-4 3/4 5.3 1.5 5.8 6.8 0.5 19.9 <3 6.3 3.9 9.5 15.8 1.5 37.0 Total 20.4 9.2 25.5 39.6 1.9 N = 79 juvenile opal inus H = 0 G = 0 R = 0 10.5-20 0 0 0 0 0 0 5-10 0 1.3 1.3 19.0 0 21.5 3-4 3/4 0 0 0 17.7 0 17.7 <3 0 0 9.5 50.0 1.3 60.8 Total 0 1.3 10.8 86.7 1.3 N = 122 male lineatopus H = 0 G = 0 R = 0 10.5-20 0 0 0 0 0 0 5-10 4.1 3.3 15.6 14.8 0 37.7 3-4 3/4 5.7 3.3 10.2 18.4 0.8 38.5 <3 4.1 3.3 9.0 7.4 0 23.8 Total 13.9 9.9 34.8 40.6 0.8 36 BREVIORA No. 368 Table 3 . (concl'd) . "--■-^^Diameter (feet) ^^^^ >5 5-2 1/2 2 1/4-7/8 7/8-1/8 lee ives 3.6 Total N = 112 female -sized lineatopus H = 0" G = R = 0 10.5-20 0 0 0 0 0 0 5-10 0 0.9 0.9 8.9 0 10.7 3-4 3/4 0 0.9 4.9 21.0 0. 9 27.7 <3 2.7 1.8 16.5 30.8 6. 3 58.0 Total 2.7 3.6 22,3 60.7 7 2 N = 32 juveni le lineatopus H = 0 G = 3.1 R = 0 10.5-20 0 0 0 0 0 0 5-10 0 0 0 6.3 0 6.3 3-4 3/4 0 0 0 6.3 0 6.3 <3 0 3.1 9.4 53.1 18 8 84.4 Total 0 3.1 9.4 65.7 18 8 N = 13 valencienni H = 0 G = 0 R = 0 10.5-20 7.7 7.7 15.4 7.7 0 38.5 5-10 0 0 15.4 15.4 0 30.8 3-4 3/4 15.4 7.7 0 7.7 0 30.8 <3 0 0 0 0 0 0 Total 23.1 15.4 30.8 30.8 0 197; JAMAICAN ANGLES 37 Table 4. whitehouse. Percent observations in various structural habitat categories. H = 20'; G ■= ground; R = rocks; N = sample size. Diameter >5 5-2 1/2 2 1/4-7/8 7/8-1/8 leaves Total Ht.^'^-.^dn.) (feet) N = 219 10.5-20 5-10 3-4 3/4 <3 Total N = 284 10.5-20 5-10 3-4 3/4 <3 Total N = 54 10.5-20 5-10 3-4 3/4 <3 Total N = 88 10.5-20 5-10 3-4 3/4 <3 Total male g rahami H = 1.4 G = 0.5 R = 0 1.8 1.8 2.3 1.4 0 2.7 9.6 21.5 9.1 2.3 0.5 7.3 10.1 4.1 0.5 1.8 10.1 7.1 4.3 0 6.8 28.8 41.0 18.9 2.8 fema] .e- -sized 2£ ahami H = 0 G = 2.5 R 0 0 0.4 0 7 0 0.7 2.8 14.1 21 8 4.6 1.8 1.8 5.8 7 9 2.5 1.4 9.9 12.3 7 7 1.1 7.3 45.2 22.5 28.3 = 0.4 1.1 44.0 19.8 32.4 2.9 14.5 32.6 38.1 8.2 juvenile gr ahami H = 0 G = 0 R = 0 0 0 0 0 0 0 0 3.7 38.9 0 0 1.9 1.9 31.5 0 0 0 6.5 13.9 1.9 0 42.6 35.3 22.3 1.9 12.1 J4.3 1.9 male opalinus H = 0 G = = 0 R = 0 0 0 0 0 0 1.1 6.8 18 2 4.6 0 2.3 13.6 15 9 1.1 0 1.1 20.5 9 1 4.6 1. 1 0 30.7 32.9 36.4 4.5 40.9 43.2 10.3 1.1 38 BREVIORA No. 368 Table 4. (cont'd) . ^"^^Diameter Ht7""^^^in.) (feet)^5 5-2 1/2 2 1/4-7/8 7/8-1/8 leaves Total N = 72 female -sized opalinus H = 0 G = 2.8 R = 0 10.5-20 0 0 0 0 0 0 5-10 0 5.6 7.0 1.4 0 14.0 3-4 3/4 2.8 4.2 16.7 1.4 0 25.1 <3 2.8 31.9 19.5 4.2 0 58.4 Total 5.6 41.7 43.2 7.0 0 N = 10 juveni le opal inus H = 0 G = 0 R = 0 10.5-20 0 0 0 0 0 0 5-10 0 0 10.0 20.0 0 30.0 3-4 3/4 0 10.0 10.0 20.0 0 40.0 <3 0 10.0 10.0 10.0 0 30.0 Total 0 20.0 30.0 50. D 0 N = 263 male sagrei H = 0 G = 6.5 R = 2.7 10.5-20 0 0 0 0 0 0 5-10 0 0.8 1.5 0 0 2.3 3-4 3/4 0 6.5 8.4 2.3 0 17.2 <3 2.3 30.0 27.8 11.4 0 71.5 Total 2.3 37.3 37.7 13.7 0 N = 393 female -sized sagrei H = 0 G = 13.2 R = 4.8 10.5-20 0 0 0 0 0 0 5-10 0 0 0.8 0.3 0 1.1 3-4 3/4 0 1.6 1.3 0.3 0 3.2 <3 2.5 31.8 22.5 21.0 0 77.8 Total 2.5 33.4 24.6 21.6 0 1971 JAMAICAN ANGLES 39 Table 4. (concl'd) . Diameter (feet) N = 251 10.5-20 5-10 3-4 3/4 <3 Total N = 7 10.5-20 5-10 3-4 3/4 <3 Total N = 8 10.5-20 5-10 3-4 3/4 <3 Total N = 7 10.5-20 5-10 3-4 3/4 <3 Total >5 5-2 1/2 2 1/4-7/8 7/8-1/8 leaves Total juveni le sag rei H = 0 G = 31 9 R = 4 0 0 0 0 0 0 0 0 8 1.6 0 0 0.8 0 4 2.0 0 2.0 17.8 12 6 25.6 0. 4 29.2 14.3 0 42.9 42.9 female-sized valencienni H = 0 0 0 0 0 0 0 0 0 0 0 37.5 0 0 25.0 0 37.5 37.5 62.5 0.4 G = 0 0 0 0 0 0 0 0 2.4 3.2 58.4 ma] Le va lencienni H = 0 G = 0 R = 0 0 0 0 0 0 14 3 0 28.6 42.9 0 0 0 14.3 0 0 0 0 0 0 0 juvenile valencienni H = 0 G = = 0 R = 0 0 0 0 0 0 0 42.9 14 3 0 0 0 0 28 6 0 0 0 14.3 0 0 0 85.8 14.3 0 R = 0 0 25.0 37.5 37.5 0 0 57.2 28.6 14.3 57.2 42.9 40 BREVIORA No. 368 Table 5. Percent observations for Whitehouse species in climatic categories. Sun Shade Clouds male sagrei 17.4 74.4 8.3 female-sized sagrei 15.9 73.8 10.2 juvenile sagrei 33.8 58.8 7.4 male grahami 12.4 74.6 12.9 female-sized grahami 20.5 61.4 18.1 iuvenile grahami 15.7 66.7 17.6 male opalinus 3.5 72.1 24.4 small opalinus 14.5 66.3 19.3 valencienni 36.4 50.0 13.6 1971 JAMAICAN ANGLES 41 TUUaTOUa-[BA P3ZTS -aXBuraj p TuuaxouaxBA 10 c 0 (0 •H snuTxedo XX^uis 10 a % p snuTxedo 3XT"9AnC n D> « O rH c > 3 w a ftj o •p ID c •H 73 Q) N « •H 3 to a I o o -P (0 0) 8 -H to 3 a o -p ro- Qi c to 3 •0 C T3 'O •rl ■H OJ -rl rH c N C (T5 c •H C 0) tn a Q) CO a; rH 3 0 •H 1 -H H C U d) o C •iH r-\ c rH c ID a iH e rH 3 (i R ID (U m r-i O w > <*A > 42 BREVIORA TuuaTouaxEA pazTS -aXBuiaj p TuuaxouaxBA snuT-[Bdo xiBuis p snuTxedo __9Xtus»AnC sndoieauTx c o u « fH Si « sndot^BauTx pazTS -aXBuiaj p sndo^eauTx aXTU3i^£i£ Tureqeab TureqBab pazTS -aXBUiaj p TureqBjB N CO N N 2 N M M N m 03 CQ iH O m o o < CQ N tq O O O O rH O O O O O O 03 CQ O r-l < O t-i T3 01 N •H (0 O E iH (3 10 x: u tT>l M-i OT CP 4> f-l •H c > 3 (0 3 a o ■p IT] OJ c: 0) N to •H 3 (0 a I o OJ 4-1 iH (0 n) 0) e c 3 a (u O -H 4J -H to 3 c •H o (0 a o No. 368 -3 i« C3 > f u Si^ « G rt 2 — c ^ "*-■ o ^ o _ c: a o o •— ■«-» ■*-• o C3 =« ^ -y) a< u h .S: 1) o D. •- '■^ u. DQ II ^ iH *• ^ u • ' x: 22 o a u ■ga O 1^ o -c _ U CQ ?i ^ rH i« 2 ^_^ (U CQ II B ^^ a i-l N ^ , r. ^^ 5^ "rt ^ 5 S « — c CO rH ^ II m ^ CQ ^^ cS ., 5 c/l rH cant rche tC > CO w C 1/5 - . ^- o ^ o E g (0 e en T3 H flj -H c: N C c •rH C U-l > c .ii II - "^ on CO CO B -^ m M S V4 M •H QJ -H •H -1 0 Oi| i*-i cH cH rH in rH iH -r-i o 10 E •H •H c c c c w > 44 BREVIORA No. 368 TuuaxouaxEA XI GUIS p TUuaxouaxBA snuTXBdo XTSUJS p snuTx^do aXTuaAnC sndo:;BauTx sndo:;BauTx pazTS -aXBuiaj p sndot^Bauxx u a o u H n) aX_£uaAnC TuiBqeaB TuiBqBJb pazTS -aXBuiaj p TUIBqBjB Si IT) 0) N •H M I • 0) .-H e 0) to N N ^3 en 5 O r-K < < < O rH .H iH •H c > 3 re u 3 a o +j (1) c •rH T3 01 N W •H 3 m D- I O Q) +J rH (TJ 53 -H >4-l rH ta 3 c o ■p o c •H e > Ui u 5 !3 5 .HO rH O O O < 5 O rH rH O O H •rt C C C c 0) (U H ■H O o c .H C 0) rH 01 rH (0 r^ «J E "0 > w > o .^ *" -o *- c a. c/) (U a, 0$ in rt ^ s t/1 c o > #. - . 'r: 5 ^ u Si ■? n. •- ^ =" o — w 0) ■4-J o E E 5 •5 II - C/5 Z c^ 1971 JAMAICAN ANGLES 45 (A C o w ■H u m o. B O o ■a 01 Ui o o o •H c o 4-1 c < 4-1 0 (X o dJ u c u c o •H 4J w •H 4-1 (0 4J Eh TuuaTouaxBA ^sXj[uaAn£ sndoieauTx sndo:;BauTX pazTS -aXBuiaj p sndo^BauTX ajTuaAnC snuTx^do snufxedo pazTs -axBuiaj p snuTXBdo * * -a •0 T3 0) 01 N to N (0 W •H 3 •H 3 3 (0 tn (A 0) a w a D. 0) 1 3 3 rH 0 1 0 O •-t o c C ■H +j OJ -P 4J •H rH -H •H c (C rH (0 fl C (0 i-H iH OJ OJ (0 c e c c > 46 BREVIORA No. 368 juuaTouaxBA sridoieauf X sndo:^BauTX pazTs -a-[Bura3 p snSo^BaTrfx aiTuaAnC ^ * pj- snuTXBclo snuTXBdo pazTs -axBuiaj p snujx'sdo t) o c o o eo V <-* Xi 10 N t3 pa m u (0 > 9> 3 4J C CuO) •H B§ r-l M-H a tr-d 01 1? « •0 S 0) N (0 N n 10 •H 3 •H 3 3 0) (0 (0 a> & (0 a a c a) -H C > a a a 3 •H •H 3 tt-i o 0 •r-> f-l Ui lU eo u« i2 c« •H 3 •H (0 tn 01 1 (1) -H ■H lU m in tn •H (0 x: 10 en •a (U N •H in I -H (1) B IT3 X E (0 -y IS P.M 9 0 0 W n c / • Oi-H / M / > ^1 / (U / +J +J / 4= 3 rH O U U O O rH O O 03 03 c TJ (U OJ > N 3 I -H (U E r-l (3 TO u cr i*-i t7H Cr> 03 CQ n n 03 n r-l iH iH r-i o o o o 5 5 t-i o o m o o o o o < 03 O HO iHOOOO 10 E I 1971 JAMAICAN ANGLES 49 JUU3JOUaXBA snuTxedo XT'^uis p snuTXBdo 9X_TuaAn£ TureqBjB "fureqioB pazTS -aXBuraj p TureiiBa6 aXJu3An£ TaaBes T) P3ZTS -aXBuraj p xaxEiis o o o o o o o o o o o o o o o o o 3 3 CM ro o o o o 3 3 rH n O O O S S 1^ h3 to rH O rH O iH O O S3 £ S 3 3 rHiHO rHr-lrHOiH (0 •a N •H (0 I 0) -H 10 n •H c > 3 (0 0) N •H tn I -H l iM cri -H C > (0 IH (0 10 a o 1) O tn ii _' 1J 3 5 Ml c _o t> i-i c CO c « •a o o •- p; ITS -r E 5 •- 3 « t; £^^ ^ c ^ - -J ^ (U CO C3 W) '^ « 4J CO j:: o CO a s: Ml •3 C 1) — C CO 00 1) X3 3 C C o CO *-• c ■rr O 13 C CO C S '^ • ' p— r- 5 •5 = c II « + O) 73 c CO c 3 b'll > •' I- u U J^ CSlj o CO aJ — Q, CO S J= O D. f O CO -O CO u •a •= CO -S f. *- .3 U (« CJ IE c CO c 3 c« M C — 1^ OJ c CO tn In *-• c CO a' CJ -a .H C CO a> 2 "^ 9- "5 c en 4J O II — c lo Ji E ° tn E II II < Q C en « 1 *J Q) / 0) jG / (3 c / •H -H / -o — vo in 00 ro vo r- V£> vo xr CM r» vo in (N xr V \ n CM ro V V T3" \ V. \ .H ■-( m ^ (N T 00 m 00 00 CO TT »» \ \ \ \ \ r>- m in n n c T) Q) "O •0 N 3 w N tn W •rH •n 3 •H 3 3 (0 a (n a a (U w •■H 1 -H •H o 1 0 o rH 3 1 (1) E c +j O AJ +j ■r( C rH S (3 ffl rH (0 (0 c •H £ (0 £. £ 0) (0 Q) Q) (U <-i m B n) m c ^ <=: C > <0 M 9 V4 h •H S -H •H 3 a w 'u m w rH IW iH rH rn ol (0 o (0 E •o r4 i-t 00 1^ 00 in 00 in CO in ■* 00 "a* V '9' \ \ \ \. ■V •-* in ro n ro T3 H 0) -rH C N C c •H C (U (0 0) H 1 -H o 0) U c rH C 0) m 0) iH e •-^ (0 9 Id > MH > 1971 JAMAICAN ANGLES 51 TUU3JOUaXBA _aXiuaAn£ sndoieauTX sndo:;B3UTX pazTS TJ p sndo^BSUTT 0) "^ a O i-i u o •H c o ■p 4J o 01 aTT"9AnC snuTxeao snuTxedo pazTs -aXBuraj p snujXBdo V 01 CO (0 3 O 4J rO 0) C (U N (0 •H 3 m a I 0 0) -P rH nj 10 Q) 00 in 00 in C > 3 00 00 in m 1-1 00 ID 00 in ro 00 \ in 00 00 •H m c 3 C O 0) 0 •H +J O (0 c 0) 0) c I-I •H 10 r-i > 52 BREVIORA No. 368 TuuaxouaxBA XT^uis p IUU3TOUaiBA snuT-[Bdo XT^u^s c s- o •H c o ■p u o 04 p snuTXBdo 9XTU3AnC sndo:iBauTX sndo:iB3UTX pazTs -aXBuragp p sndo:iBauTX ajQUBAnC Tureqeab pazTS -aXBuraj p ■tureqBJb +)^ / J3 -P / cr 0) / ■H Q) •o •o N 3 w N tn tn •H •r-1 3 •H 3 D W a w a a OJ w •H 1 -iH •H o 1 o o rH 3 F 0) g c 4J Q) +J 4J •H C (0 rH (0 (9 IT] i-l (0 (D c •H x: (0 £ jC <1) m (u 0) (U r-\ (0 E «J 10 c E c C > 10 M 0) u M •H S ■•H •H 3 U cH M-4 W cri fH <*-! rH i-H ■n o 10 o e u 00 00 00 00 \ \ \ \ in m in in 00 CO 00 00 00 en 00 00 00 00 \ \ \ \ \ \ \ •X m ro n n r~ r~ r~- ro ■H -H C c C c (U (1) •H ■H C) u c ^ c <0 ,-i OJ 1-i (0 rH 10 e fl > w > 1971 JAMAICAN ANGLES 53 0) n O Xi 0) ■p •H o c o u Si Eh TuuaTouaxBA snuTXBdo XTCUis p snujx^do aXTU9AnC Tureq¥j5 TureiiBab PSZTS -aXBurag p TureqBj5 ariuaAnC ^-1 TaxbBS xaa&Bs pazTS -aXBuiaj p TajBBS fo vo ^ V f T ^ T V, \ \ \ \ \ ro n rH ro .-1 ro a pointed posteriorly, although they deviated somewhat by orienting themselves along an axis extending from the zone of erosion toward the nearest growth zone. The reason for this is obvious. Should a lithophage burrow perpendicularly to the antero- posterior axis and only in the region of the mucro, it would risk having its surrounding substrate eroded away. By burrowing from the zone of erosion toward a growth zone, the lithophage is assured of an increasing substrate in which to bore. This explanation clar- ifies the otherwise haphazard orientation of the burrows in the posterior valve, where the oldest shell material is near the center, not at the posterior edge (Fig. 2). As the lithophage increases in size, it faces problems caused by the restricted space in which it can grow. Although some Lithophaga enlarge their burrows dorsally, most penetrate ventrally and eventually reach the mantle of the chiton. When the latter situation occurs, the chiton secretes a thin calcareous shield in an effort to contain the intrusion of the lithophage. Sometimes more than half of the ventral portion of the lithophage is situated below the ventral shell-plate surface. In spite of the efforts of the Lith- ophaga to increase in size and the chiton's effort to contain it, the lithophage soon reaches a point where further growth is impossible. Whether this represents a truly stenomorphic condition or not is uncertain, because we do not know if the L. aristata we observed ever reached sexual maturity. Lithophaga aristata is known to attain a length of 52 mm (Turner and Boss, 1962); the largest specimen extracted from a C. stokesii was 9.0 mm, while the aver- age length was about 7.0 mm. There is evidence that at least a few L. aristata outgrow their restricted polyplacophoran substrate (Fig. 6). Some of the largest burrows that we examined were exposed along their entire dorsal surface, indicating that the lithophages may have fallen from their 6 BREVIORA No. 369 burrows. It is probable that specimens that outgrow the chiton perish in the external environment.^ The effect of numerous Lithophaga in the shell-plates must be detrimental to Polyplacophora. In addition to damaging the spe- cialized portion of the nervous system that makes up the aesthetes in the tegmentum, Lithophaga weakens the valves and, at times, probably affects the maneuverability of the chiton. The seriousness of damage to a large amount of the tegmental nervous tissue is open to question and may depend upon the species involved. Aes- thetes are photosensitive, being activated both by light of constant intensity and by a decrease of light intensity (Arey and Crozier, 1919). Studies on the growth and behaviour of Chiton tubercu- latus indicated that normal erosion of shell-plates and concomitant loss of tegmental aesthete photosensitivity produced at least a partial inversion in its phototropic response. While young C. tiiber- ciilatiis are photonegative, older individuals may be irresponsive or photopositive (Arey and Crozier, 1919; Crozier and Arey, 1920), although Glynn (personal communication) found older C. tubercu- latus from Puerto Rico also to be photonegative. It appears, then, that Lithophaga mainly affects the durability of the shell, rendering the chiton more susceptible to predation. REFERENCES CITED Arey, L. B., and W. J. Crozier. 1919. The sensory responses of Chiton. J. Exper. Zoo!., 29: 157-260. Brucker, a. 1897. Sur un noiivel Acarien marin. C. R. See. Biol., 49: 632-633. Brucker. A., and E. Trouessart. 1900. Seconde note sur un Acarien marin (Halacaride), parasite de VAcanthocliiton porosiis. C. R. Soc. Biol.. 52: 107 109. Crozier. W. J., and L. B. Arey. 1920. On the ethology of Cliiton tiiber- ctilatiis. Proc. Natl. Acad. Sci.. 5: 496-498. 1 Although Hodgkin (1962) maintained L. pitimiila kelseyi Hertlein and Strong, 1946 outside their burrows for over one year under laboratory con- ditions. Otter (1937) was unsuccessful in his attempts to rear L. cumingi- ana (Reeve, 1857) and L. teres (Philippi. 1846) that had been removed from their burrows. It is doubtful if Lilliophaga could survive out of its burrow under exposed conditions. 1971 LITHOPHAGA IN CHITONS 7 Glynn, P. W. 1968. Ecological studies on the associations of chitons in Puerto Rico, with special reference to sphaeromid isopods. Bull. Mar. Sci., 18: 572-626. Helfman, E. S. 1968. A ctenostomatous ectoproct epizoic on the chiton Ischnochiton mertensii. Veliger, 10: 290-291. Hodgkin, N. M. 1962. Limestone boring by the mytilid Lithophaga. Veliger, 4: 123-129, 3 pis. MacGinitie, G. E., and N. MacGinitie. 1968. Notes on Cryptochiton stelleri (Middendorff, 1846). Veliger, 11: 59-61, pi. 6. Menzies, R. J., AND P. W. Glynn. 1968. The common marine isopod Crustacea of Puerto Rico. A handbook for marine biologists. Studies Fauna Curasao, 27: 1-133. Otter, G. W. 1937. Rock-destroying organisms in relation to coral reefs. British Mus. (Nat. Hist.) Great Barrier Expedition 1928-29, Sci. Repts., 1: 323-352, 6 pis. Soot-Ryen, T. 1955. A report on the family Mytilidae (Pelecypoda). Allan Hancock Pacific Expeditions. 20: 1-154, 10 pis. Turner, R. D., and K. J. Boss. 1962. The genus Lithophaga in the western Atlantic. Johnsonia, 4: 81-116, pis. 57-75. Webster, S. K. 1968. An investigation of the commensals of Crypto- chiton stelleri (Middendorff, 1846) in the Monterey Peninsula area, California. Veliger. 11: 121-125. BREVIORA No. 369 1971 LITHOPHAGA IN CHITONS Figure 1. Radiograph of disarticulated shell-plates from a specimen of Chiton stokesii from Culebra Island. Panama Bay, Canal Zone (MCZ 277122), showing a high degree of infestation by Lithopliaga aris- tata (0.84 x ). Figure 2. Enlargement of posterior valve of Fig. 1 (3.2 X ). Figure 3. Typical siphonal opening of L. aristata burrow in an uneroded portion of a C. stokesii shell-plate (12 X )• Figure 4. Intervening shell-plate musculature located between valves IV and V in Fig. 1, showing damage resulting from penetration of Lithophaga (9.2 X). Figure 5. Example of L. aristata (MCZ 277123) in shell-plate cross section of C. stokesii from Punta Mala, Panama (3.9 x )• 10 BREVIORA No. 369 Figure 6. Exposed burrow of Lithophaga aristata in Chiton stokesii from Panama City, Panama (MCZ 78821) (4.2 x ). Figures 7-8. L. aristata extracted from burrow illustrated in Fig. 5 (MCZ 277123) (6.5 x ). BREVIORA Miaseuni of Comparative Zoology Cambridge, Mass. 31 March, 1971 Number 370 ECOLOGICAL OBSERVATIONS ON A LITTLE KNOWN SOUTH AMERICAN ANOLE: TROPIDODACTYLUS ONCA James P. Collins^ Abstract. The little known anole Tropidodactylus onca on the island of Margarita is typically restricted to belts of low xerophytic vegetation adjacent to the open sandy area of natural beaches. Most animals are found on the ground or in low bushes and occur up to a height of 30.0 cm. They are poor climbers and will occasionally escape into holes made by ghost crabs. INTRODUCTION A brief visit (from July 8 to July 21, 1968) to the Fundacion La Salle on Margarita Island, Venezuela, afforded me an oppor- timity to collect and observe a little known anole, Tropidodactylus onca. Margarita is a continental island approximately twenty-one miles off the district of Sucre on the northern coast of Venezuela. It is approximately two hundred air miles northeast of Caracas. Tropidodactylus onca is a speciaUzed derivative of Anolis but with keeled scales underneath the digits instead of the expanded digital pads with adhesive lamellae so characteristic of the latter primarily arboreal genus. The difference in morphology should be reflected in ecology, but there have been no detailed reports on the habits, habitat, or even color in Ufe of T. onca. This paper attempts to remedy this gap in information. Description and color in life. Tropidodactylus onca is a rela- tively large anole. The largest of the specimens collected is 75 mm snout- vent length. The tail is round with no dorsal crest and is approximately equal to the snout-vent length of the animal. The most distinctive specialization of the anoles in general, the dewlap, is extremely well developed in this genus. It has a bright yellow ground color in which individual red scales are embedded. 1 Museum of Zoology, University of Michigan, Ann Arbor. Mich. BREVIORA No. 370 The dorsal color of the lizard varies from a very light ashy- gray to a dark gray-brown with a disruptive pattern of black and white longitudinal markings. These vary from barely visible to very prominent. The ventral surface of the lizard is white. The species is not sexually dichromatic and is very difficult to sex externally. The cryptic coloration of the species is perfect. In the field the animals are indistinguishable from their sur- roundings. In fact, an animal can usually be located only when the collector inadvertently frightens it into movement. Habits and habitat. Tropidodactylus onca is a beach anole. Its distribution is typically restricted to the belt of low xerophy- tic vegetation adjacent to the open sandy area of a natural beach. The exact width of this belt and its continuity varies according to the geological and ecological factors of the particular locality. The animal was never observed in the open sandy areas devoid of vegetation, and penetrates only slightly into the more land- ward areas where the vegetation is higher than 0.75 to 1.00 m. 20 "T" 10 64" 50 10 50'- 40 LOS FRAILES El Aguo Punta Montadaro Playa Guacuco ISLA CUBAGUA ISLA COCHE PENINSULA DE ARAYA (0' SO Figure 1. Margarita Island. Localities at which Tropidodactylus onca was observed or collected are shown as solid circles. 1971 TROPIDODACTYLUS ONCA 3 All the specimens, with the exception of one taken at a height of 1.25 m, were collected on the ground or up to 30.0 cm high on low bushes, vines, etc. Observation and collecting was restricted to six localities (Fig. 1), all on the eastern half of the island. Roze (1964), however, has reported the lizard from the western half (Macanao). Of the six localities in which the species was observed, five are on the coast. The exception, El Aguila, is a town approximately 2.5 km from Punta de Piedras. Possible reasons for the animal's appearance here will be discussed later. Outside of this single exception, no specimens were ever observed in noncoastal por- tions of the island and all were collected at sea level. The vegetation of the zone in which Tropidodactylus is found consists of only low ground growth. The sand-shrub communities of the five coastal localities in which Tropidodactylus is found each contain at least two and as many as four of the following species of plant: Sesuvium portidacastrum, Philoxerus vermicu- laris, Bat is maritima, Sporobolus (virginicus?), Ipomoea (pescap- rae?), and Mallotonia gnaphalodes. Howard (1950) lists these six species as pan-Caribbean in distribution. These are found in association with less abundant and restricted species, the exact taxa varying from locality to locality. The landward edge of this coastal community typically contains representatives of the genera Opuntia, Melocactus, Lemairocereus, Philoxocereus, and Proso- phis. Tropidodactylus penetrates slightly into this zone. The majority of the animals collected were taken in large patches of Ipomoea found at two of the localities investigated (El Agua and Playa Guacuco). The reptile was found both on the plants and in the open space between them. If the patch was dense enough, the animals were typically found with their head on top of the leaves and their trunk and tail among the tangled vines. Some were also found on an occasional outcropping of rock or log. When frightened, the animal would either duck into the mat of vegetation (if dense enough) or scamper across the open sand until it reached a vine or series of vines which it would then proceed to run clumsily over. Being nonwoody, the plants do not give much support and the animal could easily be taken. In the area just north of Punta Montadero where Mallotonia, a woody-stem plant, is dominant, the animal's behavior was very different. Here, when first sighted, the lizard was always on the BREVIORA No. 370 > o c o e c 0) -it: O ik_ JD x> c o ^ -a -•— c (U o JD C/5 O XI c >> o SI Cl If) o ~ w V 0) ^ X in {/) "oj to TD C c o Q) (/) -s: o T) V- C JD o o 3 5 ri 1971 TROPIDODACTYLUS ONCA 5 ground. When pursued, the majority of animals observed would merely run among the ground cover. A few specimens, however, were observed to climb the Mallotonia, some to a height of 30.0 cm. Their climbing was clumsy and ineffective. The toe structure of this genus is not well adapted for tree climbing. Unlike most anoles, T. onca is not arboreal. Another means of retreat should also be pointed out. At times, a specimen, being pursued, would run into a large hole in the sand opening into a tunnel. Ruthven (1922) also reports this species as escaping into holes. It should be noted, however, that these holes are resting places made by ghost crabs (Ocypode) and are not dug by Tropidodactylus. It should also be noted that this was a rather infrequent means of escape, used by the lizard only when almost completely exhausted. At each of the six localities, T. onca is found sympatric with Cnemidophorus lemniscatus lemniscatus. In those localities (El Agua and Punta los Cocos) where the landward border of the coastal area is occupied by a semi-desertic community, the terri- tory of Tropidodactylus partially overlaps that of Tropidurus tor- quatus hispidus. The Tropidodactylus penetrate this zone for only a very small distance. In some areas, two other organisms also found sympatric with Tropidodactylus are the gecko Gonatodes vitatus vitatus and the microteiid Gymnopthalmus laevicauda. Roze (1964) has the following note concerning the diet of Tropidodactylus: "The stomach contents examined in various specimens of this species revealed the remains of grasshoppers (Grillidae), Coleoptera, spiders, and various species of Diptera, as well as the remains of other unidentifiable arthropods." Tropido- dactylus then, hke most anoles, is insectivorous. Similarly, like most anoles, the animal is diurnal in its activity. All but one of the twenty-five specimens were captured during the day. The single exception was collected alongside the road near the town El Aguila approximately 2.5 km from Punta de Piedras. It was on a branch of a low bush, Jatropha gossypiifolia, in the cleared margin alongside the road. The animal was in typical anole sleeping posture, snout toward the main stem, but with its eyes open. Just prior to being seized, the animal moved its head but did not attempt to flee. This single exception to the otherwise complete coastal distribution of the animals on the island most probably migrated to this inland area along the corridor of low vegetation bordering either side of the roadway. This habitat is ecologically similar to that of the coastal zone. 6 BREVIORA No. 370 ACKNOWLEDGMENTS I am grateful to Dr. Janis Roze and Dr. Ernest E. Williams for critically reading the manuscript and for their valuable sugges- tions. This study was carried out at the Fundacion La Salle, Mar- garita Island, Venezuela; I thank Hermano Gines for making the facilities there available to me. Field expenses were partially met by NSF-GY-4183, administered by Manhattan College, and NSF- GB-6944 to Ernest E. Williams. REFERENCES Howard, R. A. 1950. Vegetation of the Bimini island group. Ecol. Mono. 20: 317-349. Roze, J. A. 1964. La herpetologia de la Isia de Margarita, Venezuela. Mem. Soc. Cien. Nat. La Salle, 24 (69): 209-241. RuTHVEN, A. G. 1922. The amphibians and reptiles of the Sierra Nevada de Santa Marta, Colombia. Miscellaneous PubUcation No. 8, Museum of Zoology, University of Michigan. BREVIORA Miaseiuioi of Comparative Zoology Cambridge, Mass. 31 March, 1971 Number 371 A NEW SPECIES OF BROMELIAD-INHABITING GALLIWASP (SAURIA: ANGUIDAE) FROM JAMAICA Albert Schwartz^ Abstract. A new species of anguid lizard, Diploglossus fowleri, is described from two specimens collected from bromeliads at the northern edge of Jamaica's Cockpit Country. The affinities of the new species are with D. hewardi and D. diiqiiesneyi; both D. fowleri and D. duquesneyi appear to be geographic or ecological isolates of the widespread D. hewardi. The Antillean islands of Jamaica and Hispaniola have excep- tionally large numbers of species of the anguid lizard genus Diploglossus Wiegmann. The latter island has six extant species, whereas Jamaica likewise had six species of which one {occiduus Shaw) is presently considered extinct. Cousens (1956) summar- ized the then-known Jamaican galliwasps and regarded cnisculus Carman, harbour i Grant, hewardi Gray, and duquesneyi Grant as valid species. Since that time, D. microblepharis Underwood has been named from a single specimen from the northeastern Jamaican coast. Cousens (1956), followmg Grant (1940b), sep- arated the four forms then recognized into two major groups: one group {crusculus, barbouri) with short legs and the other {hewardi, duquesneyi) with long legs. Schwartz (1970), in dis- cussing D. occiduus, suggested that the species crusculus-hewardi- barbouri-occiduus might represent a phylogenetic series, despite the interposition in this sequence of both long- and short-limbed species. D. microblepharis stands alone; its relationships are with the Puerto Rican D. pleei Dumeril and Bibron and the Cuban D. delasagra Cocteau. In the summer of 1961, while cutting bromeliads in the decid- uous forest at the northern edge of Jamaica's Cockpit Country, 1 Miami-Dade Junior College, Miami, Florida 33167 2 BREVIORA No. 371 the extensive karst region in northwestern Jamaica, we secured a single immature galUwasp. Despite the peculiar habitat (no Antil- lean Diploglossus had ever been recorded from bromehads) the Hzard bore resemblances to D. hewardi, and it was so considered in the field. Not until 1969, when the paper dealing with D. occiduus (Schwartz, 1970) was in preparation, was the lizard re-examined in a routine study of D. hewardi for comparative purposes. At that time, the difference in scutellation between D. hewardi and the 1961 juvenile specimen quickly became appar- ent. Further examination of the specimen indicated that, although it resembled D. hewardi in general (being a long-limbed form), it differed chromatically and in pattern from that species. But, since it was immature, no further course of action was planned. It was thus with great pleasure that I accepted the invitation of Dr. Thomas H. Patton of the Florida State Museum to visit Jamaica and stay at Worthy Park Estate during August 1970. Although it was hardly likely that, even with persistent bromeliad cutting, we would encounter another specimen of the arboreal galliwasp, plans were made to revisit the site of capture of the first individual (Windsor, Trelawny Parish). Thanks to the efforts of my assistants and native help, we were successful in securing another and adult specimen of the same form. Study of both indi- viduals convinces me that they represent a new species, related to D. hewardi, which has apparently taken to a bromeliad-inhabiting niche — a niche that is virtually unoccupied by Antillean reptiles. In the summer of 1961, I had the capable assistance of Ronald F. Klinikowski and David C. Leber. Our activities were facilitated by C. Bernard Lewis of the Institute of Jamaica. The 1970 trip was made both pleasant and profitable by the presence of Dale E. Becker, Michael T. Felix, and Danny C. Fowler, whose energy expenditures in bromeliad cutting were noteworthy. In addition, I have examined specimens collected by Richard Thomas in 1967, and by Robert Brenner and Paul Moravec in the same year. All specimens are in the Albert Schwartz Field Series (ASFS) with the exception of the holotype of the new taxon and two speci- mens of D. duqiiesneyi, which are in the Museum of Comparative Zoology (MCZ) at Harvard University. All measurements are in millimeters and color designations are from Maerz and Paul (1950). I am especially grateful to Dr. Patton for making the Worthy Park facilities available to us, and to Dr. Ernest E. Wil- liams for the loan of the holotype of D. duquesneyi. 1971 BROMELIAD GALLIWASP 3 In honor of Danny C. Fowler, whose endeavors on my behalf can only be recognized in a token fashion by associating his name patronymically with the species, I propose that this bromeliad- inhabiting galliwasp be called Diploglossus fowleri, new species Holotype. MCZ 125601, a female, from Windsor, elevation about 500 feet (153 meters), Trelawny Parish, Jamaica, taken 15 August 1970 by Danny C. Fowler. Original number ASFS V19902. Paratype. ASFS 14421, same data as holotype, 12 July 1961, D. C. Leber. Diagnosis. An apparently moderately sized (only known adult 105 mm snout-vent length), long-limbed, bromeliad-dwelling gal- liwasp distinguished from all other Jamaican species by a com- bination of: 1) low number (101-103) of ventral scales between mental and vent, 2) low number (41-43) of scales around body at midbody, 3) angular subocular scale modally between supra- labials 6 and 7, 4) enlarged postmental scale contacting 5 infra- labial scales, 5) fourth toe lamellae 18-21, 6) ratio of head width to head length high (80.0), 7) auricular opening small, 8) dorsal trunk and dorsal caudal scales keeled and striate, 9) ventral scales smooth, 10) dorsal pattern of tans and browns arranged in a distinct chevronate pattern, and 11) with dark markings on the head shields. Distribution. Known only from the type locality. Description of holotype. An (apparently) adult female with a snout- vent length of 105 mm and tail (almost entirely regener- ated) 75 mm; ventral scales between mental and vent 103, 41 scales around body at midbody; fourth toe lamellae 21, angular subocular between supralabials 6 and 7 on one side, between 7 and 8 on the other; head length 18.5, head width 14.8; ratio of head width to head length 80.0; median enlarged postmental (= first unpaired chin shield) small and contacting 5 infralabials. In life, dorsal pattern consisting of a series of about 16 or 17 wood brown chevrons, their apices pointing posteriorly, from the neck to the sacrum, on a tan ground; sides with somewhat lighter brown continuations of these chevrons both on the neck and between the limbs, the lateral continuations forming a series of more or less vertical brown bars which extend ventrad to about the level of the limb insertions; a few scattered paler tan dots or 4 BREVIORA No. 371 flecks in two vague lateral horizontal rows, associated with the lateral brown vertical bars; head tan, with more or less symmet- rical wood brown markings (a pair on the snout, an unpaired median blotch in the preorbital region, a median unpaired blotch on the posterior portion of the frontal, and the interparietal- parietal region with the scales dark edged); a black preorbital line on the lores; temples longitudinally streaked with very dark wood brown; a series of three brown lines on the supralabials, one below the eye, the two others extending vertically across the supralabials in the loreal region, all continuous ventrally across the infralabials; a series of three very dark brown to black nuchal blotches, the posteriormost the largest and located above the insertion of the forelimb; Umbs mottled brown and black dorsally, the forelimbs additionally with some intermixed tan areas and consequently appearing more mottled or marbled than the hindlimbs; underside pale orange, with discrete brown longitudinal streaks (four scales in length) or flecks on throat, and deep orange streaks on venter; underside of limbs and tail (unregenerated portion) pale orange; iris brown with orange pupillary ring. Variation. The paratype is a juvenile lizard with a snout- vent length of 66 mm. Scale counts are: 101 scales between mental and vent, 43 scales at midbody, angular subocular between supra- labials 6 and 7 on one side, between 8 and 9 on the other, fourth toe lameUae 18; head length 12.7, head width 9.3 (ratio 73.2). In hfe, the paratype was tan (PI. 14G6) dorsally with about 18 dorsal chevrons between the neck and the sacrum; the sides were paler tan (PL 13D3). The snout was olive, with the jowls and the base of the tail slightly reddish. The Hmbs were tan, spotted with dark brown to black. The venter was translucent gray, marbled with brown on the throat, and marked with reddish on the trunk and underside of the hindlimbs and tail. The facial markings, described for the holotype, were equally as prominent in the juvenile paratype. The dorsal chevronate pattern was dark brown to black, and on the sides the chevrons were continuous with weakly defined lateral vertical brown bars, each of which was followed by a creamy bar. Three nuchal-supra-axillary blotches were black, and the frontal head shield had a dark ante- rior margin with some additional dark sulTusions on the dorsal surface of the head. The postmental scale in the paratype is very small, but it contacts 7 infralabials; the contact on the right side 1971 BROMELIAD GALLIWASP 5 between the postmental and the third infralabiai is slight, but the contact on the left side is slightly more broad. Comparisons. D. fowleri needs comparison only with the three long-limbed Jamaican species (occiduus, hewardi, duquesneyi) . The new species differs from the short-limbed cnisculus and barboiiri in having much larger Umbs and from microblepharis in having the frontal longer than broad (in microblepharis, the frontal is broader than long). Counts of ventral scales (101-103) in fowleri overlap (of the other species) only the counts of crus- culus (97-122, data from Grant, 1940b); all other Jamaican species combined have ventral counts ranging from 107 to 150, with the low count of 107 in the giant occiduus, the high count of 150 in barbouri. In midbody scales, fowleri (41-43) overlaps only cruscuhis (36-49) and microblepharis (43). Combined midbody counts for all other Jamaican species (with the excep- tion of cruscuhis and microblepharis) vary between 47 {bar- bouri) and 59 {hewardi). In having the angular subocular between supralabials 6 and 7, fowleri resembles barbouri and cruscuhis but differs from hewardi and duquesneyi (7 and 8), microblepharis (5 and 6), and occiduus (8 and 9). In fowleri, the dorsal scales are striate and keeled; this condition occurs in all other Jamaican galliwasps with the exception of occiduus (dorsals striate but not keeled). In having smooth ventrals, fowleri resembles barbouri, hewardi, occiduus, and microblepharis. The ventral scales are striate in crusciilus and duquesneyi. Note, however, that hewardi may have weakly striate ventrals, and duquesneyi may have smooth ventrals. Finally, the striate and keeled dorsal caudal scales of fowleri are Hke those of cruscuhis and duquesneyi; all other Jamaican species have smooth {hewardi, occiduus) or keeled {microblepharis) superior caudals. From the above summary, it is obvious that fowleri combines features of scuteUation of several Jamaican species in new and different ways, and that the new species differs in combination of these characteristics from all other Jamaican species. Presumably, as will be noted below, D. fowleri is a local deriva- tive of the widespread D. hewardi. The fact that the juvenile fowleri was, in the field, considered as hewardi suggests the simi- larities between the two species. However, in addition to the structural differences noted in the above paragraph, the two spe- cies differ strikingly in coloration in life and perhaps less so in 6 BREVIORA No. 371 dorsal pattern. More importantly, the size of the auricular open- ing in fowleri is much the smaller; comparison of the opening in the fowleri holotype and a similarly sized hewardi (ASFS 14892; female with snout- vent length of 109 mm) reveals, even upon casual inspection, that the auricular opening of fowleri is slightly more than half the size of that of the hewardi. In addition, the two species differ in that fowleri has larger ventral scales (101- 103 between mental and vent in fowleri, 113-135 in hewardi), and fewer scales at midbody (41-43 versus 49-59 in hewardi). The enlarged postmental contacts 7 infralabials in all hewardi ex- amined, whereas at least in the fowleri holotype this scale con- tacts only 5 infralabials (weak contact with 7 infralabials in the paratype). In fowleri, the angular subocular modally hes between supralabials 7 and 8, whereas in hewardi it lies between suprala- bials 6 and 7. D. fowleri exceeds D. hewardi in number of fourth toe lamellae (15-19 in 22 hewardi, 18 and 21 in two fowleri). Finally, the head width/head length ratio in adult female hewardi varies between 70.7 and 74.1, whereas in the fowleri holotype, this ratio is 80.0. There are no comparably sized juvenile hewardi for comparison of this ratio in the paratype of fowleri, but the ratio (73.2) in this specimen lies near the upper extreme of ratios in hewardi with shorter snout-vent lengths (45-52 mm; ratios 66.4 to 73.8). The dorsal coloration of hewardi has been repeatedly recorded as greenish brown to greenish tan, but some individuals have the dorsum very dark brown (almost black) to metallic tan. The head regularly is unmarked dorsally, and vertical subocular and loreal lines are absent. The dorsal pattern consists of a trans- verse series of confused bars or bar fragments; these pattern ele- ments are usually so broken that no meaningful count can be taken. There is a strong tendency for the hewardi dorsal pattern elements to consist of bars, rather than chevrons as in fowleri. Perhaps the most distinctive pattern feature of hewardi, in con- trast to fowleri, is that of the throat. In hewardi, the throat has a broad dark reticulum, the pattern extending as far posterior as the forelimb insertions. This pattern is expressed even in the smallest juveniles and becomes more intense with increasing size. No hewardi has the discrete brown throat lines and flecks of fowleri. In addition, the ground color of the throat in hewardi is often blue to purplish, not pale orange as in fowleri. The deep orange belly markings of fowleri are absent in hewardi; some 1971 BROMELIAD GALLIWASP 7 hewardi have belly markings that are not discrete as in fowled and are gray in life. Comparisons in detail with D. occiduus are hardly necessary. In addition to the scutellogical differences noted above, the huge size of occiduus (to 305 mm snout-vent length) and its presumed terrestrial habits, coupled with its general bulk, immediately dis- tinguish it from fowled. The only other long-limbed Jamaican galliwasp is D. duques- neyi. Scutellogical differences have already been noted between this species and fowled. I have examined the holotype (MCZ 45194) and one other specimen (MCZ 45181) of duquesneyi. The species was casually defined by Grant (1940a: 6) on the basis of one juvenile specimen, and Cousens (1956) gave addi- tional pigmental and pattern differences between two specimens of duquesneyi, and hewardi. I have examined the two extant duquesneyi and both are damaged about the body so that accurate scale counts are difficult. Ventral scales between the mental and vent are about 116 and 122, midbody scales are about 48 and 49, and fourth toe lamellae are 19 and 23. The angular subocular lies between supralabials 7 and 8 on both sides of both specimens. D. hewardi and D. duquesneyi are comparable in dorsal pattern: the transverse markings in both are distinctly straighter and more barlike than the chevronate pattern in fowleri. In addition, com- parison of equally sized hewardi and duquesneyi shows that du- quesneyi (like fowleri) has a larger auricular opening than hewardi. D. fowleri differs from D. duquesneyi in that the former has fewer ventrals between mental and vent (101-103 versus 116- 122), fewer midbody scales (41-43 versus 48-49), the angular subocular between supralabials 6 and 7 rather than between 7 and 8, and smooth rather than striate ventrals (although the holo- type of duquesneyi has smooth ventrals). Two pattern elements differentiate duquesneyi from fowleri: the former has the tail (at least in juveniles) banded alternately black and sky blue (Grant, 1940b: 106), a feature unknown in any other Antillean galliwasp, and duquesneyi has an immaculate throat and venter (color un- known), a feature that separates duquesneyi from both fowleri and hewardi. D. duquesneyi also lacks the prominent facial mark- ings of D. fowleri. Finally, the head width/head length ratio (X 100) in duquesneyi is much less than this ratio in both fowleri 8 BREVIORA No. 371 and hewardi. In an apparently subadult female duquesneyi with a snout-vent length of 96, the HW/HL ratio is 69.8, below that of similarly sized female hewardi (ratios 70.7-13.1) and much below that of the slightly larger female holotype of jowleri (80.0). The HW/HL ratio in the juvenile holotype of duquesneyi (snout- vent length about 65) is 66.4, whereas this ratio in the jowleri paratype (snout-vent length 66) is 73.2; the HW/HL ratio in the duquesneyi holotype falls at the lower extreme of this ratio in smaller hewardi (ratio 66.2 to 67.3 in hewardi juveniles with snout-vent lengths of 49 to 52). Interestingly, Richard Thomas noted that a D. hewardi from Darliston, Westmoreland Parish, in western Jamaica, (and far removed from the known range of D. duquesneyi, which has been taken only on Portland Point, Clarendon Parish, in south-central Jamaica) with a snout-vent length of 48 mm, had pale blue distal tail bands alternating with brown bands. There seems little doubt that hewardi, duquesneyi, and jowleri are closely related, and that duquesneyi and presumably jowleri are peripheral geographic satellite species derived from parent hewardi in special situations. Remarks. The two specimens of D. jowleri were collected under the following circumstances. Both specimens were taken from bromeliads along the edge of the steep trail from Windsor Great House to Windsor Cave and thence up the escarpment of the Cockpit Country. In the case of the holotype, a Jamaican had been hired to cut arboreal bromeliads and had climbed a tall tree of moderate girth (0.5 meters) just below the path in decidu- ous forest. He had cut all but the last one or two bromeliads when the galliwasp rapidly descended the trunk of the tree and paused in confusion on a leaf about six feet above the ground. There is no doubt that the animal had been disturbed from its diurnal retreat by the chopping of adjacent bromeliads and had decided to abandon its place of retirement. The small paratype was taken from the moist center of a bromehad that had been growing 2.5 meters above the ground. When the bromeliad was cut and thrown onto the narrow path, the lizard was found inside the whorls of leaves. In both cases, the adjacent area was well forested. The elevation along the path is about 500 feet (153 meters). The Jamaican Cockpit Country is a karst region in north- western Jamaica. Its extent is about 20 miles (32 kilometers) east-west and about 10 miles (16 kilometers) north-south; the region centers in Trelawny Parish but extends for short distances 1971 BROMELIAD GALLIWASP 9 into St. James Parish on the west and St. Elizabeth and Man- chester parishes to the south. No roads penetrate it, but a series of peripheral roads allows some ingress into the region. An excep- tion is a relatively newly constructed road north of Quick Step on the southern border of the Cockpit, where penetration of about five miles (8 kilometers) is possible into virtually virgin territory. In search of Sphaerodactylus and hylid and leptodactylid frogs, we cut both terrestrial and arboreal bromeliads in several regions associated with the Cockpit periphery and elsewhere: between Spring Vale and Mulgrave (St. James and St. Elizabeth parishes), between Stonehenge and Burnt Hill (Trelawny Parish); south of Moneague on Mt. Diablo and west of Lluidas Vale (St. Cath- erine Parish), in the Dolphin Head region between Askenish and Town Head (Hanover and Westmoreland parishes), between Plum Park and Garlands (St. James Parish), and between Raheen and north of Quick Step (Trelawny Parish). In no case did we secure D. fowleri, although a single D. cnisculus was secured from a terrestrial bromehad north of Cave in Westmoreland Parish. The possibiUty remains that D. fowleri is not an obligate inhab- itant of bromeliads, and that it is a terrestrial galliwasp that, in the pitted, pocked, and rock-strewn Cockpit Country, finds diur- nal sanctuary in terrestrial situations from which it would be a lucky collector indeed who would secure it. On the other hand, there is no evidence to controvert the apparent fact that D. fowled is indeed a bromeliad dweller and that it occurs in no other situ- ation. If such is the case, it must be either extremely uncommon, remarkably elusive, or ecologically or altitudinally restricted in some presently unknown fashion. Along these lines, see Under- wood's (1959: 1) comments on his inability to secure a second specimen of D. microblepharis. The fact that the area where the microblepJiaris was secured backs upon limestone hills suggests, as Underwood stated, that it may have wandered from its usual habitat into a situation where it was fortuitously secured with relative ease. Diploglossiis fowleri is not known to be sympatric with any other species of galhwasp. However, D. hewardi has been taken 1.5 miles NW of Windsor, and D. cruscidiis has been secured 3.0 miles N W of Windsor — both in terrestrial situations. In addi- tion, D. barbouri has been collected along the eastern margin of the Cockpit Country between Stonehenge and Burnt Hill. The 10 BREVIORA No. 371 lack of precisely sympatric records between fowleri and any of these three species is not surprising, since, as pointed out above, collecting galliwasps within the Cockpit Country itself is a difficult and well-nigh impossible task except in especially favorable local- ities. If fowleri is truly bromeliadophilous, then it may in places be syntopic with D. crusculus, but such syntopy remains to be encountered. As presently understood, then, D. fowleri is a bromehad- inhabiting galliwasp that is presumably limited to the Cockpit Country area and possibly to lower elevations in that region. It is rather surprising that the bromeliad niche has been so neglected by Antillean reptiles, in contrast to Antillean amphibians. Cer- tainly Jamaica has the highest share of bromeUadicoles, both amphibians and reptiles; in addition to D. fowleri, Hyla brunnea, H. wilderi, H. marianae, and Eleutherodactylus jamaicensis are obligate bromeliad dwellers, and several other frogs {E. grabhami, E. cundalli, E. pantoni) are encountered with regularity in terres- trial bromeliads. Among reptiles, Sphaerodactylus oxyrhinus appears to be confined to this situation, and a new species of Sphaerodactylus, to be described by Richard Thomas, likewise is thus limited in habitat. S. argiis, D. crusculus, and Tropidophis haetianus are encountered in bromefiads upon occasion. This list of both obligate and facultative bromeliadicoles far exceeds that from any other Antillean island. On the other hand, no one has systematically cut arboreal and terrestrial bromefiads elsewhere than on Jamaica. It seems likely that this is a niche that will well repay investigation on other Antillean islands. LITERATURE CITED CousENs, Peggy N. 1956. Notes on the Jamaican and Cayman Island lizards of the genus Celestiis. Breviora, No. 56: 1-6. Grant, Chapman. 1940a. Notes on the reptiles and amphibians of Jamaica, with diagnoses of new species and subspecies. Jamaica To-day. London and Aylesbury, Hazell, Watson, and Viney, Ltd. Chapter 15: 151-157. 1940b. II. The Reptiles. In Lynn, W. G., and C. Grant, The herpetology of Jamaica. Bull. Inst. Jamaica, Sci. Ser., 1: 1-148. Maerz, a., and M. Rea Paul. 1950. A Dictionary of Color. New York, McGraw-Hill Book Co., pp. i-vii, 1-23, 137-208, 56 pis. Schwartz, Albert. 1970. A new species of large Diploglossus (Sauria: Anguidae) from Hispaniola. Proc. Biol. Soc. Washington, 82(60): 777-788. Underwood, Garth. 1959. A new Jamaican galliwasp (Sauria, Anguidae). Breviora, No. 102: 1-13. BREVIORA Mmsemim of Contiparative Zoology Cambridge, Mass. 31 March, 1971 Number 372 THE PALEONTOLOGY AND EVOLUTION OF CERION II: AGE AND FAUNA OF INDIAN SHELL MIDDENS ON CURACAO AND ARUBA Stephen Jay Gould Abstract. Cerion iiva has been found in great abundance in three Meso- Indian (preceramic) shell middens on Curasao. Shells from all three sites yield radiocarbon ages of about 4000 years B.P. Different groups of Meso- Indians from Venezuela reached Curasao and the nearby island of Cubagua at about the same time. A Neo-Indian (ceramic) midden on Aruba is ap- proximately 1500 radiocarbon years old. Lists of the molluscan fauna from all sites contain only intertidal and shallow water species. Collecting areas can be specified by noting differences among sites in the presence of species from various environments (rocky intertidal, mangrove, shallow grassy and shallow rocky). In the shell middens, Cerion presents two outstanding features: 1) almost all shells have had the apical whorls removed artificially and, 2) shells are larger than any living today. The apical whorls were removed by striking; flint tools found at the sites accomplish this task easily. This was done to release the internal vacuum and allow the animal to be sucked out through the normal aperture. Larger shells might indicate, since modern Cerion is so phenotypically variable, that the climate of Curasao 4000 years ago was more moist (and therefore more hospitable) than today. But there is no independent evidence for more rainfall at that time. If the effect is mainly genetic, these shells might come from relict populations, adapted to the pluvials of the previous glaciation. Cerion iiva has been found in a shell midden in Venezuela; this establishes the reciprocity of trade between main- land and offshore islands. INTRODUCTION Only a few mollusks have won entry into the Papiamento lan- guage of the Dutch Leeward Islands. These are mostly edible species — kiwci (Cittarium pica), karko (Strombus gigas), and tcipa koncha ("cover shell" — a general name for chitons). Yet 2 BREVIORA No. 372 Cerion uva, the ubiquitous pulmonale of these islands, stands out for the plethora of names attached to it, names that distinguish small from large and beach from bush. Nevertheless, Cerion plays almost no role in the economy of these islands today — though one of its names, kokoUshi kalakuna (turkey shell), reflects the fact that it is sometimes fed to turkeys as a source of lime. It is never eaten, save as an aphrodisiac by some older residents who believe that sea shells preserve sexual potency (and do not realize that this halophilic pulmonale, which lives just landward of Tectariits muricatus, does not come from the sea). But to another people, the original Indian inhabitants of Curasao, Cerion uva was a major source of food, for the oldest middens of the island are crammed with their shells. Of the many sheU sites that have been studied (Van Heekeren, 1960: 103-109, for review of archaeological work and Van Heek- eren, 1963), Cerion is known only from the older, preceramic middens of Curasao. Whenever it occurs, it presents two peculi- arities: sheUs are far larger than the largest living C. uva, and most all have had the apical whorls removed artificially. Thanks to the kindness of Father Paul Brenneker and Mr. Elis JuUana, local collectors, folklorists, and historians (and my informants for the opening paragraph), and Dr. F. Creutzberg, Director of the Biological Station at Piscadera Baai, Curasao, I had the opportunity to study the sheU sites during the summer of 1968. In this paper, I shall review the archaeological setting of these islands, report on radiocarbon dating of the shell sites, tabu- late the fauna of each and present environmental interpretations, and discuss the occurrences of Cerion with special reference to the peculiarities mentioned above. CARIBBEAN PREHISTORY AND DESCRIPTION OF SITES The Dutch Leeward Islands are tied, geographically, to Vene- zuela. Aruba, only 27 km from the mainland, lies on the coastal shelf, in easily navigable waters. Curacao and Bonaire are more distant (64 and 87 km respectively), and the passage is deeper (up to 1500 m) and more treacherous (Van Heekeren, 1960: 103). The early colonization of these islands must be discussed in the context of Venezuelan archaeology (Cruxent and Rouse, 1958-59, 1969; Rouse and Cruxent, 1963; Rouse, 1960, 1964, 1966). 1971 CERION FROM INDIAN SHELL-HEAPS 3 The Pre-Columbian inhabitants of Venezuela and the Carib- bean are designated Paleo-, Meso-, or Neo-Indians on the basis of technology and inferred economy. Although the three stages do express a chronological progression, none of their artifacts function as "index fossils" in establishing contemporaneity throughout the Caribbean, for the traits of a new stage are attained at different times by different peoples. There were, for example, still some preceramic Meso-Indians on Haiti and Western Cuba when Columbus arrived (Rouse, 1966). The original inhabitants of the New World were Paleo-Indians, "hunters of mammoths and other large land mammals" (Rouse, 1966: 125). Their stone tools have been found in Venezuela and designated as markers of the Joboid Series. They date, approxi- mately, from 17,000-7,000 B.P. The oldest radiocarbon date for Joboid charcoal is 16,870 years B.P. (Rouse and Cruxent, 1963). In earlier works, Cruxent and Rouse held that Paleo- Indians were not sea-farers, but Paleo-Indian sites have recently been found at Mordan in the Dominican Republic and dated to at least 4560 radiocarbon years B.P. They beheve, moreover, that the Mordan site is predated by another at Casimira that may be as much as 7,000 years old (Cruxent and Rouse, 1969). Although the mainland source of these first Hispaniolans is not known, these finds indicate that some Paleo-Indians crossed con- siderable stretches of ocean, probably on rafts and by accident (Cruxent and Rouse, 1969). Much scholarly agitation of late has been directed to the issue of whether or not Paleo-Indians were responsible for the extermi- nation of large land mammals (Martin and Wright, 1967). In any event, their demise drew our pre-agricultural people to the sea and inaugurated MesoTndian culture, characterized by "rela- tively few stone tools. Projectile points are made of bone rather than stone and shell artifacts are common, reflecting the mari- time orientation" (Rouse, 1966: 126). Meso-Indian artifacts in Caribbean Venezuela belong to the Manicuaroid Series and date, approximately, from 7,000 to 3,000 years B.P. The oldest radio- carbon date for mainland Venezuelan Meso-Indians is 5750 B.P. (Rouse and Cruxent, 1963). There is an extensive Meso-Indian site on Cubagua, another of Venezuela's offshore islands. Char- coal from the base of this deposit dates at 4275 radiocarbon years B.P. 4 BREVIORA No. 372 The subsequent Neo-Indian culture is "marked by pottery making and fully developed agriculture" (Rouse, 1966: 126). The invention of pottery was the crucial archaeological event that inaugurated the Neo-Indian period; therefore MesoTndian and earlier sites are often designated simply as "preceramic." Agri- culture, with manioc as a staple crop, and pottery were developed in the Orinoco Valley during the 2nd millennium B.C. During the 1st millennium B.C., some Neo-Indians moved out to the coast and became sea-farers. Displacing Meso-Indians as they went, they migrated to the coastal islands, up the Lesser Antilles and reached the Greater Antilles ca. 250 A.D. and the Bahamas ca. 1000 A.D. This displacement was still occurring when Colum- bus reached the New World (Cruxent and Rouse, 1969). The Cerion sites of Curasao are all Meso-Indian in nature. I studied the following three sites: 1. Rooi Rincon — North coast, west of Hato Airfield; in soil at the base of a small cave in a raised Pleistocene reef that also houses the larger cavern of Hato and several others; approximately 40 m above present sea level and 1 km from the coast. This well- known site was excavated by Cruxent in 1965 (Tamers, 1967) and by Van Heekeren in 1960 (Van Heekeren, 1963). Crudely chipped stone tools and flint flakes are common but, after dig- ging for 14 days. Van Heekeren found only one other artifact, a shell disc bead (Van Heekeren, 1963: 5). The naturally broken columellar tips of Strombus gigas are similar in form to some of the fashioned shell gouges common in the Manicuaroid deposits of Cubagua (Cruxent and Rouse, 1958-59); they may have been used for digging meat out of sheUs. Many other natural objects could have been used as tools. Particularly suspect are the smoothly eroded and fairly pointed branches of the stag horn coral, Acropora cervicornis, that are fairly common at this site and at Kintjan (site 2). These, obviously, have no nutritional value and must have been carried to the site for some other pur- pose. Other objects, land crab claws for example, might have been used for digging meat from shells after their own contents had been consumed. I found a few bits of charcoal: some of the shells are strongly scorched. Cruxent says of this deposit: "A Meso-Indian complex of collectors with industry of stone chips. Classified as a marginal development of El Jobo. No archaeologic station of this type presently known in Venezuela" {in Tamers. 1967: 244). 971 CERION FROM INDIAN SHELL-HEAPS 2. Kintjiin — Near south coast, east ^ Willemstad. The area, a hillslope, is being cleared for construction and shells are loose at the surface; their presence in a small area indicates original concentration in a coherent deposit. Flint chips and crude stone tools are, as at Rooi Rincon, common at this site. 3. Tafelberg — Near south coast, just east of the Tafelberg Santa Barbara. Only a few shells could be collected from the recently blasted rubble of these phosphate workings. Mr. Harry Evers, engineer at the Tafelberg phosphate workings, informs me that, prior to the blasting, the shell heap was a coherent deposit with two layers, marine shells at the base and decapitated Cerion at the top. I found no artifacts at this much disturbed site. Dr. P. Wagenaar Hummelinck, pre-eminent natural historian of these islands, has told me (personal communication, 1970) of one additional Cerion locality at Hato Cave; I have not seen this site. He also states that he knows of no other Cerion site on any of the three islands. For comparison, I add to the Cerion sites of Curacao one later, Neo-Indian deposit from Aruba: t:.t V- ^pt>^^ Figure 1. Artifacts from Ceru Canashito, Aruba. la) left: rock drawing, presumably depicting a pregnant woman, lb) right: shell disc made from Melongena melongena. Actual height: 43 mm. 6 BREVIORA No. 372 4. Ceru Canashito — North slope of this Hmestone terrace. I chose this among the many Neo-Indian sites of Aruba for two of its outstanding features. Good skeletal material has been col- lected from the caves near its summit (Tacoma, 1959), and these caves contain some of the best of the celebrated and mysterious rock paintings of these islands (Hummelinck, 1953, 1957). One of these, probably depicting a pregnant woman, is reproduced as Figure la. (There is, of course, no reason to assume that the rock drawings are contemporaneous with the shells; Van Heek- eren (1960), in fact, suspects that they were fashioned by Meso- Indians and venerated by later inhabitants.) Shells occur at all levels of the slope, but are concentrated by gravity at the base in an inhomogeneous deposit. Sherds of a coarse, unornamented, grit-tempered pottery are common. Shell artifacts include the columellar points of Strombus gigas and the unperforated shell disc, made from the outer whorl of Melongena melongena, shown in Figure lb. Such unperforated shell discs are common on the islands; their function is unknown (Van Heekeren, 1960: 112). AGE OF THE SHELL MIDDENS Tamers (1967) reported the first radiocarbon dates from archaeological sites in the Dutch Leeward Islands; all samples were charcoal and all were supplied by Cruxent. Included are five dates for the Rooi Rincon shell midden, two from a pit previ- ously excavated by Van Heekeren and three from two new pits. The dates range from 3900 ± 50 to 4490 ± 60 with a mean of 4194 radiocarbon years (see Stuiver and Suess, 1966 on the relationship between radiocarbon and calendar years). These are the only dates previously calculated for preceramic sites on these islands. Radiocarbon ages were determined for 1 1 shell samples by Geochron Laboratories, Inc., Cambridge, Massachusetts (Chama macerophylla and Cittarium pica from each of the five sites and Anadara notob'iUs from Ceru Canashito) . "The shells were cleaned of foreign material and were thoroughly leached with dilute HCl in an ultrasonic cleaner to remove the surficial layer of carbonate and expose fresh material. The cleaned shells were then hydro- lyzed to recover CO., for the analysis" (personal communication from H. W. Krueger of Geochron). Dates are based on a half- life of 5570 years and referenced to 1950 A.D. 1971 CERION FROM INDIAN SHELL-HEAPS 7 Dates based on shells are not as reliable as those determined for pure carbon (charcoal), for CaCO,, is often altered by per- colating, acidic groundwaters. I was anxious to determine the correspondence between shell and charcoal dates for Rooi Rincon; I found no charcoal at any of the other sites. All dates are shown in Table 1. The correspondence at Rooi Rincon is satisfactory, and all pre- ceramic sites of Curacao are about 4000 radiocarbon years old. This date is particularly interesting since it corresponds so well with the base of the great Meso-Indian site at Punta Gorda, Cubagua Island (p. 21). The artifacts of this Cubagua complex of the Manicuaroid series differ greatly from those of Rooi Rin- con (Cruxent and Rouse, 1958-59) and we must assume that different groups of Meso-Indians from Venezuela colonized the coastal islands at about the same time. The great spread of dates for the Neo-Indian site of Ceru Cana- shito can be explained in two ways. It is a very inhomogeneous deposit of shells artificially concentrated at the base of a slope and may represent a long span of habitation. Alternately, the Cit- tarium date could be spuriously young. Cittarium has been and remains a staple food of the islands. The kiwa is sold at all native market places; shells are carried and discarded all over the island. If this date has been falsified by the inclusion of a fairly modem shell, then the Canashito midden may represent a more coherent deposit, about 1500 radiocarbon years old. FAUNA OF THE SHELL MIDDENS In presenting these faunal lists, I have excluded the micro- molluscs that could have played no role in the economy of the Indians (though Tnmcatella and other rissoids are reasonably common as accidental transports). In each site, there are a few species that clearly dominate; these are merely listed as common. Numbers of specimens are given for other species. I have used Warmke and Abbott (1961) and Coomans (1958) as guides to identification; order of listing and family allocations follow the former source. 1. Rooi Rincon AMPHINEURA A cant ho pleura graniilata — common 8 BREVIORA No. 372 GASTROPODA PROSOBRANCHIA Trochidae Cittarium pica — common TURBINIDAE Astraea tecta — 1 Astraea tuber — 1 Neritidae Merita peloronta — 11 Nerita versicolor — 6 Nerita tesselata — 4 LiTTORINIDAE N odilittorina tuberculata — 4 Echinus nodulosus — 1 Tectarius muricatus — 1 0 Vermetidae Petaloconchus mcgintyi — 3 Strombidae Strombus gigas — 4 apices and 3 columellas MURICIDAE Murex brevifrons — 8 Magilidae Coralliophila abbreviata — 2 Coralliophila caribbea — I Fasciolariidae Leucozonia nassa — 1 Xancidae Vasum capitellum — 1 GASTROPODA PULMONATA Cerionidae Cerion uva — common; 18 of 129 specimens have in- tact apices BIVALVIA Arcidae Area zebra — 12 valves Area imbricata — 4 Anadara notabilis — 4 Mytilidae Brachidontes exustus — 2 Pteriidae Pinctada radiata — 1 3 1971 cerion from indian shell-heaps 9 Pectinidae Pecten ziczac — 2 LiMIDAE Lima scabra — 10 OSTREIDAE Ostrea jrons — 1 2 Crassostrea rhizophorae — 9 . Chamidae Chama macerophylla — common Pseudochama radians — 2 Nonmolluscan remains: a few branches of stag-horn coral {Acropora cervicornis) , land crab claws (common), a few barna- cles, fish bones and a small fragment of an echinoderm test. Not all these animals were eaten. Many, especially among the snails, are small and rare at the site (turbinids, magilids, fasci- olariids, and xancids); others (Petaloconchus and barnacles) cement to other shells and surely won a free ride on their edible hosts (probably Chama). The main food sources were the land snail Cerion, land crabs, intertidal chitons, the intertidal and just subtidal snail Cittarium and the shallow water clam, Chama; all are very abundant and easily gathered. Less common but still important as food sources are the conch Strombus gigas, Nerita peloronta, and Murex brev- ifrons among the snails (the last two artificially broken in char- acteristic ways — Figs. 2 and 3 ) and arcids, oysters, and limids among the clams. The shells provide an excellent picture of the environment from which they were gathered. All the major intertidal rock-clingers are represented (all three common West Indian Nerita, chitons, and the famous homeomorphic series Nodilittorina-Echinus- Tectarius). These species inhabit rocky shores in areas of active surf. All other species can be found in less than 10 feet of water on a varied bottom containing reefy and rocky areas (Chama, Area, Lima) and stretches of sand and grass (Anadara, Strom- bus). There may have been a lagoon with mangroves nearby, for many important elements of the mangrove-root community are present (Murex brevijrons, Ostrea jrons, Crassostrea rhizo- phorae, and Brachidontes exustus). Van Heekeren (1963) stated, correctly no doubt, that the shells were collected on the nearby north coast (Fig. 4b). Since 10 BREVIORA No. 372 Figure 2. Miire.x brevifroiis shells from Kintjan (left) and Rooi Rincon (right). Note characteristic breakage pattern in both. This can be achieved by placing the shell face down upon its aperture and striking the apex. Actual height of Kintjan specimen: 54 mm. Figure 3. Neritids from Rooi Rincon broken in characteristic fashion. Left: apertural portion from rear; Right: apertural portion from front. Such a break is made by placing the shell face down upon its aperture and striking the body whorl with a blunt object. This is also the natural break- age pattern in most cases. Right-hand fragment is 18 mm high. 1971 CERION FROM INDIAN SHELL-HEAPS 11 the unremitting trade winds blow against this coast (producing a strong surf most unconducive to shell gathering), Van Heekeren suggested that sea level at this earlier time was 6-7 m higher than today. This would submerge the extensive raised reef that forms the lower terrace all around Curacao and produce a broad area of calmer, shallow water. (And from the supposed extent of this change in level, he postulated a great age for the deposit and classified it, tentatively, as Paleo-Indian.) This hypothesis of a major shift in sea level is unnecessary for two reasons: 1) With an age of 4000 radiocarbon years, any eustatic fall in level is ruled out; if anything, mean sea level then was a bit lower than today (Redfield, 1967; Milhman and Emery, 1968). This leaves tectonic uplift. Curacao has, indeed, been uplifted during the Pleistocene (the oldest terrace, atop the Tafelberg, lies at 140-200 m, but 7 m in 4000 years is not likely). 2) The trade winds do produce a strong surf along the north coast. But Rooi Rincon lies on that part of the coast that runs due east- west; here the winds run along the coast and the waters are fairly calm. Modern Cerion populations illustrate the climatic results of changes in coastal direction. Cerion lives atop the first terrace all along the coast. In areas continually buflfetted by the strong dry wind, they aestivate for much of their lives and remain small as adults; they grow bigger in calmer areas. A graph of Cerion size vs. distance from Westpunt (Fig. 4a) is a good map of coastal direction (Fig. 4b). Cerion is small where the coast runs north- south and large where it runs east-west. They reach their greatest size at Rooi Rincon. Thus, Rooi Rincon lies in the only area of Curasao that provides good conditions for shell gathering on the north coast. 2. Kintjan GASTROPODA PROSOBRANCHIA Trochidae Cittarium pica — common Strom BiDAE Strombus gigas — common Cymatiidae Charonia variegata — 1 MURICIDAE Murex brevijrons — 3 (broken as at Rooi Rincon, Fig. 2) 12 BREVIORA No. 372 Melongenidae Melongena melongena — 2 GASTROPODA PULMONATA Cerionidae Cerion uva — common, 7 of 347 specimens have in- tact apices BIVALVIA Arcidae Area imbricata — 13 Barbatia cancellaria — 3 Anadara notabilis - Pteriidae Pinctada radiata — - Pectinidae Pecten ziczac — 7 Limidae Lima scabra — 6 common Figure 4. Correlation of coastal direction and shell size. 4a) left: map of Cura^,ao. 1. Rooi Rincon at point where coast runs east-west. 2. Kintjan. 3. Tafeiberg. 4. Schottegat (where shells at Kintjan were collected). 1971 CERION FROM INDIAN SHELL-HEAPS 13 OSTREIDAE Ostrea frons — 7 Crassostrea rhizophorae — 6 Chamidae Chama macerophylla — common NonmoUuscan remains: branches of stag-horn coral (Acropora cervicornis), barnacles, and fish bones. The shallow water fauna of Kintjan is very similar to that of Rooi Rincon, both in species composition and order of dominance {Chama and Cittarium followed by S trombus, arcids, oysters, and limids). Since shells are not so common at Kintjan, several spe- cies, rare and unimportant at Rooi Rincon, are not found here. 10 20 30 Distance from Westpunt (miles) 4b) right: Mean shell heights (20 adults per sample) for local populations living in similar microhabitats directly on the first terrace along the east coast of Curasao. Shells are largest where trade winds do not hit coast directly. 14 BREVIORA No. 372 I found no land crabs at Kintjan, but Cerion uva is even more common here than at Rooi Rincon. There is, however, one out- standing difference between the two sites: there are no intertidal rock-dwellers at Kintjan (neritids, littorinids, or chitons), while all the common forms are found at Rooi Rincon. This difference permits us to specify the collecting area for Kintjan shells. The entire periphery of Curasao is framed by an uplifted Pleis- tocene reef; intertidal forms are common all around the coast. But the central areas are underlain by volcanic rocks that erode more easily than the coastal limestone. During the last glacial period, when sea levels were lower, extensive drainage systems were developed on the volcanic terrain; these breached the harder limestone rim in only a few places. These valley systems were drowned when sea level rose and produced the outstanding pro- tected harbors that characterize all three islands: narrow inlets with expansive inland waters. Willemstad, the capital of Curacao, is built on both sides of the largest harbor, the Schottegat. The inland shores of the Schottegat are volcanic; in the absence of strong surf and a rocky coast, the rock-dwelling intertidal forms do not inhabit these shores. I conclude that the Kintjan shells were collected in the Schottegat (Fig. 4b); the extensive, calm, shallow waters provided an excellent site for gathering. 3. Tafelberg GASTROPODA PROSOBRANCHIA Trochidae Cittarinm pica — several fragments LiTTORINIDAE Tectarius muricatus GASTROPODA PULMONATA Cerionidae Cerion uva — common, 5 of 1 1 1 have intact apices BIVALVIA Arcidae Area imbricata — 1 Chamidae Chama maeerophylla — common The site has been thoroughly disturbed by blasting. 1971 CERION FROM INDIAN SHELL-HEAPS 15 4. Ceru Canashito AMPHINEURA Acanthopleiira granulata — 4 plates GASTROPODA Trochidae Cittarium pica — 8 Turbinidae Astraea tecta — 1 Neritidae Nerita tessellata — 3 Littorinidae Tectarius miiricatiis — 5 MODULIDAE Modulus modulus — 1 Cerithiidae Cerithium algicola — 1 Cerithium Utteratum — 1 Strombidae Strombus gigas — common Muricidae Murex pomum — 7 Murex brevijrons — 1 Thais deltoidea — 1 Melongenidae Melongena melongena — common Xancidae Vasum muricatum — common BIVALVIA Arcidae A nadara notabilis — common Pteriidae Pinctada radiata — 1 LUCINIDAE Codakia orbicularis — common Chamidae Cliama macerophylla — common Pseudochama radians — 1 16 BREVIORA No. 372 Intertidal rock-dwellers are found here, but the series is not nearly so complete as at Rooi Rincon (only one Nerita, Tectarius, but neither Echinus nor Nodilittorina). Among shallow water forms, there are two major differences between Canashito and both Rooi Rincon and Kintjan. The Curasao sites contained a suite of mangrove-dwellers that are completely absent here (Cana- shito yielded one Murex brevijrons, a common mangrove form, but Murex pomum, an open water species absent from both Cura- sao sites, is the common Murex here). In addition, Canashito contains a suite of shells {Modulus, the two Cerithium species and, especially, the common Codakia orbicularis) that inhabit grass and algal beds; none of these occur in the Curasao deposits. The shells were probably collected in calm waters off the leeward south coast, near the site of the present airport. CERION UVA IN THE PRECERAMIC MIDDENS OF CURACAO In all three preceramic middens of Curagao, the most common moUuscan shell is that of the land snail Cerion uva. These shells present two outstanding features: more than 80 percent in each locality have lost their apical whorls and shells are larger and more variable than modern specimens. 1. Removal of the apical whorls. By reason and experiment, one of a list of possible proposals can be identified as the cause of removal. I list the suggestions made to me by many friends and colleagues. A) Natural removal B) Artificial removal i) by biting ii) by rubbing iii) by crushing (striking with the shell held up- right) iv) by slicing (striking with the shell placed on its side). Although the apical whorls form the weakest part of the shell, I do not believe that they could have been lost naturally by so many specimens. I have extensive collections of much older fos- sils from fissure-fills on Aruba. These tumbled, often down sev- eral meters, into the fissures, suffered strong compaction, under- went tectonic uplift and still retain, in almost all cases, the apical 971 CERION FROM INDIAN SHELL-HEAPS 17 whorls. I have never seen a natural accumulation, cither recent or fossil, in which many specimens are missing their apical whorls. After suffering one dental misfortune, I am quite sure that the tops cannot be bitten off. Apices can be removed by rubbing either against limestone or volcanic rock, but the process is much too laborious and time-consuming. I am convinced that the tops were removed by striking. They were not crushed by striking the top of the shell while holding the bottom against a substrate (and keeping the shell vertical), for this process invari- ably breaks the lower lip of the aperture before crushing the top. If, however, the shell is placed on its side, horizontally against the substrate, the top can easily be removed by striking with a sharp instrument. In fact, the flint chips and stone tools of Rooi Rincon and Kintjan, are excellent devices for this purpose. With a bit of practice, the apices can be removed with a single blow. This leaves open the question of why the apices were removed. 1 can imagine three interpretations: A) Removal is unrelated to eating; the shells were used for an ornamental or other purpose. B) When the top is removed, the animal can be sucked out through the apical hole thus produced. C) Removal of the top aids, somehow, in sucking the animal out through its normal aperture. I cannot imagine what nongastronomical purpose so many thousand decapitated shells could have served. Moreover, the following demonstration that decapitation is an aid to removal of the animal argues strongly against A. Figure 5. X-ray photographs of decapitated Ccrion uva from Kintjan (left 2 specimens) and Rooi Rincon (right 2). Since internal whorl parti- tions are intact, animal was not removed through apical hole. Specimen on left is 32.8 mm high. 18 BREVIORA No. 372 If the animal were sucked out through the top, some of the internal whorl partitions would have to be broken, for the large foot could not fit in the small whorls left near the top of the shell. X-ray photographs of decapitated shells (Fig. 5) show clearly that the whorl partitions are never disturbed. The animal could not have been extracted through the apical hole. If you take an intact shell with its animal inside and suck as hard as possible at the aperture, the animal cannot be extracted. But, when the apex is removed, a single hard suck upon the aper- ture will extract either the large foot of the animal or the entire body itself. Removal of the top breaks the vacuum inside the shell and facilitates the extraction of its contents. The entire process is really quite efficient: one strike, one suck, and the animal is removed. Several can be eaten in a minute (though I recommend Cerion only to the starving). Somehow, I find it satisfying to think that the Meso-Indians of Curacao discovered an important physical principle for such a practical procedure. This idea, so obvious to all of us who were raised in the pre pop-top age of the beer can industry, is by no means a self-evident principle. 2. Variation and jorm of Cerion uva. Any sample from a shell midden is, of course, strongly biased from a biometrical point of view. The probable bias, in these cases, is twofold: the selection of large individuals (for Cerion is not a large snail and much work must be expended for little nutrition), and the amalgama- tion of shells from several local populations. Much has been made in the literature of the extreme intraspe- cific variability of land snail shells. This indeed is true, but it is usually of a particular kind (and this is rarely emphasized). The variation is interpopulational, i.e., the shells of any local popu- lation are not unusually variable, but differences among the means of local populations are often extreme. Thus, it is likely that our two biases will afi'ect the mean of a midden sample in opposite ways: the selection of large shells will augment the mean, but the amalgamation of large individuals from several local popula- tions will produce a midden mean smaller than the true mean of a local population with large shells. The rise in variability from amalgamation of local populations can be gauged by comparing coefficients of variation (C.V.) (Simpson, Roe, and Lewontin, 1960: 89-95) of midden samples 1971 CERION FROM INDIAN SHELL-HEAPS 19 and modern local populations for the same character. Table 2 presents C.V/s for shell height of the three midden samples and a mean value for 69 modern local populations (Gould, unpub- lished data for monograph in preparation; N = 20 for all samples, midden and modern; values for midden shells are estimates for actual height with decapitated apical whorls restored; all shells are adults with completed growth ) . All midden means are above the modern grand mean. Rooi Rincon and Tafelberg are within the span of modern C.V.'s (4.03 to 10.18), but, at 15.45, shells from Kintjan are far more variable than those of any modern local population. The striking feature of midden samples is the large size of some of their shells. Fortunately, Cerion iiva is among the world's best known land snails from a biometrical point of view. Three major studies have been done in this century: by Baker in the earlv 1920's (Baker, 1924), by Hummelinck in the late 1930's (Hum'- mehnck, 1940) and by myself during the past two years. Table 2 compares the heights of shells in midden and modern samples. Each modern study has uncovered a local population with greater mean height than the smallest midden sample, and one of Hum- melinck's local populations exceeds the largest midden sample in mean height. Still, of course, the midden means are all well above T^'*' Figure 6. Comparison of largest shell heap (left, from Kintjan, 34.3 mm high) Cerion and largest modern shell. Difference is much more striking in actual shells in which areal artifact of two dimensional representation is lost and judgment of size is made more properly by volume. 20 BREVIORA No. 372 the grand mean of means for each modern study. However, as mentioned previously, the midden means are almost surely lower than the true means of local populations with large shells living at that time. A more appropriate comparison might be made using maximal size. Among almost 12,000 modern snails from 248 local popula- tions over 50 years, no snail greater than 30 mm in height has ever been found. (In only one of Hummelinck's local populations did any individuals exceed 29 mm; neither Baker nor Gould found any taller than 28.5 mm.) Yet snails exceeding 30 mm in height are very common in two of the three midden samples and, at 34.3 mm, the largest snail from Kintjan dwarfs my modern "giant" (Fig. 6). Two separate factors can make a snail tall, and both operated to produce the large midden shells. First, a snail can increase in height simply by adding more whorls. Each of the decapitated shells of Figure 5 shows 1 1 whorls below the break; the complete shell would have had one or two more postprotoconch whorls. Modern shells with more than IOV2 postprotoconch whorls are a great rarity (Baker and Hummelinck included protoconch whorls in their count, hence their larger figures). Secondly, a tall snail may have as many whorls as a smaller one, but simply have larger whorls. Protoconch size is a good measure of general whorl size (Gould, 1969). Only Rooi Rincon has enough complete shells to permit the calculation of mean protoconch width. At 1.67 mm, mean protoconch width for Rooi Rincon is at the top of the range of modem mean widths (1.41-1.69 mm for 69 samples, N ^ 20 for each sample ) . The midden shells grew more whorls than any modern sample and had larger whorls than most. Why were the midden snails larger than modern snails? All three modern studies have demonstrated the extreme phenotypic plasticity of Cerion iiva. Shell size of adults is a direct function of microenvironment; snails are large when habitats are moist, calm, and well vegetated. Curasao today is an arid island. It receives only 17-22 inches of rain per year, most in brief downpours. It is hard to imagine a less hospitable area in the West Indies for pre-agricultural Meso-lndians. I do not know what they could have found, in this cactus-covered land, to supplement a diet of sea food. It is therefore tempting to think that the large midden shells indicate a wetter climate that might have supplied to Meso- lndians some of the tropical fruits that adorn most West Indian 1971 CERION FROM INDIAN SHELL-HEAPS 21 islands. Unfortunately, there is no other evidence for greater rain- fall 4000 years ago. If Curasao were much larger or higher than it is today, continental effects might lead to increased rainfall. But the eustatic rise of sea level has not been more than 10 feet during the past 4000 years (Redfield, 1967; Milliman and Emery, 1968) and the direction of tectonic movement has been upward (Weyl, 1966). Rouse and Cruxent (1936: 38) believe that tem- peratures and rainfall have not varied appreciably during the past 5000 years in Venezuela and surrounding areas. If large size is not an immediate phenotypic response to local conditions more favorable than today's, then I suspect that the midden snails were programmed to be large, i.e., that the effect is mainly genetic. In this case they probably represent the relict populations of snails that had been genetically adapted to more favorable conditions during pluvial cycles of the previous glacial period. In any event, they served the Meso-Indians well; it would be hard to make a meal of modern Cerion. There is an interesting postscript to the relationship of Cerion with Meso-Indians. There is considerable evidence for trade between the mainland and coastal islands, but it is all unidirec- tional. Rouse and Cruxent (1963: 45) found trade pottery from Venezuela in the Punta Gorda complex of the Manicuaroid Series on Cubagua. Du Ry (1960: 85) discovered that the oldest pot- tery of Aruba is finer in texture than later examples. He assumes that this first pottery was imported from northeastern Venezuela and that the later work is indigenous. In a nearly-forgotten work, Berry (1934) found Cerion iiva in an Indian shell heap near Lake Valencia, Venezuela. Berry was not convinced that these shells were imported from the Dutch Leeward Islands. But his argu- ment that Cerion might have once inhabited the shores of Lake Valencia can be discounted because this halophile would not sur- vive so far inland. I also doubt that Cerion inhabited the coast of Venezuela, for it has never been recorded from shell heaps there. Since there is no evidence that Cerion uva ever lived elsewhere than the Dutch West Indies, I conclude that the Valencia speci- mens establish the reciprocity of transport between Venezuela and the islands. 22 BREVIORA No. 372 ACKNOWLEDGEMENTS I thank Father Paul Brenneker and Mr. Ehs Juhana, local folk- lorists and archaeologists, who collected with me at Rooi Rincon; Dr. F. Creutzberg, Director of CARMABI, who showed me the Kintjan site, and Mr. Harry Evers, who allowed me to collect at the Tafelberg midden. This work was supported, in part, by N.S.F. Grant No. GB-12553. 1971 CERION FROM INDIAN SHELL-HEAPS 23 X) CD < O o "2 03 St V C « u "=c 3 u c C3 ^ ■5 c c !^ ^ o Q, c/l ^ U B o o C3 OJ C^ O" O' JO C3 « 3 k. u. u 3 3 < u o c c rt C3 C c3 ■5 c ■5 '•3 c 6 6 1 o C/5 C/5 OJ OJ /-l >i-i r<~, vO — "/". »o 00 ro m ,_, o v^ "? ^ — r— ■ *— " •— " + 1 +1 +1 o c lO ITi (^ 00 — 00 VC ■* <^ c o c o o ao c i« CS c U ■*-t u 3 c ^+— u a * •« >/-> tT Ov * O •2? ^ (^ n\ (N a: >/^ 00 t~^ vd to a C a (U ^ -5 ^ "^ ^ Tf 00 t3 1— t ^ %< a: «0 r4 .s £? G a ^ > ■*-» *+^ s; C ^ 0\ o CTv o .Op O) n-1 >0 s >rl vO ■* 5 a: . O ^ s ■*— • 1^ <^ I^ ro r- O 00 T3 \- C 0 c« ^-1 -Si "5. -^ s C 1! P •0 _3 v5 5 CO c s > CO C £ O 2' 3 0. c '5 o C3 E 3 a\ 3 0 C 3 5 oi H m X 0 * 1971 CERION FROM INDIAN SHELL-HEAPS 25 LITERATURE CITED Baker, H. B. 1924. Land and freshwater molluscs of the Dutch Leeward Islands. Occas. Pap. Mus. Zool. Univ. Mich., No. 152. 158 pp. Berry, C. T. 1934. Pleistocene remains found near Lake Tacarigua, Vene- zuela. J. Washington Acad. Sci., 24: 387-395. CoOMANS, H. E. 1958. A survey of the littoral Gastropoda of the Nether- lands Antilles and other Caribbean Islands. Stud. Fauna Curasao, 8: 42-111. Cruxent, J. M., AND L Rouse. 1958-59. An Archaeological Chronology of Venezuela. Social Sci. Monogr. VI, vols. I, 277 pp. and 11, 223 pp. Pan American Union, Washington, D.C. 1969. Early man in the West Indies. Sci. Am. 421: 42-52. Du Ry, C. J. 1960. Notes on the pottery of Aruba, Curasao, and Bonaire. Stud. Archaeol. Netherlands Antilles, I: 81-102. Gould, S. J. 1969. An evolutionary microcosm: Pleistocene and Recent history of the land snail P. (Poecilozonites) in Bermuda. Bull. Mus. Comp. Zool., 138: 407-532. Heekeren, H. R., VAN. 1960. A survey of the non-ceramic artifacts of Aruba, Curacao, and Bonaire. Stud. Archaeol. Netherland Antilles, II: 103-120. 1963. Prehistorical research on the islands of Curagao, Aruba, and Bonaire m I960. Stud. Archaeol. Netherlands Antilles, III: 1-24. Hummelinck, p. W. 1940. Mollusks of the genera Cerion and Tiidoni. Stud. Fauna Curasao, Aruba, Bonaire, and the Venezuelan Islands, 2: 43-82. 1953. Rotstekeningen van Curasao, Aruba en Bonaire. (Linear rock designs of Curasao. Aruba, and Bonaire). West-Indische Gids, 34: 173-209. 1957. Rotstekeningen van Cursn^diO, Aruba en Bonaire. West- Indische Gids, 37: 93-126. Martin, P. S., and H. E. Wright, Jr. (eds.). 1967. Pleistocene Extinctions: the Search for a Cause. New Haven, Yale Univ. Press. 453 pp. Milliman, J. D., and K. O. Emery. 1968. Sea levels during the past 35,000 years. Science, 162: 1121-1123. Redfield. a. C. 1967. Postglacial change in sea level in the western North Atlantic Ocean. Science, 157: 687-691. Rouse, I. 1960. The entry of man into the West Indies. Yale Univ. Publ. Anthrop., No. 61. 26 pp. 1964. Prehistory of the West Indies. Science, 144: 499-513. 1966. Paleo- and Meso-Indians of the Caribbean area. Qua- ternaria, 8: 125-132. 26 BREVIORA No. 372 Rouse, I., and J. M. Cruxent. 1963. Venezuelan Archaeology. New Haven, Yale Univ. Press. 179 pp. Simpson, G. G., A. Roe, and R. C. Lewontin. 1960. Quantitative Zoology. New York, Harcourt, Brace and Co. 440 pp. Stuiver, M., and H. E. Suess. 1966. On the relationship between radio- carbon dates and true sample ages. Radiocarbon 8: 534-540. Tacoma, J. 1959. Indian skeletal remains from Aruba. Stud. Phys. Anthrop. Netherlands Antilles, II: 95-112. Tamers, M. A. 1967. Instituto Venezolano de Investigaciones Cientificas. Natural radiocarbon measurements 111. Radiocarbon, 9: 237-245. Warmke, G. L., and R. T. Abbott. 1961. Caribbean Seashells. Livingston Co. 346 pp. Weyl, R. 1966. Geologic der Antillen. Berlin, Gebriider Borntraeger. 410 pp. BREVIORA Mnasemim of Comiparative Zoology Cambridge, Mass. 31 March, 1971 Number 373 THE CHANARES (ARGENTINA) TRIASSIC REPTILE FAUNA. VIII. A FRAGMENTARY SKULL OF A LARGE THECODONT, LUPEROSUCHUS FR ACTUS Alfred Sherwood Romer Abstract. Incomplete remains of a large skull, not improbably repre- S2nting a raiiisuchid thecodont from the early Middle Triassic of Argentina, are described as Luperosiichiis fractns, gen. et sp. nov. Large dermal scutes, found isolated, may pertain to this form. INTRODUCTION A moderate number of specimens assignable to the reptilian order Thecodontia are present in our Chaiiares collections. Apart from materials that are difficult of interpretation or assignment, there are definitely present: ( 1 ) a small and primitive member of the Ornithosuchidae; (2, 3) two long-snouted forms, with gen- eral proportions resembling crocodilians, but without any positive indications of affinities with that group; (4) a small form with a very lightly built skull, of which the limbs are unknown; (5, 6) two forms known from very slender hind limbs of unusual con- struction; and (7) a large animal, probably a rauisuchid. repre- sented only by a partial skull. In addition there are various isolated materials, pseudosuchian in nature. In the present short paper 1 shall describe only the last specimen listed, leaving the others for later description. Since our collections were made, Sr. Bonaparte of Tucuman has made several visits to the Chaiiares region, and found, inter alia, a fair amount of thecodont material. He invited me to make use of this in my work on thecodonts, and in June 1970 I spent a week in Tucuman studying this material. I found no identifiable forms not already present in the Harvard-La Plata collections, BREVIORA No. 373 but in a number of regards his materials supplemented ours and substantiated our conclusions. I am deeply grateful to Sr. Bona- parte and the authorities of the Instituto Lillo for placing this material at my disposal. LuPEROsucHUS FRACTUS gen. et sp. nov. Combined generic and specific diagnosis. A large thecodont, with an estimated skull length of about 60 cm, probably pertain- able to the family Rauisuchidae. A shtlike opening apparently present posterior to the nares between premaxilla and maxilla; antorbital opening large; apparently no parietal foramen; lateral temporal opening with vertical posterior border; archosaur type of otic notch partially developed. Holotype of the species. La Plata Museum 1964-X1-14-9, an incomplete skull, consisting of most of the dermal roof and part of the left side of the "face" collected from the Chanares Forma- tion in La Rioja Province, Argentina, north of the north fork of the Chanares River, about 5 km NE of the point where this river emerges into the Piano de Talampaya. The generic and specific names refer to the fragmentary and perplexing nature of the type material. I am indebted to National Science Foundation Grant GB-2454 for aid in the collecting of the material and to further grants for its preparation and for publication costs. Figure 1. Side view of the fragmentary type skull of Liiperosiicluis frac- tus. as preserved. X 1/6. 197 LUPEROSUCHUS SKULL Description. This form is represented by a single fragmentary specimen that includes most of the dorsal surface of a skull and part of the dermal bones of the left side of the skull (Figs. 1,2). The condition of the material is none too good, and sutures are generally difficult to determine. The specimen was found close beside the skeleton of a dicynodont. Near it were found a con- siderable number of weathered scraps of bone; whether they belong to the specimen here described or to the dicynodont is uncertain. The animal was a large one; the portions of the skull preserved measure 54.5 cm in length and if the missing anterior and posterior regions be restored, the length in life would have been about 60 cm. In general, the reptiles present in the Chanares Formation are of modest size; apart from this Liiperosuchus specimen, the dicynodonts are the only large animals known. As mentioned, most of the skull roof is preserved; the cranium was obviously long and slender, as in many early archosaurs. Posteriorly, the parietals are incomplete, and their posterior exten- sions, which presumably formed the median boundaries of the superior temporal fenestrae, are missing. There was no parietal foramen in the portion of the bone preserved (although it may possibly have been present in the missing posterior portion). A median longitudinal suture can be made out for almost the entire length of the roof as preserved. Other sutures are obscured by poor preservation and bone fusion in this seemingly mature skull. Figure 2. Dorsal view of the skull, restored. Parts present in stipple. Abbreviations: /. frontal; /. jugal; /. lacrimal; m, maxilla; /;. nasal; p, pari- etal; pj, postfrontal; pm. premaxilla; po, postorbital; prj. prefrontal; qj. quadratojugal; sq, squamosal. X 1/6. 4 BREVIORA No. 373 I have restored the probable sutural pattern, but it should be emphasized that uncertainties exist. The area of the posterior part of the frontals and the median portion of the parietals is depressed, and bounded on either side by prominent ridges running back along the lateral margins of the frontals and continuing backward along the parietals. Postor- bitals are present along the back portion of the supraorbital rims and extend backward and medially to meet the parietals along the anterior border of the superior temporal fenestrae. As far as can be determined, there were large postfrontals, forming part of the upper margins of the orbits and extending back between frontals and postorbitals to gain contact with the parietals. The frontals are broad posteriorly; anteriorly they become reduced in width between the prefrontals. It is possible that the frontals entered the orbital margins briefly, but imperfections of the speci- men render this uncertain, and they may have been excluded by a narrow contact between postfrontals and prefrontals. The latter elements appear to be relatively narrow, projecting somewhat outward over the anterodorsal corner of the orbits and extending a modest distance forward on either side on the dorsal surface. Anteriorly, about opposite the front margin of the antorbital vacu- ity, the nasals, as seen in side view, rise upward markedly above the general line of the skull roof on their forward course. I was at first inclined to believe this appearance was due to post-mortem distortion; however, inspection indicates that this "roman nosed" elTect is a real structural feature; the conjoined nasals form a sharp ridge below which the two bones are apposed for some dis- tance and then slant outward toward either side ventrally. It seems obvious that the nasals are incomplete anteriorly. Ventrally the point of separation of the nasal from the posterior extension of the premaxilla below it indicates the probable position of the posterior angle of the naris. Most of the premaxilla is missing. A small fragment of bone attached to the anterior end of the maxilla presumably represents the most posteroventral position of the bone. More dorsally a band of bone with well-defined margins extends dorsoposteriorly between maxilla and nasal, indi- cating a posterior extension of the premaxilla that excludes the maxilla from the narial margin, as in certain other thecodonts (and ornithischians). The whole aspect of the anterior portion of the skull, as far as preserved, strongly suggests the presence 1971 LUPEROSUCHUS SKULL 5 of an expanded narial area, and in Figure 3 I have restored the narial region on this supposition. A large portion of the maxilla is present. A ventroanterior area, much thickened, represents the ventral margin adjacent to the premaxilla. This region is excavated internally and, although preservation is imperfect, represents the area of insertion of a series of large, probably subthecodont, anterior maxillary teeth. Above this region the anterior margin of the maxilla slants upward and backward parallel to the posterodorsal extension of the pre- maxilla. In the specimen as preserved the two bones are separated here by a long if narrow slit. For much of this distance the facing margins of both bones are broadened so that they can readily be apposed to one another; hence, when I first attempted a restora- tion of the skull, I placed these margins in firm apposition. But in contrast to the close union of all other portions of the skull, in the specimen as preserved, there was here a very distinct separa- tion, suggesting that a slitlike opening was present in life. Dr. W. D. Sill, who is currently studying Saurosuchiis, a seemingly related form from the Ischigualasto Formation, informs me that such an opening was definitely present in that genus, and I have therefore indicated such an opening in the restoration in Figure 3. I have no worthwhile suggestion as to the possible function of this slit. Back of the nasal region, dorsal and lateral surfaces are sharply separated for most of the skull length and, even allowing for possible crushing, it seems certain that the side walls descended nearly vertically from the lateral dorsal ridges. A short length of maxilla is preserved ventrally; above, there is a broad plate of bone apparently formed by the maxilla, extending back above the antorbital fenestra. The margins of this fenestra are, for the most part, clearly outlined; it was an opening of considerable size. In many advanced thecodonts the fenestra is centrally situated in a depressed area of the cheek; in this specimen the anterior rim of this depression is clearly incised in the maxilla. The suture between prefrontal and lacrimal is not clear, but the latter bone apparently includes the posterior part of the upper margin of the antorbital fenestra as well as the preserved portion of a stout bar of bone that separates orbit and antorbital fenestra. Behind the orbit, the bar of bone between orbit and lateral temporal fenestra is completely preserved, and there are indications of a suture well BREVIORA No. 373 down this bar, between postorbital and jugal. A fraction of the latter bone is present, defining the lower margin of the orbit, a section of the cheek rim, and a small area of the anteroventral margin of the lateral temporal fenestra. Above this fenestra a stout bar of bone is present, presumably formed anteriorly by the postorbital, posteriorly by the squamosal (the suture between the two is not clear). An incomplete flange of the latter bone extends directly downward as part of the posterior border of the lateral fenestra. The squamosal extended backward beyond the level of this descending flange, although this extension is broken off in the specimen. The vertical descent of the squamosal flange indicates that the posterior border of the fenestra had not acquired the V-shaped contour seen in various more advanced thecodonts; on the other hand, the posterior prong of the squamosal suggests the initiation of a typical archosaur type of otic notch. In Figure 3 I have freely restored the skull in side view to give a suggestion of its probable appearance in life. Despite the incom- plete nature of the evidence I do not think that there can be too great a departure from life conditions in most regards. Most doubtful, because of lack of material, is the suspensorial region. Systematic position. As to relationships of Luperosuchiis, an early Middle Triassic form, one tends to think first of the larger erythrosuchid members of the Proterosuchia — a group most recently discussed by Reig (1970), and by Charig and Reig (1970). Primitive, for example, is the apparent presence of a large postfrontal. Liiperosuchus, however, is more advanced than Figure 3. Side view of the skull, restored. Abbreviations as in Fig. 2. X 1/6. 1971 LUPEROSUCHUS SKULL proterosuchians in various regards, such as the large size and incised nature of the antorbital fenestra, probable absence of a parietal foramen, and the apparent beginning of the pseudosuchian development of an otic notch. More reasonable is assignment to the Rauisuchidae (or Prestosuchidae), a family of large but rela- tively primitive Middle Triassic thecodonts, first seriously studied by Reig (1961). Included here may be such forms as Ticinosii- chiis from the European Anisian (Krebs, 1965), Fenhosuchus and, doubtfully, Shansisiichus from China (Young, 1964), Stcigo- nosuchus (Huene, 1938), and Mandasuchus from the Manda beds of East Africa. The presence of rauisuchids in the Middle Triassic of South America is well attested by the presence of Rauisiichiis and Prestosuclms from the Santa Maria of Brasil and Saiirosuchus of the Middle Triassic Ischigualasto Formation of Argentina. The material of Liiperosuchus is too fragmentary to warrant any extended discussion of rauisuchid relationships. Are they, as Reig believes (1970, fig. 10), a side branch from a somewhat advanced pseudosuchian stock, or could they have progressed in parallel fashion from the proterosuchian base of the Thecodontia? Are they a sterile group, without descendants, or could they be related to the ancestry of certain of the later saurischians, the Prosauropoda (Palaeopoda) or, more especially, ancestral Sauro- poda? It is possible that Dr. SilFs current studies of Saurosuchus will shed light on rauisuchid relationships. Dermal scutes. In two instances we found in the Chanares region large scutes not definitely associated with other identifiable skeletal remains (Fig. 4). They are too large to be attributed to any of the other (and much smaller) thecodonts present in our 1 2 CM J I Figure 4. Two dermal scutes, possibly referable to Luperosuchus. 8 BREVIORA No. 373 collections (and they are not, of course, attributable to the synap- sids, which make up the remainder of the materials collected). Of known forms from the Chanares, Luperosuchus is the only one to which they could have belonged and, since comparable scutes are known in other rauisuchids, we may provisionally assign them to the present genus. None of the scutes is perfectly pre- served. One type, rectangular in shape, has a thickened, saw- toothed border along one edge, indicative of an interdigitating connection with another element. Such scutes are presumably paramedian paired scutes, found in various other thecodonts. Subcircular scutes, also present, may be more lateral elements or median caudal ones. REFERENCES CITED Charig, a. J., AND O. A. Reig. 1970. The classification of the Protero- siichia. Biol. Jour. Linn. Soc, 2(2): 125-171. HuENE, F. VON. 1938. Ein grosser Stagonolepide aus der jiingeren Trias Ostafrikas. Neues Jahrb. Min. Geol. Pa!., Beii.-Bd., 80: 264-278. Krebs, B. 1965. Ticinosuchiis ferox nov. gen. nov. sp. Ein neuer Pseudo- suchier aus der Trias des Monte San Giorgio. Schweiz. Palaont. Ab- handl., 81: 1-140. Reig, O. A. 1961. Acerca de la Posicion Sistematica de la Familia Rauisuchidae y del Genero Saiirosiichiis (Reptilia, Thecodontia). Publ. Mus. Munic. Cienc. Nat. Tradic. Mar del Plata, 1(3): 73-114. 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. Young, C. C. 1964. The pseudosuchians in China. Palaeont. Sinica, new sen C, No. 19: 107-205. BREVIORA Mmseiuijni of Comparative Zoology Cambridge, Mass. 15 June, 1971 Number 374 THE FISHES OF THE MALAYSIAN FAMILY PHALLOSTETHIDAE (ATHERINIFORMES) Tyson R. Roberts^ Abstract. The tiny fishes of the family Phallostethidae, from Malaya and Thailand, develop a large penis and differ radically in structure of the bilaterally asymmetrical priapium from the somewhat larger fishes of the more widely distributed Neostethidae, the only other family in the sub- order Phallostethoidea. A morphological characterization of the Phallo- stethidae is given and its features compared to those of Neostethidae. There are three species: Phallostethus diinckeri Regan (1913), known only from the type specimens collected in the mouth of the Muar River in Johore, Malaya; Phenacostethus smithi Myers (1928), known from the types and many other specimens collected in the khlongs of Bangkok and reported in this paper from Chantaburi Province in southeast Thailand; and Phenacostethus posthon, new species, from the Indian Ocean coast of peninsular Thailand. Ph. posthon and Ph. smithi differ considerably in morphology of the priapium and penis. Furthermore, the asymmetrical priapium in Ph. posthon is invariably sinistral. In all other phallostethoids, so far as known, it may be either sinistral or dextral; in Ph. smithi the ratio of sinistral and dextral males is near equality (Hubbs and Hubbs, 1945). The ecology of phallostethids is described for the first time, excepting some brief remarks by H. M. Smith (1927; 1945), who seems to have confused Phenacostethus in the field with neostethids and perhaps with Oryzias. An hypothesis is offered that the selective advantage of internal fertilization in Phallostethoidea (an oviparous group) lies in permitting temporal separation of mating and spawning activities, corresponding, re- spectively, with periods of low water and high water in habitats subject to strong tidal fluctuations. Two trends in the reproductive biology of atheriniform fishes that might be conducive to the evolution of internal fertilization are: 1) towards eggs in which either embryonic development is slowed down or temporarily arrested, or hatching of embryos is deferred; and 2) away from expelling all ovulated eggs at once and towards expel- ling them in small batches or even singly. 1 Museum of Comparative Zoology, Harvard University, Cambridge, Massachusetts 02138. 2 BREVIORA No. 374 ACKNOWLEDGMENTS My studies in Thailand were carried out under the sponsorship of the Thai National Research Council. The College of Fisheries of Kasetsart University acted as my host, providing laboratory space, library facilities, and access to fish collections, including specimens of Phenacostethus smithi collected by H. M. Smith. I wish to thank the following people for their substantial aid, par- ticularly in the aspects of my fieldwork concerning phallostethoids: Dean Jinda Thiemeedh, Mr. Prajit Wongrat, and Mrs. Supap Monkolprasit of the College of Fisheries; Miss Prachuab Suk- charean and Mr. Sopon Chantarat of the Marine Fisheries Station of Songkhla; and Dr. Vagn Hansen, Director of the Phuket Marine Biological Center. Mrs. Monkolprasit was extremely helpful, par- ticularly in arranging my trips. Dean Thiemeedh kindly en- couraged my work and arranged transportation to Chantaburi Province, where, with Mr. Wongrat's help, not only Ph. smithi, but also large series of the neostethids Neostethus siamensis (hitherto known only from a single female) and Ceratostethus bicornis (previously unrecorded from Thailand) were obtained. Mr. Wongrat also helped find Phenacostethus near Bangkok. Miss Sukcharean arranged my travels from Songkhla to Satul; Mr. Chantarat accompanied me on this trip and helped collect the first specimens of Phenacostethus posthon. Dr. Hansen arranged my fieldwork in Pungah. For translations of the papers by Aurich and Woltereck I am obliged to Miss Deborah White and Dr. Elizabeth Deichmann. Prof. George S. Myers read the paper in manuscript. INTRODUCTION This account of the family Phallostethidae is the first paper dealing with the fresh- and brackish-water fishes collected by me in Thailand from April 15 to July 14, 1970, and deposited in the fish collection of the Museum of Comparative Zoology. Collecting phallostethoid fishes was one of the main objectives of my field- work in Thailand. In addition to Phenacostethus smithi Myers (1928) and the new phallostethid described in this paper, large series of the neostethids Neostethus siamensis Myers (1937) and Ceratostethus bicornis (Regan, 1916) were obtained. These represent the first specimens of Ceratostethus recorded from Thai- land, and the only specimens of Neostethus siamensis other than 1971 PHALLOSTETHIDAE 3 the female holotype. (/V. siamensis is close to, and perhaps spe- cifically identical with, N. lankesteri Regan (1916), the type locality of which is the mouth of the Muar River, Johore, and Singapore.) The osteology and functional anatomy of Cerato- stethus will be considered in another paper. The Phallostethoidea are small, highly specialized fresh- or brackish-water fishes in which males have a remarkable sub- cephalic copulatory organ, the priapium. The skeleton and mus- culature of this complicated bilaterally asymmetrical organ, which functions both in clasping and intromission, are derived mainly from the pelvic fins and girdle. Minor contributions come from the first pair of ribs and anteroventral part of the pectoral girdle. In females the pelvic fins are absent or vestigial. Aurich (1937) divided the Phallostethoidea into two "Familien," but gave these divisions names in subfamily form, Phallostethinae and Neo- stethinae. Berg (1940: 465-466) recognized them as families, Phallostethidae and Neostethidae, as did Rosen (1964: 261) and Greenwood et al. (1966: 398). Neostethidae, comprising eight genera and about 15 species, have been recorded from Thailand, Malaya, Sumatra, Borneo, and the Philippine Islands. Six of the genera — Gulaphallus Herre (1925), Mirophallus Herre (1926), Plectrostethiis Myers (1935), Solenophallus Au- rich (1937), Ctenophallus Herre (1939), and Manacopus Herre (1940) — are known only from the Philippines. The Phallostethidae, even more specialized (and rarer in mu- seum collections) than Neostethidae, comprise three species from the Malay Peninsula and adjacent parts of Thailand. Phallostethus dunckeri Regan (1913), the first phallostethoid to be described, is known only from the type specimens Duncker collected previous to 1904 at the mouth of the Muar River, about 20 miles south of Malacca, Johore Province, Malaya (Duncker, 1904: 171). Phenacostethus smithi, hitherto known only from several large series collected by H. M. Smith from khlongs in Bangkok, was obtained by me at Bangkhen (a suburb of Bangkok) and at Chantaburi, near the southeast corner of Thailand (near Cam- bodia). The third species, described in this paper, is from the Indian Ocean coast of Thailand. All localities where phallostethids have been collected are shown in Figure 1. The present paper presents characterizations of the family Phallostethidae and of the phallostethid species, a description of the new species just mentioned, and observations on the ecology of Phallostethidae. BREVIORA No. 374 ir 10" 5*- - 10" 15* 5* 100* lOS* Figure 1. All localities where Phallostethidae have been collected. 1. Mouth of Muar River at Bandar Maharani (type locality of Phallostethus dimckeri Regan, 1913); 2. Bangkok (type locality of Phenacostethiis smithi Myers, 1928); 3. Bangkhen, a suburb of Bangkok (Ph. smithi); 4. Chanta- buri City {Ph. smithi); 5. Khlong La Ngoo, 48 km NW of Satul Town, 6° 52' 30" N, 99° 48' 10" E (type locality of Phenacostethiis posthon n. sp.); 6. Khlong Kla Sohm, 15 km S of Pungah Town, on Pakasem Road from Pungah going towards Phuket Island (type locality of Ph. posthon n. sp.) Definition of the suborder Phallostethoidea and discussion of phyletic trends in the Phallostethoidea will be presented in a forthcoming paper on Ceratostethus. Ph. smithi provides an Asian example of the phenomenon (of which Africa and South America provide numerous examples) 1971 PHALLOSTETHIDAE 5 that the smallest fishes amidst the richest tropical freshwater faunas are representatives not of the dominant fish groups present, but of groups with a marginal distribution pattern. The ecology of such minute fishes, when known, usually proves to be highly specialized. Rosen (1964) united the exocoetoids, scomberesocoids, adri- anichthyoids, cyprinodontoids, atherinoids and phallostethoids in a new order, the Atheriniformes. Atherinoids and cyprinodont- oids were widely separated in earlier classifications, the artificial- ity of which has become increasingly apparent. Such similarities as were noted between atherinoids and cyprinodontoids had usually been attributed to convergence. Having personally investi- gated the osteology of phallostethoids (which evidently are re- lated to atherinoids), atherinoids, and cyprinodontoids, and reviewed much of the literature on osteology and reproductive biology of these groups, I am inclined to believe that they may be related. Some of the similarities in reproductive biology were first brought to my attention in a talk given by Neal R. Foster at the 1968 meetings of the American Society of Ichthyologists and Herpetologists. Note on the figures. Figures 2-5, prepared with the aid of a Wild microscope and camera lucida, are based on formalin speci- mens. Formalin specimens of Phallostethidae are more nearly normal in appearance than alcoholic specimens, since the latter invariably undergo at least some shrinkage. In specimens that have been transferred from the original formalin fixative to 60 per cent ethyl alcohol for permanent storage, teeth protrude more from the gums, scales stand out more clearly, the membranous dome atop the head has disappeared, and the caudal peduncle is decidedly narrower. CHARACTERIZATION OF THE FAMILY PHALLOSTETHIDAE The following characterization of the Phallostethidae is based on my observations of Ph. sinithi and Ph. posthon, including study of alizarin preparations, and on accounts of Ph. smithi by Myers (1928), Bailey (1936), and TeWinkel (1939), and of Phallo- stethus dunckeri by Regan (1913; 1916). 1 . Slender elongate phallostethoids, very delicate, largely trans- lucent, with deciduous scales; externally visible concentrations of melanophores restricted to the top of the braincase, middle of the 6 BREVIORA No. 374 dorsum, midlateral intermuscular septum, priapium, and bases and edges of fin rays; maximum standard length about 23 or 24 mm. 2. Dorsum of head with a translucent, membranous dome. 3. Mature individuals of both sexes with a bright orange-yellow bar on caudal peduncle. 4. Branchiostegal rays 4. 5. Main "externalized" clasping bone in the priapium is the toxactinium; ctenactinium reduced or absent. 6. A greatly enlarged, oval, concave pad, or pulvinulus, shghtly posterior to toxactinium. 7. Vas deferens terminating in a large penis that projects con- siderably from the priapium. 8. Pelvic spines or rays, if present, greatly reduced and modi- fied beyond recognition. 9. Vas deferens highly coiled, forming a sort of epididymis. Comments on the family characters. Neostethids, while small as fishes go, are all or almost all larger when adult than phallo- stethids. Most, if not all, neostethids are hardier fishes than phal- lostethids and have relatively adherent scales. While they also are largely translucent, neostethids usually have relatively more melanophores than phallostethids. The epidermis paralleling the scale margins is often well provided with melanophores in neo- stethids but invariably devoid of melanophores in phallostethids. Neostethids (Ceratostethus and Neostethus), even at comparable sizes, lack a membranous dome on the dorsum of the head, or if one is present, it is not notably elevated. According to TeWinkel (1939) this region bears sensory canals in both Phenacostethus and Gulaphallus. In neostethids large sensory pores are evident in the frontal region, whereas the membranous dome in phal- lostethids is apparently entire. With regard to characteristic number 3 (bright orange-yellow markings at the base of the caudal fin and sometimes at the origin of the anal fin), these are present in Uving specimens of Ph. smithi and in Ph. posthon. There is no way of teUing whether such spots are also present in Phallostethus. The orangish yellow coloration, contained in chromatophores (approximately 50-75 chromato- phores constituting the caudal base mark) gradually disappeared after a few weeks of preservation in formalin. Similar markings 1971 PHALLOSTETHIDAE 7 were definitely absent in live specimens of Neostethus and Cera- tostethus observed by me, nor is there any mention of such marks in the literature on Neostethidae. Concerning character 4, Neostethidae usually have 5 branchio- stegal rays. The number of branchiostegal rays in Phallostethus is unknown. Characters 5-9 concern the priapium, which differs funda- mentally from that of Neostethidae. In Neostethidae the main "externalized" bony clasping element, the ctenactinium, is appar- ently a modified pelvic fin ray or spine. The mam "externalized" bony element in the priapium of phallostethids, the toxactinium, is not homologous with the ctenactinium. My observations con- firm Bailey's view (Bailey, 1936: 463, 471) that it is homologous with the pulvinular bone, one of the anteriormost internal bony elements in the priapium of neostethids. The homologies of this element are unclear, but it is almost certainly not a modified pel- vic ray or spine. The ctenactinia of phallostethids, which may or may not be homologous with the elements called ctenactinia in neostethids, are greatly reduced in size. The ctenactinium of Phal- lostethus dimckeri, while relatively short, bears several "teeth" or sharp projections; these are perhaps comparable to the single curved hook present near the base of the ctenactinium in Neo- stethus. The pulvinulus of phallostethids probably functions as a pad in conjunction with the toxactinium. A homologous but much smaller pulvinulus is present in neostethids. A striking character- istic of the phallostethid priapium is the development of a large penis. In Phenacostethus posthon the organ is entirely smooth; in Ph. smithi its distal half bears a series of stiff ruffled pleats. The development of a large penis evidently occurred independently in the neostethid Mirophallus bikolanus (Herre, 1926, pi. 3, fig. 1 ) . In most neostethids a compUcated flap covers the opening of the vas deferens (Aurich, 1937). This flap is absent in PhaUo- stethidae. Various bony elements in the priapium of phallostethids, including a peculiar slender element lodged in the concave side of the penis bone in Ph. smithi, may be homologous with pelvic rays, but the priapium of phaUostethids does not bear any branched elements that obviously are relatively unmodified pelvic rays. Several branched pelvic rays of relatively normal appear- ance occur in the neostethids Ceratostethus, Neostethus, Soleno- phallus, and Gulaphallus (personal observations; Aurich [1937], TeWinkel [1939], Woltereck [1942 a, b]). 8 BREVIORA No. 374 The priapium of Phallostethus is clearly of the same general type as that of Ph. smithi and Ph. posthon. In Phallostethus the toxactinium and pulvinulus are very similar to these structures in Phenacostethus. It is likely that Phallostethus develops a large penis. The penis in phallostethids, like the ctenactinium of neo- stethids, only reaches its full development in the largest males. Regan (1916: 22) hypothesized that in Phallostethus the tox- actinium grips the female under the chin or is held in her mouth, while the serrated edge of the ctenactinium gives a firm hold on the pectoral region in front of and on the far side of the genital orifice, in order that the seminal papilla could be placed against it or introduced into it. In Neostethus he hypothesized that the female is held across the back of the head by the ctenactinium, the anterior descending part of which lies on the side of the female away from the male. Copulation has yet to be observed in Neo- stethus or in any phallostethids. It has been observed only in the neostethid Gulaphallus mirabilis. In this species the female is held across the back of the head by the ctenactinium; the "second ctenactinium" of Gulaphallus mirabilis, which actually is an exter- nalized pelvic bone, apparently rests or presses against the female's opercular region on the side next to the male (Villadohd and Man- acop, 1934: pi. 5, fig. 2). The reduced ctenactinium of Phallo- stethidae could hardly function in the same manner as the elongate ctenactinium characteristic of all neostethids. It may be that the toxactinium is held, not under the female's chin or in her mouth, but atop the front of her head. In Phallostethus (Regan, 1916: 19, fig. 14) the vas deferens is highly coiled within the abdominal cavity to form a sort of epididymis. TeWinkel (1939) reports a similar coiling of the vas deferens in Ph. smithi. In Neostethus lankesteri (Regan, 1916: 10, fig. 6) and in Gulaphallus mirabilis (Villadolid and Manacop, 1934: pi. 3, fig. 4) the vas deferens is unconvoluted from its origin on the testis to where it enters the priapium, then forms an expanded loop inside the priapium. The sperm of Neo- stethus lankesteri (Regan, 1916: 13, fig. 9) are concentrated into "spermatophores," more properly called spermozeugmata (Niel- sen et al., 1968: 248). Regan (1916: 19) stated that Phal- lostethus evidently did not produce "spermatophores" like those of Neostethus. In mature Neostethus and Ceratostethus (per- sonal observation) the posteriormost portion of the priapium is 1971 PHALLOSTETHIDAE 9 sometimes greatly swollen with closely packed, adherent small vesicles that presumably are spermozeugmata. The priapium apparently does not become similarly swollen in phallostethids. Insofar as can be determined from examination of the bones only, the contributions of the first pair of ribs and of the shoulder girdle to the priapium is the same in Phallostethidae and Neostethidae. THE SPECIES OF PHALLOSTETHIDAE Phallostethus dunckeri Regan 1913 Phallostethus dunckeri Regan, 1913: 550, figs. 1-4 (original description; types from mouth of Muar River at Bandar Maharani, Johore, Ma- laya; soft anatomy). — Regan, 1916 (soft anatomy, histology, oste- ology, comparison with Neostethiis). This species is known only from the specimens collected by G. Duncker at the mouth of the Muar River at Bandar Maharani, Johore, Malaya. The following statements are based on the two accounts of this species by Regan (1913, 1916). It is character- ized by an exceptionally long anal fin, with 26-28 elements (eight to ten rays more than are found in phallostethoids with the next highest number of anal fin elements), an anal base about 30 per cent of the standard length (compared to anal base 20-25 per cent of standard length in all other phallostethoids) and origin of anal considerably nearer to snout tip than to end of hypural fan (anal origin slightly to considerably nearer to end of hypural fan than to snout tip in all other phallostethoids), and by a ser- rated ctenactinium (Regan, 1916: fig. 13). Regan did not have very many specimens (some were used for histological preparations or cleared in oil of cloves for bone study). Dr. P. H. Greenwood informs me that the British Museum (Natural History) has four specimens (types?) of Phallostethus dunckeri in rather poor con- dition. Although Regan's specimens of Phallostethus dunckeri (at 23-29 mm in total length) are larger than Phenacostethus, and the males figured by him have well-developed priapia, perhaps the penis is not fully developed in them. This seems likely, be- cause in both species of Phenacostethus the largest females are only about one-half to one milUmeter longer than the largest males, whereas Regan's largest female of Phallostethus is six mil- limeters longer than his largest male. In Phenacostethus the penis is fully formed only in the largest males. 10 BREVIORA No. 374 The second dorsal fin of P. dimckeri has eight to ten elements (a number found in some Neostethidae); the highest number in Phenacostethus is seven. It is unknown whether Phallostethus has a first dorsal fin. Regan did not mention its presence, but he over- looked the first dorsal in his material of Neostethus lankesteri and Ceratostethus bicornis. Phallostethus probably has a small spinous first dorsal fin, but the absence of one would not be unique among Phallostethoidea: the first dorsal is lacking in the neosteth- ids Mirophallus bikolamis Herre (1926), (Herre, 1942: 141; Myers, 1937: 142) and Solenophallus thessa Aurich (1937: 265). Phenacostethus smithi Myers 1928 (Figures 2, 4, 6) Neostethus lankesteri (not of Regan) Smith, 1927: 353-355 (misidenti- fication). Phenacostethus smithi Myers, 1928: 6, figs. 1 and 2 (original description; types from Bangkok; holotype in American Museum of Natural His- tory). — Bailey, 1936 (osteology).— TeWinkel, 1939 (soft anat- omy).—Smith, 1945: 475 (Bangkok; synonymizes Ph. thai Fowler; ecology). Phenacostethus thai Fowler, 1937: 219; figs. 189, 190 (original description; Bangkok; holotype in Philadelphia Academy). Material studied. MCZ 47055, 13 specimens, five females 13.7-14.8 mm and eight males 13.6-14.9 mm (three with toxac- tinium arising on left side, five on right side), from Khlong Bang- khen at bridge on Nzarm Wong Wan Road, a few km west of Kasetsart University, Bangkhen, Bangkok, Thailand, 30 April 1970; MCZ 47299, 20 specimens, three females 13.4-15.6 mm and 17 males 12.7-14.9 mm (12 with toxactinium arising on right side, five on left side), from Khlong Kee Nawn, behind Cathohc church in Chantaburi City, Chantaburi Province, Thailand, 5 May 1970. Myers' figure 1 of Ph. smithi does not show the membranous dome of the dorsum of the head (Fig. 2) ; it is likely to be shrunken in alcoholic specimens. The adult male in Myers' figures 1 and 2, at 13.5 mm in standard length, does not have the penis as fully developed as in my specimens 14.3 (Fig. 2), 14.1 (Fig. 4), and 14.5 (Fig. 6) mm in standard length. In these specimens the penis is much larger, and its distal end bears a series of a half- dozen or more crenulated radial folds or extensions resembling a 1971 PHALLOSTETHIDAE 11 S m ON OS N U E "a o m N U OJ a B B m 00 o\ en u en % Oi 'o -s: a. O -s: a. 3 12 BREVIORA No. 374 set of ruffled lace cuffs (best shown in Fig. 4). These folds can be seen in one of Bailey's anatomical figures based on material of Ph. smithi collected by Hugh M. Smith. Direct comparison of Ph. smithi collected by Smith in Bangkok (deposited in the fish collection of the College of Fisheries, Kasetsart University, and in the MCZ fish collections) with my material from Bangkhen and Chantaburi indicates that only one species is involved. Live speci- mens of Ph. smithi of both sexes from Bangkhen and from Chanta- buri had a bright orange-yellow, anteroventrally-posterodorsally oriented bar on the lower half of the caudal peduncle. This bar disappeared from specimens kept a few weeks in formahn. The color was due to about 50-75 chromatophores. There is a small, shiny blue spot over the brain (well behind the eyes), and a very small silvery spot in the middle of the eyes' dorsal surface. Alizarin preparations reveal the first dorsal fin consists of a single tiny spine, as indicated by Myers. Most specimens are miss- ing many scales. Head scaleless. Abdominal keel scaleless in both sexes (abdominal keel scaled in females of Ceratostethus bicornis). Scales on body cycloid, number of scale rows corre- sponding closely with the number of myotomes. Teeth in both jaws in a single series; medial portion of premaxillary with about eight sharp conical teeth, lateral (expanded) margin of premaxil- lary with about seven to nine slightly larger conical teeth; medial portion of dentary with about a dozen conical teeth opposing but much smaller than those on medial portion of premaxillary. For counts of fin elements and vertebrae based on alizarin preparation see Table 1. Phenacostethus posthon, new species (Figures 3, 5, 7) Holotype. MCZ 47300, a 16.7-mm male from Khlong Kla Sohm about 15 km southwest of Pungah Town, where it is crossed by a bridge on the Pakasem Road (between Pungah Town and Phu- ket), Pungah, Thailand. 29 June 1970. Paratypes. MCZ 47301, 58 specimens, comprising two im- matures, 11.9 and 12.0 mm, 22 females, 12.3-17.7 mm, and 34 males 12.7-17.0 mm (of which two males, 15.5 and 16.5 mm, are cleared and stained), same data as holotype; MCZ 47302, 20 specimens, comprising three immatures 9.6-12.0 mm, 10 females 12.7-18.0 mm, and 7 males 12.9-16.7 mm, from Khlong Langu 1971 PHALLOSTETHIDAE 13 at Langu Town, 48 km northwest of Satul Town (6° 52' 30" N, 99° 48^' 10^' E). 23 June 1970. The most obvious differences between this species, from the Indian Ocean coast of Thailand (Fig. 1), and Ph. smithi involve the priapium and the position of the dorsal fin. Granted that both species have the characteristic priapial elements of the family Phallostethidae, the priapium is, in fact, so different in the two species that the question may even be raised as to whether separate generic status is indicated. Differences in the priapium of the two species involve external morphology of the penis, skeleton of the penis, ctenactinium, tox- actinium, and the nature of the laterality of the priapium itself. In Ph. smithi 1 ) the distal portion of the penis is ruflfled (see description above); 2) the penial skeleton includes a large papil- lary bone with a slender penial bone lodged in its concave surface (Fig. 6; see also Bailey, 1936: 3 and 4); 3) the ctenactinium is relatively large and externally evident; 4) the toxactinium is rela- tively slender and gently curved; and 5) the priapium itself may be either sinistral or dextral. In Ph. posthon, on the other hand, 1 ) the penis is smooth; 2) the penial skeleton has a large papillary bone but the penial bone is absent; 3) the ctenactinium, if it is present at all, is reduced and hardly detectable externally; 4) the toxactinium is stouter and distinctly more sharply curved; and 5) the priapium is invariably sinistral (toxactinium arising on left side) in the material examined. Ph. posthon and Ph. smithi of both sexes can be distinguished at a glance by the position of the first dorsal fin relative to the anal fin base. In Ph. posthon the first dorsal (which has a single spine as in Ph. smithi) originates slightly posterior to the base of the last ray in the anal fin; in Ph. smithi it originates over the middle of the anal fin base. The first dorsal fin is somewhat closer to the second dorsal origin in Ph. posthon than in Ph. smithi. Ph. posthon is more elongate (depth of body at anal fin origin about 7, vs. 6 in Ph. smithi) and evidently a larger species. The average and maximum sizes of specimens in my two samples of Ph. posthon are definitely larger than in my two samples of Ph. smithi (which include specimens larger than those previously recorded). The largest male and female specimens of Ph. smithi are, respec- tively, 14.9 and 15.6 mm; of Ph. posthon, 17.0 and 18.0 mm (standard lengths). The largest specimens (both sexes) had a bright orange-yellow bar on the caudal peduncle, as in Ph. smithi, 14 BREVIORA No. 374 Figure 4. Phenacostethus smithi Myers (1928), ventral view of pri- apium, 14.1-mm specimen, MCZ 47055 (formalin). and also a smaller orange-yellow bar on the body next to the anal fin origin. As in Ph. smithi, there appears to be a one-to-one correspon- dence between the scale rows and the myotomes. The thin scales are extremely difficult to see in specimens under alcohol or water; and most specimens are missing many scales. I find in several specimens eight oblique scale rows between dorsal midline and anal fin origin, and six oblique scale rows between dorsal fin origin and anal base. Shape of jaw bones and disposition of teeth as in Ph. smithi. For fin and vertebral counts based on alizarin preparations of Ph. posthon see Table 1. Ph. posthon seems to have, on the aver- age, one less ray in the dorsal fin (verified by counts of unstained specimens) and one more vertebra than Ph. smithi. Right- and left-handedness in the priapiiim of Phallostethidae. 1971 PHALLOSTETHIDAE 15 Figure 5. Ph. posthon n. sp., ventral view of priapium in 16.9-mm para- type, MCZ 47301 (formalin). That the priapium of Neostethus lankesteri may be either right- or left-handed is stated by Regan (1916: 5). Regan did not spe- cifically state that the priapium of Phallostethus may be either way, but in a general discussion of the priapium as compared to copulatory organs of other fishes, he refers to its being either dextral or sinistral (Regan, 1916: 23). Regan's figures of Phal- lostethus depict both right-handed and left-handed males. It is conceivable that one or more of them might be reversed images, the images being reversed either by the illustrator or by the printer, so the figures cannot be taken as definite proof that P. dimckeri may be either left-handed or right-handed. Concerning Phena- costethus smithi, Hubbs and Hubbs (1945: 294) found that in 334 males (from material collected by H. M. Smith at Bangkok) 16 BREVIORA No. 374 posttemporal c lei thrum coracoid supracleithrum scapula tozactiiiium antepleural papillary axial Figure 6. Ph. smithi Myers (1928), skeleton of priapium and pectoral fin in 14.5-nim specimen, MCZ 47301. Terminology of priapial elements according to Bailey (1936). posttemporal cleithrum coracoid supracleithrum scapula radial toxactinium pulvinulus antepleural ribs anal basipenial prepapillary uncus anterior infrasulcar' posterior infrasulcar ctenactinium Figure 7. Ph. posthon n. sp., skeleton of priapium and pectoral fin in 16.5-mm paratype, MCZ 14301. Priapial elements of uncertain homology with those in Ph. smithi are unlabelled. 1971 PHALLOSTETHIDAE 17 the aproctal side was the left side in 155 and the right side in 179. The deviation from equality is statistically insignificant. In my samples from Bangkhen and from Chantaburi (too small for statistical analysis) I find both left- and right-handed males. In Phenacostethus posthon the toxactinium arises on the left side in every male in which the priapium is sufficiently developed to have a toxactinium (27 specimens from Pungah and five from Satul). To my knowledge this is the only phallostethoid in which the sidedness of the priapium appears to be fixed. Females of Phal- lostethidae are bilaterally symmetrical, the genital openings lying in the middle of the throat. ECOLOGICAL OBSERVATIONS The only ecological information Regan had about Phallostethus dunckeri was that it came from brackish water from the Muar River at Bandar Maharani. Duncker also collected Neostethus lankesteri in the same general locahty. Smith (1927; 1945: 476) made brief ecological observations on Phenacostethus smithi but in restrospect it seems that he some- times confused this little fish in the field with Neostethus or Cera- tostethus and perhaps even with Oryzias. I did not observe either phallostethids or neostethids at the surface, and feel that they generally keep well below the surface. Smith's remarks (that Phenacostethus "normally remain at or close to the surface, where they would be difficult to see were it not for a glistening yellow area on the top of the head'') sound more like Oryzias. The glisten- ing area on the top of the head in phallostethoids is relatively small and bluish, compared to the large yellowish glistening area in Oryzias, which is comparable to the striking head spots in rivuline cyprinodontids of the genera Epiplatys and Aplocheilus. In any event. Smith records that Ph. smithi abounds in freshwater pools, ditches and smaller canals in the Bangkok region, Hving in water that is nearly always muddy or turbid, and that the species is oviparous (although spawning was not observed); the egg-bearing and spawning periods are protracted, corresponding with the rainy season, and may extend from May to December. Small numbers were maintained in aquaria for a month by the daily introduction of fresh ditch or canal water to provide food in the form of minute crustaceans, protozoans, worms, etc. Smith found the larvae of anopheline mosquitoes much too large for Ph. smithi to ingest. 18 BREVIORA No. 374 Undoubtedly Ph. smithi once lived in innumerable khlongs and ditches in Bangkok which are today so polluted that they are inhabited only by the hardiest air-breathing fishes such as Anabas testudineiis. In Khlong Bangkhen, a few kilometers from Kasetsart University, I found Ph. smithi in association with a variety of primary and secondary freshwater fishes, principal among which were cyprinids mainly of the genus Rasbora, Dermogenys, Xenen- todon, Oryzias, Gobiopterus chuno (a small translucent goby), young Fhita, and Chaudhuria. The Phenacostethus were mostly in a little backwater, close to the main current in the khlong, in very turbid water. I was unable to taste any salt in the water. At Chantaburi City, Phenacostethus was collected in a turbid ditch or small khlong, Khlong Kee Nawn, behind the Catholic church. There were a number of houses along the khlong at this point, and considerable rubbish had been thrown into it, so col- lecting efl'orts were confined to dip-netting for Phenacostethus. Dermogenys, Brachygobius and Gobiopterus were collected inci- dentally. About two kilometers downstream, where the khlong traversed a large open field, we found minnows such as Oxygaster, Esomus and Rasbora (but no Phenacostethus) which indicates the water in the khlong was not brackish. At Satul and at Pungah Ph. posthon was obtained along the margins of khlongs or tiny branches of khlongs in which the water was swiftly flowing, highly turbid, and probably completely fresh. A few of the largest females (collected June 1970) were replete with ripe ova. Both localities were at places fully subject to tidal fluctuations, but far upstream from branches having water the least bit sahy to the taste. At Khlong Langu in Satul Province the water level was undoubtedly high because of heavy rains for the preceding 12 hours or more; Phenacostethus were collected in the khlong both on the falling tide and rising tide. No primary freshwater fishes were obtained. Species in the khlong where Ph. posthon was collected included Oryzias, Dermogenys, Chanda, Gobiopterus, Butis, and Tetraodon. In Khlong Kla Sohm, near Pungah, Ph. posthon was coUected in a narrow side channel more or less uniformly 3-4 feet deep, swiftly flowing, with hard-packed mud bottom, well shaded by Nipa palm and mangrove. Oryzias, Dermogenys, Gobiopterus, Periophthalmus, Tetraodon, and young Scatophagus were the only other fishes collected or observed in this channel. 1971 PHALLOSTETHIDAE 19 In contrast, on the more numerous occasions when I obtained either Neostethus or Ceratostethus in Thailand, the water was usually brackish or even very salty to the taste. At one locality Neostethus and Ceratostethus were collected together in large numbers but neither were collected in association with Phenaco- stethus. I did not find Neostethus or Ceratostethus in association with primary freshwater fishes except on one occasion (afternoon, 1 1 July 1970) when Ceratostethus was found in a khlong (strongly influenced by tides) about midway between Bangkok and Samutsa- korn, in association with Toxotes, Dermogenys, Rasbora and Eso- mus. The water was more or less fresh to the taste. A number of dead Esomus were carried by the current of the khlong, and a Rasbora or two may have been amongst them. I thought at the time that mortality of these minnows was perhaps due to in- cursion of salt water. The dead fish were carried by the out- going tide. My impression is that, in Thailand, phallostethids occur in water that is turbid and fresh; and neostethids in water that is turbid and brackish or even quite salty. I would guess that at some of the seven locahties where I collected neostethids the water was at least a third and perhaps one-half or more as salty as sea water. The only phallostethoid in which mating and egg-laying have been observed (but not in sequence) is the neostethid GulaphaUus mh-abilis Herre (VilladoUd and Manacop, 1934). It seems prob- able that all phallostethoids are oviparous. No females carrying embryos have been observed. It is unclear how much time elapses between copulation and egg-laying in G. mirabilis, or whether several egg-layings follow a single copulation. Judging from the range in size of young individuals found with adults in my collec- tions of phallostethoids, I suspect that in Thailand species some reproduction goes on throughout the year. It seems Hkely that reproductive peaks occur towards the latter part of the rainy sea- son (November and December). The distribution of phallostethoids is marginal to the rich East Indian marine and freshwater fish faunas. They do not occur in the sea, nor do they penetrate very far into fresh water where there is a continental fauna of primary freshwater fishes. The Phallostethidae penetrate further into waters inhabited by primary freshwater fishes than any other phallostethoids, but they do not 20 BREVIORA No. 374 get very far inland. Usually the canals or creeks they inhabit are strongly influenced by tides, and thus the faunal composition (with regard to fishes at any rate) is subject to considerable temporal variation, involving retreat of primary freshwater fishes and inva- sion of brackish-water forms, and vice versa. In places where the tidal changes are greatest, these invasions and counterinvasions would be a daily event, varying of course with the extent in the tides. In other places they might occur only during the strongest tides. Gulaphallus on the island of Luzon probably pass their en- tire lives in fresh water — where no primary freshwater fishes occur. DISCUSSION Selective advantage of internal fertilization in phallostethoids. One usually associates copulatory organs in teleostean fishes with viviparity, as in Poecihidae, Embiotocidae, and Brotulidae. It is a striking fact, however, that several groups of oviparous teleosts normally have internal fertilization. Among these are some Cot- tidae, some Scorpaenidae, probably some Clinidae, and probably some glandulocaudine characids. According to Nelson (1964) the evolution of internal fertihzation in the oviparous glandulo- caudine characids of tropical South America appears to be a response to well-marked wet and dry seasons. The presumed advantage of a temjxDral separation in mating and spawning is as follows: mating occurs during the dry season, when populations are crowded together in small pools, food is scarce and conditions for survival are generally unfavorable; spawning, on the other hand, occurs during the height of the rainy season, when the popu- lation is maximally dispersed (so much so that males and females may no longer be together) and conditions for survival of the young are optimum (abundance of food, well-oxygenated waters, access to areas where most predators are too large to enter). A very similar set of conditions may apply to the evolution of internal fertihzation combmed with oviparity in phallostethoids, with temporal separation of mating and spawning corresponding, respectively, with periods of low water and high water. This hypothesis should be extended to the oviparous atheriniform fishes Horaichthys setnai Kulkarni (1940) and Tomeurus gracilis, males of which have independently evolved excessively complicated copulatory organs through modification of the anal fin. Tomeurus occurs in brackish water along the northeastern coast of South 1971 PHALLOSTETHIDAE 21 America, including the mouth of the Amazon River, Horaichthys in brackish water along the Bombay and Kerala coast of India. The sperm in Tomeurus are transmitted in spermozeugmata (Niel- sen, et al., 1968: 253), as in the viviparous poecilioids (the nature of the phyletic relationship between Tomeurus and poecilioids is unclear; they probably are closely related). Horaichthys, on the other hand, is among the very few teleosts known to have a true spermatophore (Kulkarni, 1940; Nielsen, et al., 1968). (In true spermatophores the sperm are enclosed in a capsule. A spermo- zeugma is an unencapsulated group of sperm held together by a mucoid substance.) Trends in the reproductive biology of atherinijorm fishes that might be conducive to the evohition of internal fertilization. When oviparous fishes with internal fertilization exhibit a marked delay between mating and spawning, it may be presumed that fertiliza- tion actually occurs just before spawning, the sperm having been stored. Storage of sperm and delayed fertiUzation evidently char- acterize at least some viviparous fishes, such as those poecihoids (including forms with superfoetation and forms without it) in which females are capable of producing up to four or five suc- cessive broods after a single mating session. It should be noted, however, that the eggs of a number of atheriniform fishes with external fertilization normally exhibit arrested embryonic develop- ment or delayed hatching. A lengthening of the period between fertilization and hatching, particularly if it could be subjected to hormonal control, might be highly preadaptive to the evolution of internal fertilization with delayed spawning. While there is no evidence of arrested embryonic development or delayed hatching in atheriniforms with internal fertilization, it seems worthwhile to review briefly what is known about these phenomena in forms with external fertilization. Wourms (1967) found that the eggs of annual cyprinodont fishes of the subfamily Rivulinae (Cynolebias, Pterolebias, Racho- via, Nothobranchius, and Austrofundulus) are subject to develop- mental arrest at one or more stages. In Austrofundulus myersi dispersion of amoeboid blastomeres occurs early in development (stages 19-22) so that there is no embryo or aggregation of cells. This is followed by a slow reaggregation of cells and resumption of development. This arrested phase (Diapause I) is facultative, and also occurs in other annuals. Obligate developmental arrests 22 BREVIORA No. 374 in annuals were found in presomite embryos (stage 33, Diapause II) and in prehatching embryos (stage 44, Diapause III). The duration of these phases is rather variable. Wourms suggested that "the net effect of a developmental system which can undergo dia- pause of variable duration at several stages is to generate a wide distribution of eggs in any single developmental stage, and to make the transition from stage to stage a variable phenomenon. The continued survival of the population is ensured in spite of climatic cycles which are variable in periodicity and intensity" (Wourms, 1967: 341 1 ). The eggs of annual cyprinodonts have exceptionally hard chorionic membranes. Harrington (1959) reported delayed hatching in stranded eggs of marsh killifish, Fundiilus confiuentus, a member of the cyprinodontid subfamily Fundulinae, and Jones (1944) reported delayed hatching in Oryzias melastigma, a mem- ber of the cyprinodontoid family Oryziatidae (see Rosen, 1964, for a characterization of this family). In O. melastigma hatching normally occurs in 8-14 days, yet can be delayed up to six weeks; hatching can be induced by adding water of lower salinity. The only noncyprinodontoid fishes in which delayed hatching has been reported as a normal phenomenon are one or two members of the atherinid subfamily Atherinopsinae, including the grunion, Leu- resthes tenuis. In this species, "spawning begins just after turning of tide during the lower (bright moon) series of high tides, but somewhat later than turn of tide during the higher (dark moon) series of high tides. Thus the eggs are always deposited near the same point on the beach profile so that after 2 weeks the lower series of high tides washes out the eggs deposited during the higher series, and vice versa. During the interim of low tides, the eggs are above tide level in moist sand 4 inches below the surface" (Harrington, 1959: 434-435, after Thompson and Thompson, 1919; Clark, 1925). Grunion eggs presumably have a highly protective covering. It seems likely that more atheriniform fishes will be found with either delayed embryonic development or else deferred hatching (with viability of embryos sustained far beyond the usual incubation periods). The outer covering of the egg in many atheriniform fishes is highly protective. These reproductive features obviously could be advantageous to fishes such as cyprinodontoids, which generally exploit niches in ephemeral waters (Foster, 1967). Finally, it should be noted that Laale and McCallion (1968) experimentally induced developmental arrest in Brachydanio rerio, 1971 PHALLOSTETHIDAE 23 a cyprinid. Exposure to extracts from Brachydanio, frog, or chick embryos caused Brachydanio embryos at stages 17-18 to stop developing. Returned to buffered water, the embryos resumed normal development and hatched. Laale and McCallion suggested that the inhibitory factor might be a nuclease. Foster (1967) suggested that in cyprinodontoids the shift from expelling and fertilizing all of the ovulated eggs in one continuous spawning act to expelling and fertilizing them singly or in small batches greatly increased the amount of courtship behavior per fertilized egg and thereby the impact of sexual selection. The habit of expelling a few eggs at a time is also found among ather- inoids (see Breder and Rosen, 1966) and is apparently char- acteristic of phallostethoids (Villadolid and Manacop, 1934, for Gulaphallus mirabilis; Myers, 1935, for Plectrostethus palawanen- sis) , Horaichthys (Kulkarni, 1940), and Tomeurus (Myers, 1947: 8-11; Breder and Rosen, 1966: 341-343). Whereas in many atheriniforms the tendency to deposit a few eggs at a time seems to have led to the evolution of prolonged courtship, in other lines it may have led to internal fertilization. Supposing females carry a fair number of eggs but lay only a few at a time, males that can impregnate the female and fertilize all of the eggs at once have an obvious selective advantage over males that can fertilize only a few eggs at a time. Even more to the point, there may be little to prevent the sperm from getting into the oviduct and fertilizing eggs retained inside the female. Thus, while fertilization normally is external in Oryzias, instances of O. latipes females with inter- nally fertilized eggs (developing embryos) do occur (Amemiya and Murayama, 1931). Oryzias evidently is ancestral to Horaichthys. 24 BREVIORA No. 374 TO ^. O o Q- ro -a o < «/) o o -D E O O I CO hIcm cvj hIcm CO O rt ~ « G o ■- a! .22 =3 ^ .2 2 ■" X - ^ :S >. e.E u a, cd-- o o . t: ■!-• _ ^'-S'-n S ~ c LLt.22 O 5u ^ O =0 ■"" (U U C/3 ^-J rt 4—1 C E o s u -o qj c > (U 1971 PHALLOSTETHIDAE 25 LITERATURE CITED Amemiya. I., AND S. MuRAYAMA. 1931. Some remarks on the existence of developing embryos in the body of an oviparous cyprinodont, Oryzias {Aplocheiliis) latipes (Temminck et Schlegel). Proc. Imp. Acad. Japan (Tokyo), 7(4): 176-178. AuRiCH, H. 1937. Die Phallostethiden (Unterordnung Phallostethoidea Myers). Intnatl. Rev. Ges. Hydrobiol. Hydrogr., 34: 263-286. Bailey, R. J. 1936. The osteology and relationships of the phallostethoid fishes. J. Morph., 59(3): 453-483, 4 pis. Berg, L. S. 1940. Classification of fishes, both recent and fossil. Trav. Inst. Zool. Acad. Sci. URSS, 5: 87-517. Breder, C. M., and D. E. Rosen, 1966. Modes of Reproduction in Fishes. Garden City, New York, Nat. Hist. Press, xv + 941 pp. Duncker, G. 1904. Die Fische der Malayischen Halbinsel. Mitt. Naturh. Mus. Hamburg, 21: 135-207. Foster, N. R. 1967. Trends in the evolution of reproductive behavior in killifishes. Stud. Trop. Oceanogr. (Miami), 5: 549-566. Fowler, H. W. 1937. Zoological results of the third de Schauensee Siamese Expedition. Part VIII, — Fishes obtained in 1936. Prcc. Philadelphia Acad. Nat. Sci., 89: 125-308. 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, art. 4: 341-455. Harrington, R. W., Jr. 1959. Delayed hatching in stranded eggs of marsh killifish, Fimdulus confliientiis. Ecology, 40(3): 430-437. Herre, a. W. 1925. Two strange new fishes from Luzon. Philippine J. Sci., 27: 507-513, 2 pis. 1926. Four new Philippine fishes. Philippine J. Sci. 31(4): 533-543, 3 pis. 1939. The senera of Phallostethidae. Proc. Biol. Soc. Wash- ington, 52: 139-144. 1940. Manacopus, a new name for a genus of Phallosteth- idae. Copeia, 1940(2): 141. -. 1942. New and little known phallostethids, with keys to the genera and Philippine species. Stanford Ichth. Bull., 2(5): 137-156. HuBBS, C. L., AND L. C. HuBBS. 1945. Bilateral asymmetry and bilateral variation in fishes. Papers Michigan Acad. Sci. Arts Letters, 30: 229-310, 1 pi. 26 BREVIORA No. 374 Jones, S. 1944. On the occurrence of diapause in the eggs of Indian cyprinodonts. Current Science (Bangalore), 13: 107-108. KuLKARNi, C. V. 1940. On the systematic position, structural modifica- tions, bionomics and development of a remarkable new family of cyprinodont fishes from the province of Bombay. Rec. Ind. Mus. Cal- cutta, 42: 379-423. Laale, H. W., and D. J. McCallion. 1968. Reversible developmental arrest in the embryo of the zebra-fish, Brachydanio rerio. J. Exp. Zool., 167(1): 117-127, 3 pis. Myers, G. S. 1928. The systematic position of the phallostethid fishes, with diagnosis of a new genus from Siam. Amer. Mus. Novitates, No. 295, 12 pp. 1935. A new phallostethid fish from Palawan. Proc. Biol. Soc. Washington, 48: 5 6. 1937. Notes on phallostethid fishes. Proc. U. S. Nat. Mus., 84(3007): 137-143. 1947. The Amazon and its fishes . . . Part 3. Amazonian aquarium fishes. Aquar. Journ., 18(5) (May, 1947): 6-13, 32. Nelson, K. 1964. Behavior and morphology in the glandulocaudine fishes (Ostariophysi, Characidae). Univ. California Pub. Zool., 75(2): 59-152. NiELSON, J. G., A. Jespersen, and O. Munk. 1968. Spermatophores in Ophidioidea (Pisces, Percomorphi). Galathea Report, vol. 9: 239-253. Regan, C. T. 1913. Phallostetlius dunckcri, a remarkable new cyprino- dont fish from Johore. Ann. Mag. Nat. Hist., 12: 548-555. 1916. The morphology of the cyprinodont fishes of the subfamily Phallostethinae, with descriptions of a new genus and two new species. Proc. London Zool. Soc, 1916: 1-26, pis. 1-4. Rosen, D. E. 1964. The relationships and taxonomic position of the halfbeaks, killifishes, silversides and their relatives. Bull. Amer. Mus. Nat. Hist., 127, art. 5: 217-267, pis. 14 and 15. Smith, H. M. 1927. The fish Neostethus in Siam. Science (n.s.), 65: 353-355. 1945. The fresh-water fishes of Siam, or Thailand. Bull. U. S. Nat. Mus., No. 188, 622 pp. TeWinkel, L. E. 1939. The internal anatomy of two phallostethid fishes. Biol. Bull. Woods Hole, 76(1): 59-69. ViLLADOLiD, D. v., AND P. R. Manacop. 1934. The Philippine Phal- lostethidae, a description of a new species, and a report on the biology of Gulaphallns mirabilis Herre. Philippine J. Sci., 55(3): 193-220, 5 pis. 1971 PHALLOSTETHIDAE 27 WoLTERECK, R. 1942a. Stufen der Ontogenese und der Evolution von Kopulationsorganen bei Neostethiden (Percesoces, Teleostei). Intnatl. Rev. Ges. Hydrobiol. Hydrogr., 42: 253-268. 1942b. Neue Organe, durch postembryonale Umkonstruk- tion aus Fischflossen entstehend. Intnatl. Rev. Ges. Hydrobiol. Hydrogr., 42: 317-355. WouRMS, J. D. 1967. A naturally occurring vertebrate dispersion- reaggregation system subject to developmental arrest. Dissertation Abstracts, Sect. B, 27(10-11): 3410-3411. BREVIORA Miaseiiam of Comparative Zoology Cambridge, Mass. 15 June, 1971 Number 375 STRUCTURAL HABITATS OF WEST INDIAN ANOLIS LIZARDS II. PUERTO RICAN UPLANDS Thomas W. and Amy Schoener^ Abstract. This paper reports differences in structural and climatic iiabitat between the commonest Anolis species of two upland localities — Maricao and El Verde — • in Puerto Rico. Five of the seven Anolis species recorded at Maricao were studied in detail: two- — cristatelhis and stratitlus — are major components of the warmer lowland fauna, and three — krugi, evcrmanni, and gundlachi — are representative of the cooler upland fauna. A. stratiihis and evermanni tended to occupy higher and thicker perches than the other species; krugi was found on especially thin perches and often on leaves. The trunk- crown species evermanni and stratiilus, of very different size, were virtually totally syntopic over the study area, while the trunk-ground species cristatelhis and gundlachi, of very similar size, were virtually allopatric. Similarities and differences in climatic habitat between the species reflected this microgeography: stratulus and evermanni were found in fairly similar situations, though the former tended to be in opener areas, while cristatellus and evermanni were confined to open and closed habitats, respectively. At El Verde the same species were studied, but the open-area species cristatel- his and stratulus were relatively rarer. Structural and climatic habitats paralleled those at Maricao, and the trunk-crown species were again more syntopic than the trunk-ground species. In a separate analysis, it is shown that the two trunk-crown species had significantly more collecting localities in common than did the two trunk- ground species. The grass-bush species pulchellus and krugi were inter- mediate in this respect. It is hypothesized that the greater body size difference of the often syntopic trunk-crown species may be an adaptation preventing food overlap, an adaptation which the nonsyntopic trunk-ground species neither have nor need. 1 Biological Laboratories and Museum of Comparative Zoology, Harvard University, Cambridge, Mass. 02138. 2 BREVIORA No. 375 The possible evolution of climatic and size differences is discussed for the upland and lowland series. It is suggested that in the broken, drier forest often found in the lowlands, the trunk-crown species would occur in more shaded conditions on the average than would the trunk-ground species; whereas in the wetter, more continuous forests of middle and upper eleva- tions, the reverse situation would be true. Therefore, the trunk-crown species may have been predisposed for greater spatial overlap in advance of their coming together. That the lowland trunk-crown species should be smaller and less sexually dimorphic in size may be a result of greater aver- age overlap between it and the corresponding trunk-ground species: possibly the lower average vegetation height and the greater need to avoid desicca- tion makes separation by height more difficult in the opener vegetation of the lowlands. No evidence to support this latter speculation is found in this study, however. This is the second in a series of papers describing the structural habitats of some West Indian Anolis lizards. Rand (1964) has documented extensively differences between seven Puerto Rican species in their structural and chmatic habitats. Our study will also elucidate gross interspecific differences, but its primary pur- pose is to demonstrate how the size classes of each species studied are deployed over the vegetation, with respect both to others of the same species and to similarly sized lizards of other species. Its secondary purpose is to investigate the spatial relationship be- tween stratuliis and evermanni, two "trunk-crown" forms (after Rand and Williams, 1969; Williams and Rand, in preparation) of very different size and sexual dimorphism, occurring in differ- ent but overlapping climatic and geographic areas. FORMAT AND METHODS Data on the structural habitat were gathered in the way described previously (Schoener and Schoener, 1971): for each lizard seen in the area, the type of perch, perch height, and perch diameter were noted. In addition, when the sun was shining the lizard was scored as to whether it perched mostly in the sun or shade. Obser- vations during cloudy weather were lumped into a third chmatic category. As before, the two of us simultaneously canvassed an area in staggered fashion so as to minimize the possibility of miss- ing lizards. In addition, we were assisted during one afternoon at El Verde by E. E. Williams and W. P. Hall, III. 1971 PUERTO RICAN ANOLIS 3 Results are presented here in two ways: 1) standard, structural habitat tables are given for each area and lizard class of interest (Tables 3-7), and 2) the occurrence in sun, shade or clouds is given for each of the same classes (Table 2). In order to test for statistically significant differences in struc- tural and climatic habitat between the hzard classes, the multiway contingency approach used previously (Schoener, 1970; Schoener and Schoener, 1971) was again employed. This procedure deter- mines the strength of associations between the lizard classes and habitat variables while simultaneously taking into account possible associations between the habitat variables themselves. It is based on methods developed by Deming and Stephan (1940), Kullback (1959), and Bishop (1969), and is exposited for ecologists by Fienberg (1970); the appendix (page 18) outlines its particular application to our data. In addition, the appendix contains Tables 9-11, indicating the statistical significance of the comparisons we make below. Our presentation begins with a description of the study sites and their anohne inhabitants, continues with a verbal summary of differences in habitat between the lizard classes, and terminates in a comparison of our results to previous ones and in a speculation on the evolution of size in relation to space for certain Puerto Rican anoles. THE SPECIES Ten species of Anolis are known from Puerto Rico (Table 1). These can be divided roughly into three groups on the basis of their climatic and correlated geographic ranges (Rand, 1964; Wil- liams, Rivero and Thomas, 1965; Heatwole et al., 1969; Web- ster, 1969). One group comprises those species which occur modally in rela- tively humid, dark vegetation at high, cool elevations. This group consists of 1 ) evermanni — a medium-large, green, trunk-crown species, relatively sexually dimorphic in size; 2) gundlachi — a medium-large, rust-brown, relatively dimorphic trunk-ground species; 3) krugi — a small, striped yellow and brown, relatively dimorphic grass-bush species; and 4) occultus — a very small, grey-white, relatively nondimorphic twig species. Two of the three species, evermanni and krugi, are also found at scattered wet 4 BREVIORA No. 375 lowland localities, probably in remnants of the nearly destroyed wet lowland forest. The second group occurs modally in more illuminated, drier forest at the warmer low to moderate elevations. This group consists of 1 ) stratulus — a small, grey-brown, relatively non- dimorphic trunk-crown species; 2) cristatelhis — a medium-large, brown, relatively dimorphic trunk-ground species; and 3) pulchel- lus — a small, striped yellow and brown, relatively dimorphic grass-bush species. This group is also widespread in the highly disturbed wet lowlands and overlaps the first most commonly at mesic to wet intermediate altitudes, such as the Maricao locality to be described below. Additionally, George Drewry (personal communication) has found cristatelhis predominating in high- altitude, mossy forest, most of which is above the altitudinal range of gundlachi. The third group is restricted to the arid southwestern lowlands in xeric vegetation. It consists of 1 ) cooki — a medium-large, grey-brown, relatively dimorphic trunk-ground species, occurring sympatrically with cristatelhis over much of its range; and 2) ponceiisis — a small, striped yellow and brown, grass-bush species of relatively moderate sexual dimorphism. Notice that in this group there is no trunk-crown representative. A final species, the nondimorphic green giant ciivieri, appears from museum records to be commonest at middle and upland ele- vations, with pockets of abundance in the wetter lowlands such as those along the north coast. LOCALITIES The study to be reported below concentrates on the three commonest mostly upland species — evennanni, gundlachi and knigi — and two of the three commonest widespread species — cristatelhis and stratulus. All five of these species were found in close proximity to one another in the Maricao Insular Forest, a preserve in the ''monadnock" region (Pico, 1950) of the Cordillera Central. Consequently, several adjacent study sites of varying shapes were marked ofl" in the preserve (altitude = 860-900 m, slightly less than 4 km due south of the town of Maricao), and the anoles were observed therein. We distinguished four such areas: 1. Forest interior. This area was the most natural of the four studied. It contained forest of medium height and somewhat 1971 PUERTO RICAN ANOLIS 5 broken canopy, clinging to a 20-40° mountainside. The under- story in many places was quite sparse and easily traversed, but where the canopy had broken, tangled grasses and ferns obstructed passage. The commonest species seen in this area was gundlachi (Table 2). The two trunk-crown species, evermanni and stratuliis, were also common and appeared to be of nearly equal abundance. In addition, krugi, cristatelliis, and cuvieri occurred as "trace" species. We saw no occultus though they have been collected there (Williams, Rivero, and Thomas, 1965). 2. Forest with cleared understory. This area was adjacent to the more undisturbed forest and consisted of a strip about 30-40 feet wide that bounded an open, grassy parking area. Although the understory had apparently been cleared, nearly all the large trees were left standing and very little sun penetrated to the ground, which itself consisted mostly of bare mud. In this area, gundlachi was again the most abundantly seen species, followed by evermanni and stratuliis, again in nearly equal proportions. A few cristatellus were also seen at the edge of the area. 3. Open forest. This area was less elongated than Area (2) and faced it from directly across the parking area. Trees were arrayed in an open, parklike fashion; some overlaid a 2-to-3-foot herbaceous understory while others grew among short grass. Much sun penetrated to the ground, in contrast to Area (2). The most abundant species seen here was cristatellus. Again, evermanni and stratulus occurred in strikingly similar proportions. A few gundlachi were seen along the border between this area and Area (1). In addition, a few krugi inhabited the understory where it had not been chopped away. 4. Secondary road edge. This area comprised a strip about 1 0-20 feet wide bordering Area ( 1 ) . The understory graded from a lawnlike growth to dense, tangled typical forest-margin vegetation. The area faced full sunlight during the morning hours, that time of day when the sun was most likely to be shining during the period of our study. Above the understory rose trees of varying heights, with 40- to 60-foot Cecropia emerging above the rest. This area showed the most equitable distribution of apparent abun- dances for the five species. A. stratulus was commonest, but gund- lachi and evermanni were also rather frequently observed. Less abundant but far from rare were cristatellus and krugi. Wc studied the Maricao area during the period June 23-26, 6 BREVIORA No. 375 1969. Weather throughout this time was mostly sunny during the morning, generally partly cloudy during midday, and somewhat rainy during the afternoon — on one occasion in the form of a severe thunderstorm. Rainfall at the nearby but lower town of Maricao averages 111 inches per year (Pico, 1950). To further elucidate the relationships between the upland species, a second, more humid area was investigated at El Verde in the Luquillo rainforest (340-440 m). The study sites were in parts of the forest that were and are still the subjects of intensive and extensive ecological investigation, including a study of its two commonest Anolis species by Turner, Gist, and Rowland (Odum, 1965). The area is described in detail in various reports (Odum, 1965; Kline, Jordan, and Drewry, 1967, 1968). In comparison to the Maricao forest, that at El Verde was considerably taller and of more continuous but still somewhat broken canopy; the understory was consequently sparser and there were frequent boulders strewn about the forest floor. Two study sites are dis- tinguished in Table 2. The first begins near the biological station and penetrates the interior of the forest along a trail that climbs upward, eventually terminating at an observation tower. Our records were all taken adjacent to the bottom third of the trail where the grade was closest to horizontal. This area was in coffee plantation until the early 1930's and had not reached climax but was still in middle succession at the time of our study (G. E. Drewry, personal com- munication). The second study site contained roadside secondary vegetation along the forest margin, very close to the first study area. In both study areas, gimdlachi was the species most frequently seen, although it appeared more dominant in the forest interior. The next most frequently encountered species was evermanni, rela- tively more abundant along the forest edge. Three other species, stratulus, knigi, and ciivieh, were rarely encountered in the forest interior; the first two were always observed in relatively open, sunny areas. The margin had, however, a more equitable distribu- tion of species: stratulus, cristatellus and krugi did not appear uncommon. One cuvieri was also seen. All the relative abundances given in Table 2 are, of course, apparent only — they represent what we saw and are therefore biased in favor of three species {gundlachi, cristatellus, and krugi) frequently occurring at or below eye level. Observations were made June 29 -July 1, 1969. Weather dur- ing this time was, in general, considerably overcast and rainy. 1971 PUERTO RICAN ANOLIS 7 the rain occurring intermittently at all times of the day. Even though we tried to avoid sunless periods, a heavy incidence of clouds is apparent in Table 2. Rainfall in the vicinity was esti- mated from the weekly tables in Kline, Jordan, and Drewry ( 1967, 1968) as 189 inches per year from September 1964 through August 1966. RESULTS Differences between lizard classes. In the following discussion, all comparisons, unless otherwise stated, are statistically significant in the sense explained in the appendix. Each species dealt with is divided into two classes: 1) small individuals, comprising adult females, subadults, and juveniles, and 2) adult males. The great majority of the former class (85-100%) were too large to be labelled "juveniles." Maricao open areas (Tables 3, 4, 9). In order to compare spe- cies classes from relatively open areas with respect to their struc- tural and climatic habitats, observations from the "open forest" and "secondary road edge" were lumped together. Only the three commonest species of these areas — evermanni, stratulus and cris- tatellus — were treated statistically. Male evermanni perched higher than did all other groups and were significantly higher than all but male stratulus. The latter class occurred higher than small evermanni or cristatellus. Small- sized stratulus were higher than small evermanni and all cristatel- lus. Small evermanni perched higher than either male or small cristatellus. Male evermanni occurred on thicker perches than all other classes. Male stratulus were on thicker perches than all classes but male evermanni. Small evermanni were on thicker perches than small stratulus and small cristatellus. Male cristatellus occupied thicker perches than did small stratulus. Within each of the three species, males perched on both higher and thicker branches than did small-sized individuals. There were significant differences in the climatic habitats of stratulus males and the two classes of cristatellus (Table 2). The former occurred more often in the sun and less often in the shade than did either of the latter. There was also an overall tendency for male stratulus to be seen more often on cloudy days. Small stratulus too occurred more often in the sun than did either class 8 BREVIORA No. 375 of cristatellm and were seen less often in the shade and on cloudy days. Small-sized evermanni were observed more often in the shade and less often on cloudy days than were male stratulus; sunny observations were of about equal proportions in the two classes. Small evermanni were seen more frequently in the sun than male cristatellus and less frequently on cloudy days; the per- cent observations in the shade were nearly identical. Small ever- manni also occurred more frequently in sunny positions than did female cristatellus; the latter appeared more often during cloudy weather and, to a lesser extent, in the shade. No significant climatic differences were found between the classes of the same species. Maricao closed area (Tables 5, 6, 10). Observations for two study sites, the interior forest and the forest with cleared under- story, were lumped to compare the three most common species — gundlachi, evermanni and stratulus. There was no significant difference in perch height between male stratulus and male evermanni, the two uppermost classes of anoles. Male evermanni were, however, seen significantly higher than small stratulus or the classes of gundlachi. Male stratulus did not differ significantly in height from small evermanni but were observed higher than either class of gundlachi. Small evermanni were seen slightly higher than small stratulus and all gundlachi; small stratu- lus also occurred higher than did the classes of gundlachi. In perch diameter, male evermanni significantly exceeded all interspecific classes but male stratulus. Male stratulus, in turn, significantly exceeded all other interspecific classes but male gund- lachi. Small evermanni were next in perch diameter, occupying thicker perches than either class of gundlachi. Small stratulus and male gundlachi were similar in perch diameter, and the former occurred on thicker perches than small gundlachi. Once again, within the same species, the larger classes occurred on higher and thicker perches than did the smaller-sized classes. Compared to the other species, both classes of gundlachi strikingly avoided sunny places. Small gundlachi were found sig- nificantly more often in the shade or on cloudy days than any class of stratulus or evermanni. Male gundlachi were found more often in the shade than small stratulus, and more often in the shade and on cloudy days than either class of evermanni. As was the case for the more open area, small evermanni were found considerably more often in the shade and less often during cloudy 1971 PUERTO RICAN ANOLIS 9 weather than male stratidus; inside the forest, however, they were also seen slightly more often in the sun. In contrast to the opener areas, here small evermanni as well differed significantly in the above respects from small stratidus. No intraspecific comparisons were significant. El Verde (Tables 7, 11). Observations for the two areas dis- tinguished in Table 2 were combined to test structural habitat differences at El Verde. Only the two most abundant species, evermanni and gundlachi, were considered. As before, male and small evermanni were higher than either class of gundlachi. Male and small evermanni also occupied thicker perches than did small gundlachi. Male gundlachi, how- ever, were on greater-diametered perches than small evermanni and did not differ significantly from male evermanni in this respect, though the latter at Maricao occupied thicker perches. Intraspe- cifically, males once again were distributed over greater perch heights and diameters than small individuals. There were no significant climatic differences between the lizard classes — this is unsurprising because of the small sample size and preponderance of cloudy weather at El Verde. The common species at El Verde, evermanni and gundlachi, are similar in size and sexual dimorphism (Table 1) and appear to stagger their sizes in such a way as to overlap most small ever- manni and large gundlachi, with respect to both perch height and perch diameter. The situation becomes more complex where a third species, stratulus, is more abundant, as in the Maricao interior forest. The two species evermanni and gundlachi still maintain the same spatial relationship between their size classes. However, male stratulus are now inserted between male evermanni and small-sized ever- manni. They differ markedly from the former in size (Table 1) but are only slightly smaller than the latter. There is also rather strong overlap between small stratulus and small evermanni; the latter, however, average a good bit larger. Small stratulus also extensively overlap male gundlachi but are very different in size. They differ greatly in both perch height and diameter from the similarly sized small gundlachi. In the opener areas at Maricao, the situation is essentially the same for both diameter and height except that cristatellus virtually replaces gundlachi. Given that substitution, the only difference 10 BREVIORA No. 375 that appears to exist is a reversal in relative height of small stratii- liis and small evermanni. This reversal allows a more nearly per- fect staggering of sizes: now male stratulus are further from small- sized evermanni . All overlaps, of course, are further reduced by the existence of climatic differences between the niches of the species. It is important to point out that in many places in the lowlands evermanni is absent; there, presumably, cristatelliis and stratulus often stagger their sizes in much the same way as do evermanni and gundlachi. However, even if the classes of adult males over- lapped the most (see below), the two lowland species probably do not take very similar food since they are so different in size — in contrast to gundlachi and evermanni. Moreover, it is interesting to note that where stratulus and evermanni overlap extensively, as at Maricao, there is only a slightly greater tendency than at "pure" upland or lowland localities for similarly sized classes to overlap, as a result of the great difference in size between the two trunk- crown species. Because of their small sample size, classes of krugi were not tested against any other. However, if all observations for krugi from the three areas are lumped, and to these are added observa- tions made in high grass and other secondary growth near the mid- elevation town of Adjuntas, it is possible to compare large and small krugi in structural habitat. Table 8 shows that males of krugi occurred significantly higher and on thicker diameters than did small-sized lizards. However, even more striking was the much greater frequency of small krugi on leaves. This result is consistent with those for other species that occupy the leaf habitat in a major way: carolinensis on Bimini (Schoener, 1968) and grahami aquarum on Jamaica (Schoener and Schoener, 1971). No climatic difference was found between the two classes. Differences between habitat variables (Tables 9-11). In the Maricao open area the only significant height-diameter associations were those in which thinner perches tended to be high and thicker perches tended to be low. Fewer comparisons were significant in the interior forest at Maricao, but those that were showed a uni- formly opposite tendency. This could easily reflect our observa- tional bias: fewer lizards can be seen in the canopy when within the forest. The one significant interaction at El Verde was the same as those for the Maricao closed area. 1971 PUERTO RICAN ANOLIS 11 In the open area at Maricao, there was a tendency, when we combined male-evermanni perches with those of interspecific classes, for the highest perches to occur in the sun and the lowest in the shade, as would be expected from the physiognomy of the vegetation. A somewhat different result was true for the three significant interactions in the interior vegetation: high perches were more often used on cloudy days; the lowest perches were still more frequently found in the shade, however. At El Verde, the one significant interaction (for male evermanni vs. small gimd- lachi) was the same as those for the Maricao open area. There were no significant interactions at all between perch diameter and insolation in the Maricao closed area, and only one such interaction was significant for the Maricao open area. This was a tendency for the combined perches used by small stratulus and small cristatellus to be thinnest in the sun and thickest on cloudy days. At El Verde, in contrast, perches of male evermanni and male giindlachi were significantly thickest in the sun and thin- nest in the shade. The combined perches for male evermanni and male gimdlachi also were thinnest in the shade, but the thickest perches were used during cloudy weather. Both these interactions may again primarily represent the greater ease of seeing a lizard on the thin branches and twigs of the understory than on similarly sized branches belonging to the high canopy of the El Verde rain- forest. They could, however, also reflect real differences in utili- zation of the habitat by lizards: for example, during the rather brief sunny periods at El Verde, lizards may have moved onto broader surfaces to bask more effectively. CONCLUSION Although we have concentrated on differences between the size classes of the lizards at Maricao and El Verde, our results for species agree well with those reported by Rand (1964) in his pioneering study of the structural habitat of species from all areas of Puerto Rico. In that study Rand pointed out the size difference between the two trunk-crown species, stratulus and evermanni, and suggested that, because of their similarity in climatic habitat (as measured by body and air temperature, as well as shade vs. sun records), these two species should in general show a greater amount of spatial overlap than the trunk-ground or grass-bush 12 BREVIORA No. 375 species. Although in an area near the town of Maricao the few stratulus seen by Rand did not seem to overlap evermanni much in horizontal habitat, we have shown above that in our areas at the Maricao Insular Forest, the two species were quite syntopic and sometimes were of nearly identical apparent abundance (Table 2). Thus we saw 62 evermanni and 66 stratulus in the forest in- terior, 20 evermanni and 17 stratulus in the forest with cleared understory, 70 evermanni and 72 stratulus in the open forest, and 36 evermanni and 138 stratulus along the road edge. Even at our study site at El Verde, where evermanni was seen more fre- quently (67 evermanni and 7 stratulus in the interior and 31 evermanni and 11 stratulus along the forest margin), stratulus was interspersed throughout evermanni's horizontal range, although in the interior it was seen only in fairly open places, such as on sunny spots along the trail, high on emergent trees, and on stream- side boulders. G. E. Drewry (personal communication) has, how- ever, seen stratulus more commonly than evermanni in the more closed, relatively less leafy canopy of certain other areas at El Verde. The extensive syntopy of the trunk-crown species is in striking contrast to the trunk-ground species we studied. For example, in the forest interior at Maricao we saw 243 gundlachi and 1 cristatel- lus. In the open forest at Maricao we saw 175 cristatellus and 8 gundlachi. In the interior forest at El Verde we saw 243 gundlachi and 0 cristatellus. All three of these areas were relatively equi- lateral in shape. In the more elongated, marginal areas we saw more equitable proportions of the two: 33 gundlachi and 6 cris- tatellus in the forest with cleared understory at Maricao, 50 gund- lachi and 25 cristatellus along the Maricao secondary road edge, and 51 gundlachi and 14 cristatellus along the edge locality of El Verde. At the second of these areas, where the numbers are closest to identity, we noted that along a strip about ten feet wide cris- tatellus and gundlachi were perching upon the same vegetation, although under different climatic conditions: cristatellus was espe- cially common during sunny days, whereas the majority of the gundlachi individuals were seen under overcast skies. Thus, even though there is spatial overlap of cristatellus and gundlachi in marginal areas, it is largely nonsynchronous. We can test differences in the relative abundances of trunk- ground vs. trunk-crown species by constructing 2x2 contingency 1971 PUERTO RICAN ANOLIS 13 tables and running chi-square or exact tests. If this is done for the six study areas, it is seen that the relative proportions of the two trunk-ground species are significantly different from those of the two trunk-crown species in the El Verde interior forest, Maricao interior forest, Maricao road edge, and Maricao open forest at the 1 percent level and in the Maricao forest with cleared understory at the 5 percent level. Only the ecomorphs in the El Verde forest margin show nonsignificant differences in propor- tions. In summary, we can conclude that at our study sites the trunk-ground species were practically allopatric, whereas the trunk-crown species overlapped broadly. There exists a second way in which the degree of horizontal spatial overlap between the species belonging to various habitat categories may be detected. Museum localities are usually dis- tinguished horizontally, that is, as points on a map. Therefore, we have Usted all the localities at which specimens of the six common Puerto Rican species from the Museum of Comparative Zoology (Harvard) and the University of Michigan Museum have been collected. There were 91 such localities for cristateUus and gund- lachi combined, and only 7 of these (7.7% ) were held in common between the two species. In contrast, 9 of 45 locahties (20%) were held in common between evermawu and stratiilus. This pro- portion was significantly different by a chi-squared test (P < 0.05). The grass-bush species, kriigi and pulchellus, were intermediate in this regard: 8 of 59 localities (13.6%) were held in common. This proportion did not differ significantly from trunk-crown or trunk-ground species. While the comparisons involve the error that not all species inhabiting a given locality will be taken by collectors, usually because of differential abundance and/or acces- sibility, that error should counteract the pattern brought out in this analysis. That is to say, because the two trunk-crown species are seen less often and are more difficult to capture, there is less likelihood of both being collected at any given locality than both trunk-ground forms. It therefore appears clear that the relatively great body-size difference between the trunk-crown species in Puerto Rico is associated with their relatively great spatial overlap. Presumably this size difference in part helps the species to avoid resource com- petition by being associated with and facilitating differences in the diet, especially with regard to prey size. However, while the asso- ciation is presumably adaptive, the mechanism whereby it came 14 BREVIORA No. 375 about is unclear. That is, did the species differ in size to begin with and so were able to overlap spatially when they came together to a great degree, or were the size differences largely evolved in response to a predisposition for spatial overlap in the trunk-crown forms? We can crystallize the relevant problems by asking three questions: 1. (a) Why should any of the three structural habitat groups — trunk-crown, trunk-ground, and grass-bush — contains species which differ markedly in size? (b) Why should it be the trunk-crown group, and that only, which contains such species? There is thus an existence and uniqueness part to this question. 2. Given that the trunk-crown species show the greatest size differential, why should the lowland dry- area form be the smaller, and the upland wet-area form be the larger? 3. Why should the lowland trunk-crown species be much less sexually dimorphic in size than the upland species? We can gain some insight into the first question by ranking separately for upland and lowland forest the modal habitats of the three kinds of species on a darkness or humidity or coolness gradient (we assume the three to be highly correlated). In middle and upper elevation forest, such as that at El Verde and Maricao, the darkest vegetation is the understory, and the most exposed vegetation lies along margins and in the canopy. Therefore we would expect the trunk-ground species, which primarily inhabit the lowest layer of the forest, to be better adapted to dark, cool conditions than either the trunk-crown species — found largely in the canopy or somewhat lower — or the grass-bush species, found mostly along margins. However, the situation is different for the lowland forest. Much lowland forest in Puerto Rico, par- ticularly in the south, is and must have been for some time of xeric aspect. It is characteristic of such forest, as well as of much coastal woodland elsewhere, to have a relatively broken canopy, with much light penetrating to the understory and with the largest trees being often quite dispersed. The patchiness is intensified during the dry season by some but not all trees undergoing partial or complete leaf drop. The darkest places for an arboreal lizard in such veeetation are often associated with the largest trees, 1971 PUERTO RICAN ANOLIS 15 especially often within or directly under the crown. Although that fraction of the understory immediately beneath such trees will also usually be shaded, the average degree of shading of the under- story as a whole should be considerably less than in the humid forest of middle and upper elevations. The grass-bush habitat should be even less shaded. Consequently, the trunk-crown habi- tat should in the lowlands be on the average the darkest and cool- est; the trunk-ground habitat should be intermediate and the grass-bush habitat should be on the average the hottest and most illuminated, even though there are places in these latter two habi- tats as shaded as any in the crown and on the upper trunk. It also follows that within the lowland trunk-ground habitat, there should be a greater diversity of shade regimes than in the upland trunk-ground habitat. This implies that in the lowlands large adult males, which prefer thick perches, are more likely to find such perches shadier than the average small-diametered perch preferred by the females. This difference should scarcely exist for the upland species. Therefore, there should be a greater intersexual difference in climatic habitat for the lowland trunk-ground form than for the upland trunk-ground form. Matching species with structural habitat and combining the upland and lowland species into a single ranking, as would be appropriate for the many intermediate localities on the island, we should obtain from sunniest to shadiest: piilchelliis > cristatel- lus > stratulus >= evermanni = krugi > gundlachi. Notice that the trunk-ground and grass-bush species are widely separated, but that the trunk-crown species are adjacent. That is, the lowland trunk-crown species should occur in the darkest part of its habitat and the upland trunk-crown species in the sunniest part of its habitat. Consequently, there should be a great deal of spatial overlap in intermediate areas on the basis of climatic habi- tat alone. This means that much of the size difference could well have evolved as a necessary adjunct to climatic preferenda of the trunk-crown species, preferenda which have developed inde- pendently of one another as adaptations to the prevalent vegeta- tional structure. No such size difference need have evolved for the trunk-ground or grass-bush species, farther apart on the climatical spectrum. The temperature data of Rand (1964, fig. 4), particularly those for the air, fit for the most part the inequality stated above. Thus, 16 BREVIORA No. 375 among the primarily lowland species, pulchellus occurs at the warmest temperatures, cristatellus at cooler temperatures, (with great variance, perhaps because of intersexual differences) and stratulus the coolest. Almost identical to stratulus is evermanni, whereas gundlachi is about the same as evermanni and therefore warmer than predicted, possibly because the highest evermanni could not be sampled. Values for gundlachi fall below the bulk of the cristatellus observations. Air temperatures for krugi are somewhat higher than those for evermanni but markedly below the majority of the pulchellus observations. Rand (1964) also provides information on species within five locahties as to their degree of occurrence in shade. This again depicts stratulus and evermanni as differing little, while the other groups differ a good deal. Also shown is that cristatellus and stratulus can be fairly similar in their shade occurrences, though in all cases stratulus occupied the deeper shade. A. gundlachi and evermanni are, however, very different, the former being, of course, more often in shady positions. In the two localities where both krugi and pulchellus were found, they showed practically no over- lap in their shade occurrences. The climatic data we have gathered at Maricao, which are of a different sort than those presented by Rand, also support the greater similarity of the trunk-crown species in their climatic habitats than the trunk-ground forms. There is not even any consistent tendency for one or the other of stratulus or evermanni to be found more often in the sun among the Maricao localities, although evermanni, if any, appears commoner there. This may in part, of course, be an artifact of observation; at El Verde, as we judge from the few stratulus data, the situation appeared to be reversed. The two trunk-ground species cannot be compared in most places because one or the other is nearly absent, but along the secondary road edge, as mentioned, gundlachi was much more likely to be seen during cloudy weather. At El Verde, despite the preponderance of overcast skies, cristatellus were seen several times more frequently in the sun than gundlachi. A second aspect of the above scheme that can be checked from our data is the climatic relation between trunk-ground and trunk- crown forms. Although evermanni and gundlachi showed striking differences in the expected direction, stratulus and cristatellus did not. In fact, stratulus were seen significantly more often in the 1971 PUERTO RICAN ANOLIS 17 sun in the open area at Maricao (see above). Possibly this was an artifact of observation, but it is more likely that in the shade cristaielliis perches averaged warmer than did those of stratidus because of the more exposed nature of the vegetation on which it perched. But Heatwole (1968; Heatwole et ai, 1969) showed that individuals of stratidus and cristatellus in "an open park-like situation" at Rio Piedras did not differ significantly in their air or substrate temperatures. However, Heatwole et al. (1969) have found that even under apparently identical mean environ- mental temperatures, the body temperatures of stratulus average lower than those of cristatellus. Furthermore, cristatellus shows greater resistance to high temperatures and does not tolerate low ones as well as stratulus. Heatwole et al. conclude both from these physiological prop- erties and Rand's data that "although the two species have similar ecologies and distributions, in some cases coexisting under appar- ently identical conditions, there is a tendency for A. cristatellus to utilize the warmer (less shaded) habitats than A. stratulus." Their conclusion agrees well with the scheme we have outlined above. The second question posed above may be related to the relative degree of overlap between trunk-ground and trunk-crown forms in the two areas. If we could show that the lowland species are more likely to overlap spatially than the upland forms, then we could argue that the smaller size of stratulus may have evolved in part as a response to that overlap and thus served as a preadapta- tion for any further displacement that might have occurred between the trunk-crown species. Rand's structural habitat grids show practically no differences in percent overlap (62.5% vs. 61%) between the two species pairs. However, his data were from sev- eral localities and all size classes combined. Even if only adult males are considered and our data for particular locahties used, results are inconsistent: percent overlap for evermanni and gund- lachi is 35 percent in the forest interior at Maricao and 72 per- cent at El Verde, whereas that for stratulus and cristatellus is 30 percent along the Maricao road and 50 percent in the Maricao open forest. It is possible that these results are artifacts of the greater difficulty of seeing a trunk-crown species in the canopy when within a continuous forest than when along a forest margin or in a very open area. That error would tend to minimize differ- ences between evermanni and gundlachi and is almost certainly 18 BREVIORA No. 375 in part responsible for the high overlap value at El Verde. One might argue, a priori, that because of the greater average height of rainforest such as that at El Verde, overlap between two spe- cies segregated mainly by height should be less than in drier forest. Furthermore, even though the species are physiologically adapted to different thermal environments, it is possible that in the dry lowlands the greater danger of desiccation forces cristatelliis and stratuliis together in shaded situations during a large portion of the day; in the uplands, however, no such problem need arise, and the more exposed species, in this case evermarmi, can occur in sunny or open places most of the day. This is perhaps the rea- son why less difference was found between the climatic habitats of cristatellus and stratuliis than between gnndlaclii and evermanni in our study. Any overlap between the first pair should be espe- cially severe for male cristatellus because they more frequently occupy the relatively large trees which stratulus inhabit. If the explanation above be correct, that is, had stratulus evolved small size in part to alleviate competition with male cristatellus, then the reduced sexual dimorphism of stratulus would automati- cally follow, since it is the male that would be displaced. ACKNOWLEDGMENTS We thank G. E. Drewry, A. S. Rand, T. P. Webster, and E. E. Williams for critical comments on the manuscript and S. D. Fien- berg for statistical advice. We also thank G. E. Drewry for his hospitality at the Puerto Rican Nuclear Center field station at El Verde and K. Horton for typing the tables photographed directly for this article. Rand, Webster and Williams were all in the field with us during some part of this study, and their previous experi- ence with the island greatly facilitated our work. In addition, Williams and W. P. Hall III contributed some observations to our data from El Verde. Research was partly supported by NFS grants GB 6944 and B 01 980 IX to E. E. Williams. STATISTICAL APPENDIX This appendix briefly reiterates the description of the statistical treatment of the structural and climatic habitat data given in detail elsewhere (Schoener and Schoener, 1971). Four variables — lizard class, climatic category, perch height and perch diameter — were used to set up 4-way contingency tables. For the lizard and 1971 PUERTO RICAN ANOLIS 19 climatic variables, two (the classes being compared) and three (sun, shade, and clouds) categories (levels), respectively, were used. For perch height and diameter, two levels each were chosen by separately determining the point of maximum difference in cumulative frequency between the distributions of the habitat variable for the two lizard classes being compared. Data were broken at this point, all observations less than or equal to that number being cast into one category and all greater than that number being cast into the other. Table 12 lists critical values for height and diameter groupings. As before, an iterative pro- cedure (Deming and Stephan, 1940; Bishop, 1969; Fienberg, 1970) was used to fit the data to models containing all six 2-way interactions. One by one, interactions were dropped if differences between models were not significant at the 5 percent level accord- ing to difference in the log-likelihood ratio chi-square (Kullback, 1959; Ku and Kullback, 1968). Results are summarized in Tables 9-1 1 . The following key applies to the numbers in the body of the table: 1 — the interaction was significant every time it was tested in the removal procedure; 2 — the interaction was significant at least at the termination of the procedure; 3 — the interaction was significant when removed from the most inclusive model (with six interactions) but not at termination; 4 — the interaction was significant sometime during the pro- cedure but not at the beginning or end; 0 — the interaction was never significant. As can be seen, nearly all interactions could be labelled "0" or "l." There was no set of models for a particular 4-way table which never gave a x" value denoting a satisfactory fit of the model at the 5 percent level, regardless of what interactions were re- moved. Therefore we did not test for 3-way interactions. There were only three tables with zero margins. These could be, and therefore were, handled in the way given by Fienberg (1970). LITERATURE CITED Bishop, Y. M. M. 1969. Full contingency tables, logits, and split con- tingency tables. Biometrics, 25: 383-400. Deming, W. E., and F. F. Stephan. 1940. On a least squares adjustment of a sampled frequency table when the expected marginal totals are known. Ann. Math. Stat., 11: 427-444. 20 BREVIORA No. 375 FiENBERG, S. E. 1970. The analysis of multidimensional contingency tables. Ecology, 51: 419-433. Heatwole, H. 1968. Relationship of escape behavior and camouflage in anoline lizards. Copeia, 1968: 109-113. , T. Lin, E. Villalou, A. Muniz, and A. Matta. 1969. Some aspects of the thermal ecology of Puerto Rican anoline lizards. J. Herpetology, 3: 65-78. Kline, J. R., C. F. Jordan, and G. E. Drewry. 1967, 1968. The rain forest project annual reports. Puerto Rico Nuclear Center, Puerto Rico. Ku, H. H., and S. Kullback. 1968. Interactions in multi-dimensional contingency tables: an information theoretic approach. J. Res. Natl. Bur. Standards - Mathematical Sciences, 728: 159-199. Kullback, S. 1959. Information Theory and Statistics. Dover Publica- tions, New York. Odum, H. T. 1965. The rain forest project annual report FY-65. Puerto Rico Nuclear Center, Puerto Rico. Pico, R. 1950. The geographic regions of Puerto Rico. Univ. Puerto Rico Press. Rio Piedras, Puerto Rico. 256 pp. Rand, A. S. 1964. Ecological distribution in anoline lizards of Puerto Rico. Ecology, 45: 745-752. , and E. E. Williams, 1969. Anoles of La Palma: aspects of their ecological relationships. Breviora, No. 327: 1-19. Schoener, T. W. 1968. The Anolis lizards of Bimini: resource partition- ing in a complex fauna. Ecology, 49: 704-726. . 1969. Size patterns in West Indian Anolis lizards. I. Size and species diversity. Syst. Zool., 18: 386-401. 1970. Nonsynchronous spatial overlap of lizards in patchy habitats. Ecology, 51: 408-418. , AND A. Schoener. 1971. Structural habitats of West Indian Anolis lizards. I. Lowland Jamaica. Breviora, No. 368: 1-53. Webster, T. P. 1969. Ecological observations on Anolis occult us Wil- liams and Rivero (Sauria, Iguanidae). Breviora, No. 312: 1-5. Williams. E. E.. and A. S. Rand. The structure of diversity in Anolis liz- ards, in prep. , J. A. RivERO, AND R. Thomas. 1965. A new anole (Sauria, Iguanidae) from Puerto Rico. Breviora, No. 231: 1-18. 1971 PUERTO RICAN ANOLIS 21 TABLES 22 BREVIORA No. 375 01 (1) •H O 01 o -H Pi o +) :3 1-1 o 01 +1 fH +> -H (0 a nJ 01 Q) N -H 01 (U Eh t3 (C 01 01 (1) rH < n ■r-i w 0) -p -p +J M (1) Ul OJ s o a >i >1 (1) rH 4-1 0) 01 iH _ O TJ -u e T3 01 01 C fO S-l 0) 0 >-l 0 3 0) r^ cr> 00 • • o •^ r^ U3 01 :3 4-1 (0 M 4-> 01 01 S o 01 o u I X u (TV -H o o u CN U3 in in 0) 4-1 to 4-> 0) •H >-l O 0) g o 01 Tl o s-l I Eh •H Cn 3 in 01 3 Xi I CO 01 n) en ■H Eh 4J •H O O til -—V o o\ rt VO s o o\ "* o '^ ^1' t- » ^_l o VI . , •T3 '^ i^ 2 P. c/5 _C to 1- ^"^ iH •o c o OC v: C -a a c •o w OJ c« o rt r^ CQ C/1 97 PUF.RTO RICAN ANOI.IS 23 Table 2. cliiratic and geographic occurrence of Puerto Rican anoles. Percent Occurrence Lizard Class Sun Shade Clouds Maricao, forest interior evermanni adult male everraanni small gundlachi adult male gundlachi small stratulus adult male stratulus small cristatellus adult male crista tellus small krugi adult male krugi small cuvieri large aricao, forest, clea evermanni adult male evermanni small gundlachi adult male gundlachi small stratulus adult male stratulus small cristatellus adult male cristatellus small krugi adult male I-.rugi small cuvieri large 53.8 11.5 50.0 28.6 0 0 33.3 9.1 66.7 16.7 57.1 25.0 16.0 0 18.2 33.3 Sun or Sample Shade Size 34.6 44 .4 22 .2 27.8 17 1 34 .3 45.7 15 6 32 4 48.0 31 8 4 5 63.6 31 8 13. 6 47.7 0 0 100.0 50.0 83.3 14.3 75.0 84.0 66.7 72.7 5.6 2.9 4.0 0 6.8 0 26 36 70 173 22 4 4 1 0 2 0 1 14 8 25 6 11 0 6 0 0 0 Percent Compo- sition 6.9 9.6 18.7 46.1 5.9 11.7 0.3 0.5 0.3 7.9 18.4 10.5 32.9 7.9 14.5 7.9 24 BREVIORA No. 375 ass Percent Occurrence Sample Size Lizard CI Sun Shade Clouds Sun or Shade Percent Compo- sition Maricao, open forest evermanni adult male 33.3 28.6 33.3 4.8 21 6.2 evermanni small 44.9 28.6 20.4 6.1 49 14.5 gundlachi adult male 0 gundlachi small 12.5 50.0 25.0 12.5 8 2.4 stratulus adult male 40.5 8.1 45.9 5.4 37 11.0 stratulus small 22.9 37.1 37.1 2.9 35 10.4 crista tellus adult male 22.0 26.0 50.0 2.0 50 14.8 cristatellus small 22.4 32.0 41.6 4.0 125 37.1 krugi adult male 28.6 42.9 28.6 0 7 2.1 krugi small 0 60.0 40.0 0 5 1.5 cuvieri large 0 Maricao, secondary road edge evermanni adult male 78.3 0 21.7 0 23 8.3 evermanni small 69.2 7.7 23.1 0 13 4.7 gundlachi adult male 11.1 27.8 61.1 0 18 6.5 gundlachi small 15.6 3.1 78.1 3.1 32 11.5 stratulus adult male 55.6 7.4 33.3 3.7 54 19.4 stratulus small 59.5 8.3 31.0 1.2 84 30.2 cristatellus adult male 33.3 33.3 16.7 16.7 6 2.2 cristatellus small 36.8 21.1 31.6 10.5 19 6.8 krugi adult male 0 25.0 50.0 25.0 4 1.4 krugi small 4.0 16.0 76.0 4.0 25 9.0 cuvieri large 0 El Verde, interior evermanni adult male 5.6 33.3 61.0 0 18 5.6 evermanni small 3.9 27.6 68.6 0 51 15.7 gundlachi adult male 7.1 32.5 59.8 0.6 169 52.2 gundlachi small 5.4 41.9 52.7 0 74 22.8 1971 PUERTO RICAN ANOLIS 25 stratulus adult male 100.0 0 0 0 1 0.3 stratulus small 33.3 33.3 33.3 0 6 1.9 cristatellus adult male 0 cristatellus small 0 krugi adult male 50.0 0 50.0 0 2 0.6 krugi small 0 0 100.0 0 4 1.2 cuvieri small 0 0 100.0 0 1 0.3 El Verde, edge evermanni adult male 46.2 23.1 30.8 0 13 9.5 evermanni small 27.8 16.7 55.6 0 18 13.1 gundlachi adult male 8.3 36.1 52.8 2.8 36 26.3 gundlachi small 13.3 26.7 60.0 0 15 10.9 stratulus adult male 67.7 0 33.3 0 3 2.2 stratulus small 37.5 12.5 37.5 0 8 5.8 cristatellus adult male 50.0 0 50.0 0 6 4.4 cristatellus small 25.0 25.0 50.0 0 8 5.8 krugi adult male 16.7 16.7 66.7 0 6 4.4 krugi small 21.7 26.1 52.2 0 23 16.8 cuvieri small 100.0 0 0 0 1 0.7 26 BREVIORA No. 375 Table 3. Maricao secondary road edge. Percent observations in various structural habitat categories. H = >20'; G = ground; R = rocks; N = sample size. ^^^^iameter Ht. ^-^in.) (feet) ^-v,^^^ >5 5-2 1/2 2 1/4-7/8 7/8-1/8 le aves Total N = 54 male stratulus H = 11 G = 0 R = • 0 10.5-20 4 17 4 0 0 25 5-10 9 17 6 0 0 32 3-4 3/4 9 4 4 0 0 17 <3 9 4 4 0 0 17 Total 31 42 18 0 0 N = 84 small stratulus H = 2 G = 0 R = = 0 10.5-20 4 8 7 1 0 20 5-10 4 13 13 6 0 36 3-4 3/4 2 6 5 5 0 18 <3 10 6 6 1 2 23 Total 20 33 31 13 2 N = 6 male cristatellus H = 0 G = 0 R = 0 10.5-20 0 0 0 0 0 0 5-10 0 0 0 0 0 0 3-4 3/4 17 17 0 0 0 34 <3 33 17 17 0 0 77 Total 50 34 17 0 0 N = 19 small cristatellus H = 0 G = 11 R = 0 10.5-20 0 0 0 0 0 0 5-10 0 0 0 0 0 0 3-4 3/4 5 0 0 0 0 5 <3 16 26 26 16 0 84 Total 21 26 26 16 97: PUERTO RICAN ANOLIS 27 N = 23 male everinanni H = 4 G = 0 R = 0 10.5-20 22 17 0 0 0 39 5-10 9 13 0 0 4 26 3-4 3/4 17 0 0 0 0 17 <3 13 0 0 0 0 13 Total 61 30 0 0 0 N = 13 small evermanni H = 0 G = 0 R = 0 10.5-20 0 8 0 0 0 8 5-10 0 15 0 0 8 23 3-4 3/4 8 0 0 0 0 8 <3 46 15 0 0 0 61 Total 54 38 0 0 8 Table 4. Maricao open forest. Percent observations in various structural habitat categories. H = >20'; G = ground; R = rocks; N = sample size. JDiameter Ht. ^\^(in.) (feet) >5 5-2 1/2 2 1/4-7/8 7/8-1/8 leaves Total N = 37 10.5-20 5-10 3-4 3/4 <3 Total male stratulus H = 0 G = = 0 R = = 0 0 22 0 0 0 8 16 11 5 0 3 3 0 5 0 5 19 3 0 0 22 40 11 27 14 10 N = 35 10.5-20 5-10 3-4 3/4 <3 Total small stratulus H = 0 G = = 0 R = = 0 0 3 0 0 0 3 6 19 13 0 9 0 11 9 0 14 9 6 0 0 3 41 29 29 26 18 36 22 28 BREVIORA No. 375 N = 50 10.5-20 5-10 3-4 3/4 <3 Total •M = 12 5 10.5-20 5-10 3-4 3/4 <3 Total N = 21 10.5-20 5-10 3-4 3/4 <3 Total male cristatellus H = = 0 G = 6 R = = 0 0 6 0 0 0 4 4 10 2 0 8 0 12 2 0 .20 10 7 7 2 32 20 29 11 small crista te llus H = 0 G = 12 R = 1 0 2 0 0 0 0 2 5 2 0 2 2 2 3 1 11 23 16 15 2 13 29 23 20 male evermanni H = 0 G = = 10 R = 0 5 19 7 2 0 5 5 14 5 0 10 5 0 0 0 14 0 0 0 0 6 20 22 46 2 9 9 67 33 29 15 14 34 29 21 N = 49 10.5-20 5-10 3-4 3/4 <3 Total small evermanni H = 6 G = 2 R = 0 2 6 2 4 0 4 6 12 10. 0 0 2 4 0 0 16 10 8 2 2 14 32 6 38 22 24 26 16 1971 PUERTO RICAN ANOLIS 29 Table 5. Maricao forest with cleared understory. Percent observations in various structural habitat categories. H = >20' G = ground; R = rocks; N = sample size. ^iameter Ht. ^^-.^^in.) (feet) >5 5-2 1/2 2 1/4-7/8 7/8-1/8 leaves Total N = 6 10.5-20 5-10 3-4 3/4 <3 Total male stratulus H = 50 G = 0 R = 0 0 17 0 0 0 17 17 0 0 0 0 0 0 0 0 0 0 0 0 0 17 34 0 0 N = 11 10.5-20 5-10 3-4 3/4 <3 Total small stratulus H = 0 G = 0 R = 0 0 9 0 0 0 9 18 9 9 0 0 0 0 0 0 18 18 9 0 0 9 45 0 45 27 45 N = 8 10.5-20 5-10 3-4 3/4 <3 Total male gundlac hi H = 0 G = = 0 R = 0 0 0 0 0 0 0 12 12 25 0 0 12 0 12 0 0 0 19 6 0 0 49 24 25 24 31 43 N = 25 10.5-20 5-10 3-4 3/4 <3 Total small gundlachi H = 0 G = = 0 R = = 0 0 0 0 0 0 4 4 4 4 0 0 16 8 4 0 4 12 28 12 0 0 16 28 56 32 40 20 30 BREVIORA No . 375 N = 6 male evermanni H = 0 G = 0 R = 0 10.5-20 17 0 0 0 17 34 5-10 0 33 17 0 0 50 3-4 3/4 0 0 0 0 0 0 <3 0 17 0 0 0 17 Total 17 50 17 0 17 N = 14 small evermanni H = 0 G = 7 R = 0 10.5-20 0 0 0 0 0 0 5-10 7 14 14 0 0 35 3-4 3/4 0 0 0 0 0 0 <3 36 21 0 0 0 57 Total 43 35 14 0 0 Table 6. Maricao forest interior, various structural habitat categories, rocks; N = sample size. Percent observations in H = >20'; G = ground; R = ^\Diameter Ht. ^\(in.) (feet) ^v^ >5 5-2 1/2 2 1/4-7/8 7/8-1/8 leaves Total N = 22 male stratulus H = 5 G = 0 R = 0 10.5-20 0 32 5 0 0 37 5-10 5 14 7 11 5 42 3-4 3/4 0 0 9 0 0 9 <3 5 5 0 0 0 10 Total 10 51 21 11 5 N = 44 small stratulus H = 2 G = 2 R = 0 10.5-20 2 2 2 2 2 10 5-10 5 5 16 11 7 44 3-4 3/4 0 0 2 2 0 4 <3 2 9 20 2 2 35 Total 16 40 17 11 1971 PUERTO RICAN ANOLIS 31 N = 70 10.5-20 5-10 3-4 3/4 <3 Total N = 173 10.5-20 5-10 3-4 3/4 <3 Total N = 26 10.5-20 5-10 3-4 3/4 <3 Total N = 36 10.5-20 5-10 3-4 3/4 <3 Total male gundlachi H = 0 G = 1 R = 0 0 0 0 0 0 4 4 6 4 0 1 11 13 7 1 9 11 11 13 1 14 26 small gundlachi 0 0 1 1 0 1 3 9 30 H = 0 0 0 5 23 G = 24 8 0 3 4 29 R = 1 0 0 1 11 35 24 20 4 11 28 36 12 male evermanni H = 8 G = = 0 R = = 0 15 4 8 0 0 12 12 12 8 0 0 8 0 0 8 8 0 0 0 0 0 18 33 45 0 5 11 75 27 44 16 small evermanni H = 3 G = 3 R = 0 0 8 11 0 0 0 6 17 14 3 0 3 6 3 3 0 6 8 8 0 19 40 15 24 23 42 25 32 BREVIORA No. 375 Table 7. El Verde. Percent observations in various structural habitat categories. H = >20'; G = ground; R = rocks; N = sample size. ^^"'----.Diaineter Ht.^^^in.) (feet) ^^..^^ >5 5-2 1/2 2 1/4-7/0 7/8-1/8 leaves Total N = 205 male gundlachi H = 0 G = 2 R = 1 10.5-20 0 3 0 0 0 3 5-10 15 10 12 7 0 44 3-4 3/4 8 12 9 3 0 32 <3 9 4 2 1 0 16 Total 32 29 23 11 0 N = 89 small gundlachi H = 0 G = 4 R = 3 10.5-20 0 0 0 0 0 0 5-10 0 8 8 6 1 23 3-4 3/4 4 5 9 10 0 28 <3 6 12 9 12 0 39 Total 10 25 26 28 1 N = 31 male evermcinni H = 0 G = = 3 R = 0 10.5-20 6 0 6 0 0 12 5-10 16 3 10 6 3 38 3-4 3/4 13 3 13 0 0 29 <3 10 3 3 0 0 16 Total 45 9 32 6 3 N = 69 small evermanni H = 1 G = 6 R = 3 10.5-20 6 1 0 1 1 9 5-10 1 4 12 9 6 32 3-4 3/4 4 3 3 9 0 19 <3 13 9 2 4 0 28 Total 24 17 17 23 7 1971 PUERTO RICAN ANOLIS 33 Table 8. A. krugi, all localities combined. Percent observa- tions in various structural habitat categories. H = >20'; G = ground; R = rocks; N = sample size. ^iameter Ht. ^\(in.) (feet) >5 5-2 1/2 2 1/4-7/8 7/8-1/8 leaves Total N = 33 10.5-20 5-10 3-4 3/4 <3 Total N = 83 10.5-20 5-10 3-4 3/4 <3 Total male krugi 0 3 0 0 0 0 0 0 H = 0 G 0 6 12 3 0 R = 0 3 30 6 12 21 51 0 15 6 3 24 small krugi H = 0 G = 4 R = 0 0 0 0 0 0 0 0 0 1 5 0 0 0 7 16 0 0 2 11 54 6 51 24 18 0 6 23 67 19 75 34 BREVIORA No. 375 w G O U) M e o u (fl (1) to C 0) o 0 «3 U ■M m s u o O o c (0 o •H c en ■H n3 o •H tn -p en 0) en 0) > 0) ITS tn 0 i-\ rH 0) rH +J iH fl (0 +J e tn U) ■H i-i u tn :3 iH tH (U 0) iH +J 03 (0 g -M tn •H M U H H H * * * O +J CO ^ u > Q) CO U (U 03 e (0 03 oj e en 03 e 03 e en * Q Q U m u u pq U U CQ (U CO -H pq PQ CQ tn tn 3 3 H tn rH rH •rl tn 3 rH Q) •rl C 3 H 0) 4J C c rH 3 p 03 G OJ 3 -P 03 4-> 08 e JJ 03 4J tn e ^ 03 M to •H M tu sh +) ■r) >H 4J tn U tj > (U to u 0) (U 0) e 03 (1) rH 0) rH rH 03 rH 03 OJ g 5 6 6 tn E CO S to 1971 PUERTO RICAN ANOLIS 35 t( Cn Cm fc, H H H fM O O O O O O M w 3 3 -H 03 r-l rH •H w 3 iH (D •H c 3 rH Q) -p c c <-i 3 -p (0 c nj P -P nJ 4J (0 e -P (0 •p 01 £ H (0 M w •H V-i U -P •H >-< 0) > -P in ^ o > 0) (0 0 (U 0) H 0) rH (0 to g (0 6 01 »3 CO •OT3 3 3 ^ _o "o "o T3 c C "^ ca 4-* CO v < « -Q C C lo C oj 3 D. CQ--: (u oj ■;: 3 J= t« C/5 ■2 C CO ra 3 u. X c/5 OJ X o c 5 c/5 Ca ^ OX C u O S OJ IJ Q. c 36 BREVIORA No. 375 0) u u o •H u 0) -p c -H o n3 O •H u M o m (U u c (0 u •H m •H c en •H w rH o •rH 4-1 m •H 4J (0 CO 0) > •H C c Q) 0) rH e IB )-i e Q) > 0) •H 3 •rH Xl u rH H e G U3 H rH CO W 3 rH (U 3 H H-> fl m g ^ -p CO He ■K rH tS) OJ m •H •H to ^ ■H x; •H C 3 ^ u c c rH 3 o fl c to 3 +J (0 H OJ •P (0 rH T) g p <0 U TJ C >H (U ^ -P C 3 (U > +> tn 3 Di > 0) (0 tJ> 0 (0 E (0 6 (0 6 g (U W H Q Q U U U U O O CN rH * * u u H H O iH < < tn -H w 3 •H a:: •H 3 rH X! U C rH 3 O nJ C 3 -P m H nj +> (fl rH -0 g (0 M •a C u U -P c 3 Q) •P W 3 &> > w Cn 0) (0 g H g en g -H c: g u > (0 E C/3 1971 PUERTO RICAN ANOLIS 37 5 5 (N O O O fN O O O O O •3 o o o o o O o o o O O o o O o o o o o tn •H •rH w 3 •H x: •H c 3 iH ^ u C c rH 3 u (0 C fd 3 ■P (fl H m H -P (fl rH 13 g M (0 M T) c M Q) »H -M G 3 0) > ■P m 3 Di > QJ W D^ 0) 0) iH dJ rH (1) H iH (0 iH nj >H ftj (0 g (0 e 2 s g in g ui 6 w 3 N O T! (/3 " C3 _ -C •73 ^ II -" ,, 2i ll = . ^ S.„' -^1 Wj II -S U4 c (/I a -r-^^ x: C o 3 a'^ c c *-• « « iH ^ «Si ry) "O r^ !LI C3 I. ■^■s^ -_> " c ^.s- L- II (U o •- '^ " o . Eg ° 11 ■& C3 Q E 38 BREVIORA No. 375 Table 11. Statistical significance for El Verde." ^^^^Group vs. Gro^i^^i^ht vs. ^""~\^ male small male small Diameter ^""^ evermanni evermanni gundlachi gundlachi male evermanni 1 2 1 small evermanni 1* 1* 1 male gundlachi 0 1 1 small gundlachi 1* 1* 1* ^"^^^Group vs. Hei^h^-^°1^^^°'^ vs. ^"^-^^ diameter ^~^--^^^ male evermanni 0 0 0 small evermanni 0 0 0 male gundlachi 0 0 0 small gundlachi 1(A) 0 0 ^""--.^eight vs. ^^^^insolaticn Diameter^^^^^ vs. ^^-^^^ insolation ^~^^ -^ male evermanni 0 0 1(F) small evermanni 0 0 0 male gundlachi 1(K) 0 0 small gundlachi 2(L) 0 0 " * = species at top has larger value; A = small diameters at low perches; F = highest in sun and lowest in shade; K =: thickest in sun and thinnest in shade; L r= thinnest in shade and thickest in clouds: for interpretation of numbers, see "Statistical appendix." 1971 PUERTO RICAN ANOLIS 39 •H G (0 g > •rl a c Q) (8 rH g (0 !-i > CO 3 iH -H •-i 0) rH +J (d (0 e -P CQ in •H ^1 u 0) Q) rH 4J (0 (0 S -P cn •H ^^ U CO rH iH rH 13 g nj en )-i -P (0 rH CD 3 rH 4-> (0 (0 e u -p w CD to -P Q) 0) Xi (3 c •H -H (N •^ 'S' -^ (NJ \ \ \ \ \ rH n iH iH iH rH CN IX) fN CO CM 'S' cr\ CTi ro ■^ ^ '^ "^ \ \ \ rO rH rH CM CN] (N in "* -^ n in CN CN OJ •>* CM W (!> 3 3 f-i M H iH •H to 3 M (U •H c 3 rH Q) +J c c H 3 -P to a to :3 4J td +J to g 4J to +J to ^H to iH w •H ^ QJ M 4J •H ^ 0) > •P V) ^ O > tu w U Q) fl) rH (U H rH to rH to 2 g to e g Wl g W iH to to to o to o -H to -H r-« r-J G 1 — 1 to to g w QJ > (U •H G G 0) to to fi g tu > •H H U iH to to r-\ g -73 -a p (U D' w o •H 45 o OJ to rH rH 10 tJl e C 3 Di tn 3 .H rH rH 3 to -P m . •^ ■^ CA \ \ \ ro ro rH vxi o rH 1J3 «;y CN •<3' H 1^ (» VO 1^ CXD H '^ ^ ^ \ \. \ rH n rH ^ "^ (N CN \ \o in CO CM CM CM CO (^ CO in m 03 ■H -H M 3 •H x: -H I^ 3 H ;G u G G rH 3 u to G to 3 +J (0 H to g -!J (0 rH TJ g M to ^ -13 c U 0) U +J C 3 d) > 4-> tn 3 CT > 0) m tTi OJ rH (U iH Q) rH to H tO rH (0 g to g 10 g 0) g tn g t/3 •H G rH C rH (0 10 g g IH W Q) > OJ •H G G (U to H g 0) to ^ -a g 0 M > 0) 0) > iH W •H x; rH U rH 03 lO H g T) M G 3 tji •H x: U 0) 10 .H rH nj -a e G 3 OT / 4J 1 0) 1 +) (U 1 x: MH 1 01 — 1 •H 1 OJ /u ^ K / (u m +J lU 3 CP > 0) tJi 0) rH 0) iH 10 rH to g 2 g in g (0 BREVIORA Mmseiuinii of Coiniiparative Zoology Cambridge, Mass. 15 June, 1971 Number 376 Podocnemis venezuelensis, a new fossil pelomedusid (Tes- tudines, Pleurodira) from the Pliocene of Venezuela and a review of the history of Podocnemis in South America Roger Conant Wood^ and Maria Lourdes Diaz de Gamero- Abstract. Podocnemis venezuelensis, a new species of pelomedusid turtle from the mid-Pliocene (Huayquerian) of Venezuela is described. Present evidence is insufficient for determining whether it was a marine or a fresh water form. Previously described South American fossil species that have been attributed to Podocnemis are discussed. Of these, only two — P. bassleri and P. elegans — are surely referable to this genus. It is concluded that the fossil record of Podocnemis is not yet adequate for reconstructing its evolutionary history on this continent. Resumen. Se describe una nueva especie de tortuga pelomedusida del Plioceno medio (Huayqueriense) de Venezuela, Podocnemis venezu- elensis. Las presentes pruebas son insuficientes para determinar si fue de aguas marinas o de agua dulce. Se discuten las especies suramericanas fosiles que han sido atribuidas a Podocnemis. De estas, solo dos — P. bass- leri y P. elegans — son valederas. Se concluye que el registro fosil de este genero no es aiin adecuado para la reconstrucion de su historia evo- lutiva en este continente. INTRODUCTION Only one species of fossil turtle, Podocnemis geologorum (Simpson, 1943), has ever been formally described from Vene- zuela. Other occurrences of Venezuelan fossil chelonians have, 1 Museum of Comparative Zoology, Harvard University. 2 Escuela de Geologia y Minas, Universidad Central de Venezuela, Caracas. 2 BREVIORA No. 376 however, occasionally been mentioned: Swinton (1928) referred three specimens "too fragmentary for specific determination," of Pliocene or Pleistocene age (Van Frank, 1957: 22n), to "Testuclo sp."; and Royo y Gomez (1960) published a faunal list of fossil vertebrates discovered in what is now known as the upper member of the Urumaco Formation, that included ". . . varios capara- zones completos . . . de tortugas palustres (Podocnemys) . . . ." which he believed to be of mid-Miocene age. Unfortunately, the untimely death of Royo y Gomez in 1961 prevented his describing any of this material. The specimens, however, are preserved in the collections of the Universidad Central de Venezuela in Ca- racas. Except for one report on rodent remains (Pascual and Diaz de Gamero, 1969) and another on a new gavial (Sill, 1970), nothing has been published on them. The fossil turtles are de- scribed here. Abbreviations used in this paper are: AMNH — the American Museum of Natural History, New York MCN — Museo de Ciencias Naturales, Caracas VF — Laboratorio de Paleontologia, Escuela de Geologia de la Facultad de Ingeniera de la Universidad Central de Venezuela (Caracas). CLASSIFICATION AND DESCRIPTION Order Testudines Suborder Pleiirodira Family Peloiiiedusidae Genus Podocnemis A cautionary word is appropriate concerning the attribution of fossil material to Podocnemis. Most extinct species of this genus are known solely from shell material. However, at least one other fossil genus, Botliremys, has a shell that is virtually indistinguish- able from that of Podocnemis. In fact, two purported species of Podocnemis, P. barberi (Schmidt, 1940) and P. alabamae (Zan- gerl, 1948), were originally described on the basis of shell material that, upon the later discovery of associated skull material, has subsequently been referred to the genus Botliremys (GafTney and Zangerl, 1968). Conceivably, therefore, when skulls become known for some of the fossil species now referred to Podocnemis 1971 NEW VENEZUELAN FOSSIL PELOMEDUSID 3 on the basis of their shells, it may be necessary to transfer them elsewhere. The new species described in this paper is placed in Podocnemis with this caveat. Podocnemis venezuelensis sp. nov. (Figures 1-3 and Plates I-V ) Type. VF 1176, a complete carapace, slightly flattened, includ- ing much of the axial buttresses and the pelves, the latter badly damaged. Also, VF 1173, a fairly complete but badly fractured plastron lacking only the entoplastron and epiplastra. Although bearing a diff'erent number, the plastron is of an appropriate size and state of preservation so that we believe it highly probable that both of these shell elements are from the same individual (see Plate IV). Unfortunately, none of Royo y Gomez's notes contain information bearing on this point. Hypodigm. The type, and VF 1174, a plastron lacking the posterior lobe; VF 1175, the anterior one-third of a carapace; VF 1177, a complete carapace and plastron, considerably flattened, of a relatively small individual; VF 1177a, a nearly complete carapace and plastron of the largest individual in the sample, lacking only the pygal and posterior portions of both eleventh peripherals; VF 1177b, the left epiplastron, hyoplastron, and left one-half of the entoplastron, plus several pleurals attached to part of the bridge, probably all belonging to the same individual; VF 1059, the distal end of a right humerus. Horizon and locality. The "capa de huesos" or "capa de tor- tugas" (Pascual and Diaz de Gamero, 1969: 373 and map) of the upper member of the Urumaco Formation, probably of mid- Pliocene (Huayquerian) age, north of Campo El Mamon, state of Falcon, Venezuela. The Uramaco Formation is divided into three members, desig- nated as lower, middle, and upper. The lower and middle mem- bers are largely, if not entirely, of marine origin while the upper member apparently includes both littoral or deltaic as well as more strictly terrestrial facies. Marine molluscs from the middle mem- ber were the basis for the original age determination of mid- Miocene for this formation. This was subsequently revised to late Miocene. However, the recent study of Pascual and Diaz de Gamero (1969: 379) indicates that at least the "bone bed" of the uppermost part of the upper member is probably of middle BREVIORA No. 376 Figure. 1 Reconstruction of the shell of Podocnemis venezuelensis (dorsal view). The scale represents a length of 15 centimeters. The posi- tions of the axial and inguinal buttresses as well as of the iliac scars of the pelvis on the visceral surface of the carapace are indicated by the broken lines. 1971 NEW VENEZUELAN FOSSIL PELOMEDUSID Figure 2. Reconstruction of the shell of Podocnemis veneziielensis (ventral view). Same scale as in Fig. 1. The positions of the pelvic scars on the visceral surface of the plastron are indicated by the broken lines. 6 BREVIORA No. 376 Pliocene (Huayquerian) age. Palynological evidence appears to confirm this date (A. E. Gonzalez Guzman, personal communica- tion). It is possible that there may be a temporal hiatus between the middle and upper members of the Urumaco Formation that cannot be detected structurally; an explanation of this nature would account for the different age determinations of the middle and upper members of the Urumaco Formation. Diagnosis. Differs from all other species of Podocnemis in totally lacking neural bones. Description. It is unfortunate that a gypsiferous encrustation on the external surfaces of all of the specimens obscures most of the bone sutures and virtually all scute sulci. Nevertheless, an essentially complete osteological description of the shell is possible from examination of the visceral surfaces of the type and some other, less complete specimens. Although the outlines of any par- ticular bone may differ somewhat from the external to the internal surface of the shell, as Dacque (1912: 290 and fig. 7) has demon- strated, it is nonetheless possible to make a reasonable reconstruc- tion of the shell of a typical representative of the species (see Figures 1 and 2 ) . Adult specimens of Podocnemis venezuelensis attained a rather large size. The three complete carapaces in our sample range from approximately 46 to 67 centimeters' in length. Of the eight living species of Podocnemis, only two, P. expansa and P. unifilis, are reported to reach a larger maximum size. In cross section, the arch of the carapace is very flat; in outline it is somewhat oval and greatly expanded posteriorly. The external surface of the carapace is completely smooth. No significant in- dentation occurs in the nuchal region. Well-developed axial and inguinal buttresses unite the carapace to the plastron. The plastron itself is essentially flat and of nearly uniform thickness throughout. The outstanding feature of the carapace, and indeed the char- acter permitting definition of a new species, is the complete ab- sence of neural bones. Otherwise, the carapace is typical of other South American representatives of the genus. The nuchal bone is roughly pentagonal and slightly broader than long in its maxi- mum dimensions; its postero-lateral borders are bowed outwards 1 The larger number represents the estimated total length of VF 11 77a; its actual midline length as preserved is 60.1 centimeters. 1971 NEW VENEZUELAN FOSSIL PELOMEDUSID slightly. There is no indentation at the midline of the anterior margin. The eight pairs of pleurals meet in the midline. As in all pelomedusids, there are eleven pairs of peripherals. The pygal is trapezoidal and the suprapygal is roughly subtriangular. But- tresses of the axial and inguinal notches are attached to the under- sides of the first and fifth pleurals respectively. The iliac scars of the pelvis are situated on the visceral surfaces of the seventh and eighth pair of pleurals. Of the three elements of the plastron, the bridge is the longest while the anterior lobe is the shortest. The anterior plastral lobe is U-shaped and does not extend beyond the anterior lip of the carapace. The lateral margins of the posterior lobe are straight rather than curved and are inclined medially so that the posterior lobe becomes narrower toward the rear. The entoplastron is diamond-shaped and has a slight U-shaped ridge with the open end facing anteriorly on its visceral surface; this is presumably for the attachment of neck muscles. At the midline junction of the epiplastra there is a pronounced protuberance on the visceral side. The mesoplastra are subrounded to hexagonal elements situated ^ ^ Figure 3. The shape of the anal notch in three specimens of Podocnemis veneziielensis (from left to right, VF 1177, VF 1173 and VF 1177a). The arrow points toward the anterior end of the shell. The scale represents a length of 10 centimeters. 8 BREVIORA No. 376 laterally at the base of the bridge. (These can best be observed on the internal surface of VF 1174; see Plate V.) There is some variation in the shape of the anal notch; in the smallest specimen, VF 1 177, it is V-shaped, but in the two larger examples, VF 1 173 and VF 11 77a, it is broader at its base and thus tends to be more U-shaped (Fig. 3). This structural difference may be due to sexual dimorphism. If so, however, it is not possible to specify which type represents the males and which the females in this species. Normally, male turtles can be distinguished from females by the characteristic depression on the posterior lobe of their plastrons, but it is not possible to determine whether or not these existed in the present sample because, in the course of fossilization, all the plastra have been somewhat depressed inward along the midline. Size alone is not a useful criterion for identifying the sexes in Podocneinis. In all but one of the living South American species of this genus mature females are invariably larger than males. The reverse is true, however, for P. dumeriliana (F. Medem, personal communication ) . Nor does the shape of the anal notch necessarily provide a reliable method of sex determination. Although in at least one species, P. lewyana, males can be identified solely on the basis of this character, there are other species (e.g., P. vogli) in which the two sexes appear to have similarly shaped anal notches, while in still others (e.g., P. unifilis) the shape of the anal notch varies considerably but evidently also randomly with respect to sex. Pelvic scars on the plastron appear to be disposed in essentially the same positions as in the living South American species. Although partial pelves have been preserved in several of the specimens (VF 1173, 1176, 1177, and perhaps also 1177a), none of these are well enough preserved to describe in detail. The distal half of a right humerus (VF 1059) is all that is known of the appendicular skeleton. Nothing serves to distinguish this limb fragment from comparable portions of this same bone in other species of Podocnemis. Virtually no scute sulci can be detected in any of the specimens available for study. This is disappointing because their arrange- ment, particularly on the anterior plastral lobe, is sometimes of taxonomic sianificance. 1971 NEW VENEZUELAN FOSSIL PELOMEDUSID 9 DISCUSSION Taxonomic considerations. Within the suborder Pleurodira, the presence or absence of neural bones has been accorded vary- ing taxonomic significance. Those chelyid species that do possess neurals usually have a variable number, and in certain species of this family neurals evidently may or may not be present in diff"erent individuals of the same population. Consequently, whether or not a chelyid carapace includes some neurals, and if so, how many, has never been considered a useful taxonomic character ^ Most pelomedusid species, however, have a neural series that does not deviate from a modal number (usually 6-8) by more than one or, occasionally, two. Only two exceptions to this typical condition are known within the family, both involving extinct taxa from the Eocene of Tunisia-. Gajsachelys (de Stefano, 1903; Bergounioux, 1952; 1955; 1956), hke some chelyids, appears to have had a rather variable number of irregularly shaped neurals. Eusarkia (Bergounioux, 1952; 1956), described on the basis of a single specimen, has no neurals and on the basis of this and several other characters noted by Bergounioux we believe that it was probably appropriate to propose a new genus. But in no case has the absence of neurals alone been used as a taxonomic character, either at the generic or the specific level, within the Pleurodira (or among any other chelonians for that matter). Why, then, have we described the Urumaco fossil pelomedusids as a new species of Podocnemis? In view of their strong overall resemblance to the living South American species of this genus it seems inappropriate to propose a new genus on the basis of a single character which, by itself, is not highly unusual nor of particularly great taxonomic significance among other members of the suborder. On the other hand, since the absence of neurals is clearly a constant character within the Venezuelan sample, this 1 No adequate osteological descriptions of any living chelyid species have ever been published, so that reliable data are not actually available regarding the extent of intraspecific variation in the number of neurals. -One of us (RCW) is preparing a discussion of the taxonomic status of the Tunisian fossil turtles for publication elsewhere. In this paper Euclastochelys (Bergounioux, 1955; 1956) is considered to be synonymous with Gajsachelys. 10 BREVIORA No. 376 feature can hardly be regarded as an aberrant condition of no taxonomic consequence. Thus, by a process of elimination, the only alternative is to choose a procedure intermediate between regarding the lack of neurals as of enormous taxonomic impor- tance or as of none at all and describe the Venezuelan material as a new species. Ecological considerations. The small vertebrate fauna with which Podocnemis veneziielensis is associated (Royo y Gomez, 1960: 509; Pascual and Diaz de Gamero, 1969: 370 and 374) is not adequate for determining with any degree of certainty what the probable habitat of this species might have been. The mam- mals — a eumegamyine rodent and a toxodontid — were un- doubtedly strictly terrestrial forms, while the crocodilians pre- sumably spent most of their time in streams, lakes, or swamps. The fish — sharks, sawfish, rays, catfish, and an unidentified teleost — appear to be a mixture of marine and fresh water forms. If all these fossils were collected from a single horizon, as the scanty field evidence would suggest, then the stratum in which they occur must represent an estuarine facies. On the basis of present evidence, therefore, it is impossible to determine unequiv- ocally whether P. veneziielensis was a marine or a fresh water form. To be able to do so would be particularly interesting be- cause, while all living pelomedusids are inhabitants of fresh waters, in the past some were marine and others were fresh water forms (Wood, MS). If P. veneziielensis were, in fact, marine, it would be the last recorded pelomedusid so adapted. No hving species of Podocnemis (or any other fresh water turtle) are found in the Maracaibo basin, in which the type local- ity of P. veneziielensis lies, although they are common to the south and east of this enclave in Venezuela as well as to the west of it in Colombia. Thus, P. veneziielensis occurs outside the pres- ent range of the genus. Should P. veneziielensis eventually prove to be a fresh water rather than a marine form, its extinction may be explicable in terms of the Pleistocene climatic history of tropical South America. The Maracaibo basin is ringed by mountains except on its seaward side and hence is effectively isolated from adjacent land areas. Conceivably, a period or periods of aridity during the Pleistocene (and evidence for severe climatic fluctua- tions in the tropics during this epoch is accumulating — cf . Van- zolini and Williams, 1970: 94-103) may have eliminated P. 1971 NEW VENEZUELAN FOSSIL PELOMEDUSID 11 venezuelensis, a form presumably endemic to the basin, while the surrounding mountain barrier prevented subsequent recolonization by other species. The fossil record of Podocnemis in South America. Few fossil species of Podocnemis have been described from this continent and several of these are known from such inadequate material that it is questionable whether or not they should be referred to the genus. As previously noted, Simpson (1943) has described a partial carapace and plastron from Venezuela as Podocnemis geolo- gorum^. The single known specimen was recovered from fluvia- tile beds of Miocene age. Without a doubt, this fossil represents some kind of pleurodire because of the union of its pelvis with both carapace and plastron. Whether this specimen actually rep- resents a pelomedusid instead of a chelyid, however, is not entirely clear. Although Simpson (1943: 57) commented "Es muy po- sible la existencia en esta especie de un mesoplastron tipo Podoc- nemis de buen tamafio," the presence or absence of this pair of bones, the critical character for distinguishing members of one pleurodiran family from the other, cannot be determined. Very few potentially useful taxonomic characters can, in fact, be dis- cerned. On the carapace, only two pleurals separate the last neural from the suprapygal. There appears to be a deep indenta- tion in the posterior edge of the pygal at the midline, and this, together with similar but not quite so pronounced indentations in the tenth and eleventh peripherals, gives the rear margin of the carapace a serrated appearance. Whereas the last vertebral was broader than long, the two preceding ones were longer than broad. The anal notch of the plastron is rather deep and narrow, and the lateral margins of the posterior lobe are sinuous. As a consequence, the xiphiplastral tips are much more elongate than in any other taxon yet described as a pelomedusid. Simpson (1943: 61) considered the deep anal notch and serrations along the posterior border of the carapace to be the species-specific characters of P. geologorum, and indeed, these are quite distinctive and indicate the validity of the species. Unfortunately, however, 1 When Simpson described the type of P. geologorum. it was catalogued as AMNH 6781. It now belongs to the collections of the Museo de Ciencias Naturales, Caracas, and bears the number MCN 915. 12 BREVIORA No. 376 on the basis of present evidence there is no reason to believe that P. geologorum is really a species of Podocnemis or, for that mat- ter, of any pelomedusid. It cannot yet be confidently allocated to either of the two pleurodiran families to which it must belong, the Chelyidae or the Pelomedusidae. Thus, until better material of this taxon becomes available, P. geologorum should be listed as Pleurodira incertae sedis. Three species of Podocnemis — P. harrisi (Pacheco, 1913), P. brasiliensis (Staesche, 1937)\ and P. elegans (Suarez, 1969) — have been described from three widely separated localities within the Bauri'i Formation of southern Brazil. These sediments repre- sent terrestrial deposition, with fluvial and alluvial plain beds predominating, and are probably of late Cretaceous (Senonian) age (Oliveira, 1956: 53-54). The only specimens ever referred to P. harrisi were a nearly complete right xiphiplastron and sev- eral peripherals (Pacheco, 1913: 37, pi. 3 [figs. 6a-e], pi. 4 [fig. 6]). Ischial and pubic scars on the visceral surface of the xiphiplastron clearly indicate that some kind of pleurodire is rep- resented, but no other taxonomically useful evidence exists. It is therefore impossible to determine whether P. harrisi is a chelyid or a pelomedusid. Reference of this species to Podocnemis was unjustifiable, as Schmidt noted long ago (1931: 253). Further- more, since the type material is now apparently lost (Price, 1953: 10), "P. harrisi" must be regarded as a nomen vanum. On the basis of some photographs of a partial plastron, a car- apace fragment-, and three associated pleurals, Staesche (1937) described P. brasiliensis. The xiphiplastra of this species do not appear to dift'er in any appreciable way from those of P. harrisi, and therefore Simpson (1943: 61) may well have been correct in suggesting that P. brasiliensis is a synonym of P. harrisi, although Staesche (1937: 302-303) noted that his material dif- fered in that it represented a somewhat larger individual with a difl'erent kind of sculpturing on the external surface of the shell, factors which might or might not be of taxonomic significance. 1 Staesche (1944) is merely a translation of Staesche (1937) from German into Portuguese. -This carapace fragment, together with a previously undescribed ante- rior lobe of a plastron from the same locality, was subsequently referred to a new genus and species, Roxochelys wanderleyi, by Price (1953). 1971 NEW VENEZUELAN FOSSIL PELOMEDUSID 13 Small, laterally placed mesoplastra were present, even though no longer preserved, so that P. brasiiiensis is clearly a pelomedusid. An appropriate generic determination cannot at present be made, however, because much of the anterior plastral lobe, so critical for pelomedusid shell taxonomy, is missing. Thus, the specimen can neither be certainly referred to Podocnemis nor, owing it its imperfect preservation, can any species-specific characters be established. Until better material is available, therefore, the plas- tron and pleurals to which the name P. brasiiiensis now applies should be designated as Pelomedusidae gen. et sp. indet. {"Podoc- nemis brasiiiensis" Staesche). Unlike the other two dubious "species" of Podocnemis from the Baurii Formation, P. elegans is clearly valid and referable to this genus. It is the only South American fossil pelomedusid yet described for which associated shells and skulls have been re- covered. Furthermore, it is the oldest representative of Podoc- nemis known anywhere. It was described on the basis of two specimens, an essentially complete shell and a well-preserved skull belonging to a different individual. A detailed description of this species will not be presented here since an account, based on addi- tional new material as well as the original hypodigm, is being pre- pared by one of us (RCW) for separate publication. Some of the salient characters may be briefly noted, however. In most respects the shell is typical of all South American species of Podocnemis, but the shapes of the first two neurals are unique: instead of being spindle-shaped, the first is hexagonal, with the postero-lateral sides much shorter than the antero-lateral ones; and the second, rather than being hexagonal, is subrectangular. A unique feature of the skull is the total absence of triturating ridges on the palatal surface of the upper jaw. All other species have from one to three triturat- ing ridges, the exact number being characteristic of different spe- cies. In addition, there does not seem to be an antero-posterior 1 Suarez (1969: 37) stated: "Designamos como tipo da nova especie a carapaca e plastrao com craneo e diversos elementos esqueletais . . ." The shell and skeletal elements belong to one individual and there is in fact a badly crushed skull (which was not illustrated or discussed) asso- ciated with them, but it is not the skull described by Suarez. This is an isolated one from a much larger individual. Both specimens are in the paleontological collections of the Faculdade de Filosofia. Ciencias e Letras de Presidente Prudente: they bear no catalog numbers. 14 BREVIORA No. 376 forehead groove between the orbits. Of the other South American species of Podocnemis, only P. dumeriliana lacks this groove. Perhaps the most extraordinary aspect of the skull of P. elegans is its modern appearance; archaic or ancestral features that one might expect to find in such an ancient species are notably lacking. Cattoi and Freiberg (1958) described Podocnemis argentinensis from the Santa Barbara Formation^ in the Province of Jujuy, Argentina. It is known from a large part of a plastron, lacking the terminal portions of the anterior and posterior lobes as well as much of the bridges, and most of the right epiplastron of a second individual. The systematic position of this form is uncer- tain. Laterally placed mesoplastra were definitely present, as evidenced by the semicircular excavations on either side of the plastron at the base of the bridge, so that argentinensis clearly represents some kind of pelomedusid. Too little of the shell has been preserved, however, to permit assignment to Podocnemis with any degree of confidence. At a lower taxonomic level, the characters used to define the species are in some cases questionable and in others of little or no taxonomic significance. Cattoi and Freiberg described the entoplastron as cordiform, but in their figure and plate it appears to be quadrangular. Examination of the specimen itself indicates that the entoplastron is slightly dam- aged anteriorly and that it was probably diamond-shaped orig- inally. Small, triangular gular scutes are characteristic of most pelomedusids, but the relatively small intergular, which was prob- ably pentagonal, is certainly reminiscent of the condition typical of South American species of Podocnemis and the North American Bothremys (Gaftney and Zangerl, 1968). The various scute pro- portions cited by Cattoi and Freiberg in their diagnosis yield no useful taxonomic information. Anastomosing vermiculations cover the external plastral surface as in the majority of pelomedusid 1 Various ages have been assigned to this stratigraphic unit, formerly referred to as the Margas Multicolores. Cattoi and Freiberg placed it in the late Cretaceous, while Bardack (1961) considered it to be middle Tertiary. The recent discovery of a mammal skull high in the formation indicates a Paleocene or early Eocene age (R. Pascual, personal com- munication). There is some uncertainty as to whether these beds are of marine or terrestrial origin (Cattoi and Freiberg, 1958: 59). 1971 NEW VENEZUELAN FOSSIL PELOMEDUSID 15 genera'. Until better material becomes available, it will not be possible to determine the systematic position of this turtle. In the meantime, it must be referred to as Pelomedusidae gen. et sp. indet. {'Podocnemis argentinensis" Cattoi and Freiberg). The type and only specimen of Podocnemis bassleri (Williams, 1956) is a large, exceedingly well-preserved skull. It was collected in eastern Peru from beds of the Contamana Group, which in- cludes sediments believed to range in age from Eocene to possibly at late as Pliocene. Williams, on the basis of information supplied by Kummel, reported that the skull "came from the uppermost part of the . . . group," which suggests that its age falls within the latter part of the Tertiary. He further remarked that "The fossil itself is so close to a Recent species as to tend to support the latest date geologically permissible." The skull differs only in minor details from that of the living P. expansa. It seems fairly certain that P. bassleri was closely related, if not directly ancestral, to this species. Unfortunately, the evolutionary history of Podocnemis in South America cannot be reconstructed on the basis of present informa- tion. Only three fossil species — P. bassleri, P. elegans, and P. venezuelensis — are of unquestionable validity. One of these, P. bassleri, is clearly very closely related to P. expansa. The rela- tionships of the other two extinct species to living South American forms are uncertain, owing to their distinctive shell characters. With the exception of P. lewyand'-, the living South American species of Podocnemis are all strikingly similar in terms of shell 1 Cattoi and Freiberg's figure I shows the entoplastron as being nearly encompassed by unusually large epiplastra. a condition unknown in any other chelonians that we are aware of. However, their sketch does not accurately represent the positions of the sutures between the epiplastra and hyoplastra; these are, in fact, disposed in typical pelomedusid fashion, extending outward from the lateral apices of the entoplastron. -Through the courtesy of Professor F. Medem, one of us (RCW) has been able to examine a series of six P. lewyana shells in the collections of the Instituto Roberto Franco at Villavicencio, Colombia. None of these has a suprapygal bone on the carapace; instead, each of the last (eighth) pleurals is subtriangular, not trapezia! as is the case in other species of the genus, and these pleurals are in continuous contact along the midline from the posterior end of the seventh pair of pleurals to the pygal. 16 BREVIORA No. 376 morphology. Only small structural details characteristic of each taxon permit differentiation among them on the basis of shells alone. P. venezuelensis stands markedly apart from all other spe- cies of the genus in its total lack of neurals, and certainly could not have given rise to any of the living forms. Nothing remotely resembling the shapes of the first two neurals in P. elegans is en- countered elsewhere in the genus. So conservative in structure is this part of the shell in all other species (except, of course, for P. venezuelensis ) , and so radically different is it in P. elegans, that this species could hardly have been ancestral to any or all of the later species known from South America. Thus, neither P. elegans nor P. venezuelensis has any obvious relationship to living species of the genus or to each other. A much better fossil record for Podocnemis will be necessary before a meaningful picture of its evolutionary history in South America can be formulated. ACKNOWLEDGMENTS We are particularly grateful to Senora Frances Charlton de Rivero, retired professor of paleontology at the Escuela de Geo- logia y Minas in Caracas, not only for her gracious hospitality but also for having provided laboratory facilities while the material here described was being studied. Both authors have examined the type specimen of "Podocnemis'' geologorum and one of us (RCW) has also been able to examine the types of "P. argen- tinensis," P. bassleri, and "P. brasiliensis." We would like to thank the curators of the various institutions at which these fossils are housed for permission to study them. Through the kindness of Professor Jose Martin Suarez, RCW has not only been able to study the type material of P. elegans but also to visit the locality from which it was recovered. Without a generous grant to RCW from the National Geographic Society, none of this work would have been possible. We are much obliged to Professors F. Medem, B. Patterson, P. E. Vanzolini, and E. E. Williams, and to Dr. M. Freiberg for their comments and discussions on various aspects of this manuscript. 1971 NEW VENEZUELAN FOSSIL PELOMEDUSID 17 LITERATURE CITED Bardack, D. 1961. New Tertiary teleosts from Argentina. American Mils. Novitates. No. 2041: 1-27. Bergounioux, F. M. 1952. Les cheloniens fossiles de Gafsa. Appendix to: Arambourg, C, and J. Signeux. Les vertebres fossiles des gise- ments de phosphates ( Maroc-Algerie-Tunisie). Notes et Memoires No. 92, Service Geologiqiie de Maroc: 377-396, pis. 45-46. . 1955. La famille des Eusarkiides. Comptes Rendus de I'Acad. Sci.. 240: 1455-1457. 1956. Les reptiles fossiles des depots phosphates sud tunisiens. Annales des Mines et de la Geologie (Tunis), No. 15: 1-105, pis. 1-17. Cattoi, N., and M. a. Freiberg. 1958. Una nueva especie de "Podoc- nemis" del cretaceo argentino. Physis. 21(60): 58-67. Dacque, E. 1912. Die fossilen Schildkroten Aegyptens. Geol. Palaont. Abhandl., N. F., 10(4): 275-337, pis. 36-37.^ Gaffney, E. S., and R. Zangerl. 1968. A revision of the chelonian genus Bothremys (Pleurodira, Pelomedusidae). Fieldiana. Geol., 16(7): 193-239. Oliveira, a. I. DE. 1956. Brazil. //; W. F. Jenks (ed.). Handbook of South American Geology. Geol. Soc. Amer. Mem. 65: 1-378. Pacheco, J. A. 1913. Notas sobre a geologia do valle do Rio Grande, a partir da foz do Rio Pardo ate a sua confluencia com o Rio Paranahyba. /// J. dos Dourados (ed.). Explora^ao do Rio Grande e de seus afflu- entes. Commisao Geographica e Geologica do Estado de Sao Paulo: 33-38. Pascual, R.. and M. L. Diaz de Gamero. 1969. Sobre la presencia del genero Eiimegamys (Rodentia, Caviomorpha) en la formacion Urumaco del Estado Falcon (Venezuela). Su signification cronologica. Bol. Informativo, Assoc. Venezolana de Geol.. Min. Pet.. 12(10): 369-387. Price, L. I. 1953. Os quelonios de forma<:ao Baurii, Cretaceo terrestre do Brasii meridional. Dept. Nac. Prod. Min., Div. Geol. Min., Bol. 147: 1-34. RoYO Y Gomez, J. 1960. Les vertebrados de la formacion Urumaco, estado Falcon. Mem. Ill Congr. Geol. Venezolano, 2: 506-510. Schmidt, K. P. 1931. A fossil turtle from Peru. Field Mus. Nat. Hist. Publ. 299, Geol. Ser., 4(8): 251-254. 1940. A new turtle of the genus Podocneniis from the Cretaceous of Arkansas. Geol. Ser. Field Mus. Nat. Hist.. 8( 1): I-I2. 18 BREVIORA No. 376 Sill, W. D. 1970. Nota preliminar sobre un nuevo gavial del Plioceno de Venezuela y una discusion de los gaviales sudamericanos. Ameghiniana, 7(2): 151-159. Simpson, G. G. 1943. Una tortuga del Terciario de Venezuela. Rev Ministerio de Fomento, Caracas, Venezuela, ano 5, nos. 51-52, April -Sept. 1943: 53-64. Staesche, K. 1937. Podocneinis hrasiliensis n. sp. aus der Oberen Kreide Brasiliens. Neues Jahrb. Min., Geol. Palaont., 77: 291-309. . 1944. Uma tartaruga do cretaceo superior do Brasil. Dep. Nac. Prod. Min., Div. Geol. Min., Bol. 114: 1-24, pis. 16-20. Stefano, G. de. 1903. Nuovi rettili degli strati a fosfato della Tunisia. Bol. Soc. Geol. Ital., 22: 51-80. Suarez, J. M. 1969. Urn quelonio da formagao Bauru. Dept. Geografia, Fac. de Filos., Cien. Letras Pres. Prudente, No. 2: 35-54. SwiNTON, W. E. 1928. Note on the fossil reptilia collected by Mr. Don- ald Stewart in Venezuela. Quart. Jour. Geol. Soc, 84, Pt. 3: 583. Van Frank. R. 1957. A fossil collection from northern Venezuela. 1. Toxodontidae (Mammalia, Notoungulata). American Mus. Novi- tates. No. 1850: 1-38. Vanzolini, p. E., and E. E. Williams. 1970. South American anoles: the geographic differentiation and evolution of the Anolis chrysolepis species group (Sauria, Iguanidae). Arq. Zool.. S. Paulo, 19(1-2): 1-124. Williams, E. E. 1956. Podocneinis bassleri, a new species of pelomedusid turtle from the late Tertiary of Peru. American Mus. Novitates, No. 1782: 1-10. Zangerl, R. 1948. The vertebrate fauna of the Selma Formation of Alabama, part 11. The pleurodiran turtles. Fieldiana: Geol. Mem. 3(2): 19-56. 197 NF.W VENEZUELAN FOSSIL PELOMEDUSID 19 15 cm Plate L Dorsal view of the type carapace (VF 1176) of Podocnemis veneziielensis. 20 BREVIORA No. 376 15 cm Plate II. View of the visceral surface of the type carapace (VF 1176) of Podocnemis venezuelensis. Note the absence of neural bones. 1971 NEW VENEZUELAN FOSSIL PELOMEDUSID 21 -^ ^•^^. V:^'' i t-r '■ 15 cm Plate ill. View of the visceral surface of the type plastron of Pocloc- neinis veneziielensis (VF 1173). showing the disposition of the pelvic attachments to it. 22 BREVIORA No. 376 ■^- '^■■<%5 15 cm Plate IV. Ventral view of the type shell of Potlocneiuis vcnezuelensis, with the plastron (VF 1173) positioned correctly in relation to the carapace (VF 1176). 1971 NEW VENEZUELAN FOSSIL PELOMEDUSID 23 Plate V. The visceral surface of VF 1174. a partial plastron of Podoc- nemis venezneleii.sis, showing clearly the outlines of the laterally placed mesoplastra. BREVIORA MmseiLim of Comparative Zoology CAMBRrocE, Mass. 15 June. 1971 Number 377 THE CHANARES (ARGENTINA) TRIASSIC REPTILE FAUNA IX. THE CHANARES FORMATION Alfred Sherwood Romer Abstract. The term "Chanares Formation" should be retained for the fossihferous beds so named by Romer and Jensen; the type section of the "Ischichuca Formation" is homologous with part of the overlying Los Rastros Formation. With the discovery of a rich Triassic reptilian fauna in the Chafiares-Gualo region of La Rioja Province, Mr. James Jensen and I set about the task of determining the stratigraphic sequence of the area (Romer and Jensen, 1966). It was soon apparent that, despite complex faulting, a series of formations with clear-cut boundaries could be distinguished, and in the Arroyo del Agua Escondida the entire local series could be seen in proper sequence. The area is part of a basin of late Paleozoic and early Mesozoic deposition lying in western La Rioja Province and an adjacent portion of San Juan, extending roughly from the western flanks of the Sanogasta Range on the east to the Rio Bermejo on the west, and from the region of Villa Union south to the northern end of the Valle Fertil range. The center of this area is the flat Campo de Talampaya, and I shall term this cuenca the Talampaya Basin. Little attention had ever been given to the geology of our area of interest in the eastern part of the basin, but considerable work had been done in the western part. It was obvious that the thickness and nature of the sediments varied greatly from one basin area to another, and the situation was further complicated by the fact that there had been much volcanic activity. As far as possible we utilized formation names already in the literature; \vhcn no similarities to named formations in other areas in the hasin were discernible, new names were given. The formations named by us are shown in the right hand column of the table. The 2 BREVIORA No. 377 three formations latest in time — Los Colorados, Ischigualasto^ and Los Rastros — are comparable to those so named to the west, across the Campo de Talampaya, although much thinner in our area in each instance. Below the normal beds of the Los Rastros, strongly carbonaceous in nature, are some 70 meters of evenly bedded volcanic ash sediments in which fossil reptiles are abun- dant. Despite the fact that these strata are quite conformable with the overlying Los Rastros deposits, they are, as layers of white to bluish white ash, quite distinctive in character and, since they are the bearers of our fauna, we ventured to separate them from the Los Rastros as a distinct Chanares Formation. Below them, unconformably, are red and white sandstones which are roughly comparable to those generally assigned to "Paganzo IIL" Since this is not a proper stratigraphic term, we have given them the name of the Tarjados Formation. Beneath them, again, are thick series of soft sandstones, perhaps also part of the "Paganzo III" complex, for which we have found no clear equivalents in other parts of the basin, and which we have named the Talam- paya Formation. In general, I think, this terminology has been accepted. The one exception is that Sr. Bonaparte (1967, 1969, etc.) has ob- jected to our term "Chaiiares Formation," and maintains that these beds should bear the name "Ischichuca Formation." Despite the high regard I have for Sr. Bonaparte's work in the collection and description of Triassic fossils, I believe that this usage is inde- fensible. A review of the history of stratigraphic work in this region is necessary. The first serious study of this region was that of Bodenbender (1911). His stratigraphic section in this basin is given in the first column of the table. He believed that the sediments in this area extended from the Carboniferous to the Cretaceous, and for the lower beds in the region invented the term "Paganzo," divided into "Paganzo I" for light colored sediments which he believed (apparently correctly) to be of Carboniferous age, "Paganzo II," for red sandstones which (again probably correctly) he thought were Permian, and "Paganzo III," for beds, mainly reddish sand- stones, which (with less assurance) he claimed to be Triassic. ^ Sr. R. R. de la Vega has pointed out to me that the correct spelling should be Ichigualasto, but the "improper" version has become so embedded in the literature that it seems impossible to eradicate it. 1971 THE CHANARES FORMATION 3 "Rhaetic" was a favorite term with early German workers in South American geology, favored perhaps because of its rather vague meaning, and to the "Rhaetic" he assigned a very considerable thickness of beds including yellow and variegated shales, coal shales, and coal seams. Above these (beyond some indeterminate beds in the region of Cerro Morado, which he suggested were possibly Jurassic) the depositional series terminated with thick red sandstones which he believed to be Cretaceous in age and termed the "Cretaceo Andino." A more thorough study of the beds of the western part of the basin was undertaken in the 1940's by Frenguelli and by de la Mota. Frenguelli mainly visited the region of the Ischigualasto Valley, drained to the west by the Rio de la Pena, and published his results in 1948. Bodenbender's "Cretaceo Andino" beds were termed by him the "Estratos de Gualo," and their presumed age reduced, reasonably, from Cretaceous to "Rhaetic." In Boden- bender's erstwhile "Rhaetic" series he distinguished an upper mem- ber as the Ischigualasto Formation — a series of variegated shales and some sandstones characteristic of the Ischigualasto Valley west of the red bluffs of the "Gualo." These beds are now known to possess a very considerable fauna of Triassic (? Ladinian or pos- sibly Carnian) age. Below the Ischigualasto, to the west, and traversed by the difficult gorge of the La Pena, are rugged hills of sandstones and shales including coal seams, in which footprints had been discovered (Huene, 1931). As Frenguelli notes (1948: 191), he did not penetrate much farther to the west through the rugged country in the Ischigualasto-La Pena region than the Que- brada de los Rastros, where a coal mine is located and where the footprints were found; some data were furnished him by Ramac- cioni and Heim, who studied the coal beds (Heim, 1949). They informed him that these "Rhaetic" beds were underlain, to the west, by red "Paganzo" sandstones which he cites as "Paganzo 11" (they are actually of "Paganzo III" age). The total thickness of the "Rhaetic" beds below the Ischigua- lasto Formation in this region was estimated as 600-650 meters. Frenguelli made this area the type section of the Los Rastros Formation, to which he assigned the upper 400-450 meters of these beds — i.e., about two-thirds, on his reckoning. As can be seen from the sections of Ortiz, mentioned later, under this defini- tion of the Los Rastros the formation would include essentially 4 BREVIORA No. 377 the whole of the coal-bearing portion of the "Rhaetic" beds. In de- fault of personal knowledge of the lower beds in the Ischigualasto- La Pena region, Frenguelli turned to the region of Cerro Bola, some 70 km to the north, which had been studied in detail by de la Mota (whose results are recorded in an unpublished thesis at the University of La Plata ).^ Equivalents of the "Gualo" and Ischigualasto beds are readily determinable in the Cerro Bola region. Between the Ischigualasto and "Paganzo III" there are here, much as in the Ischigualasto region, some 500-550 meters of "Rhaetic" beds of sandstones, shales, and coal seams. The upper 250-300 meters of these beds were equated by de la Mota and Frenguelli with the type Los Rastros; they contain occasional carbonaceous seams, but consist mainly of shales and fine-grained olive-green sands. The beds below, with a thickness of about 250 meters, are here the main coal-bearing strata, dominantly black or grey-black in color, with some intercalations of thin olive-green sands. These beds were named the Ischichuca Formation, the type section being located in the quebrada of that name south of Cerro Bola. Below the Ischichuca Formation lies "Paganzo III." This consists mainly of coarse red sandstone and intercalated volcanic beds; between the beds of "Paganzo 111" proper and the Ischichuca are 30-40 meters of coarse grey to reddish conglom- erates which de la Mota assigned to "Paganzo III." Frenguelli's sections of the basin, derived from two areas 70 km apart, are given in the second and third columns of our table. In 1953 Groeber and Stipanicic (pp. 87-93), in their review of the Triassic, followed FrengueUi, as given in our column 4, ex- cept that his "Estratos de Gualo" are renamed "Estratos de los Colorados," (since de la Mota had pointed out to them that the Mogote del Gualo lies at a much lower stratigraphic position than FrengueUi beheved). A decade later than the visit of Frenguelli to the Talampaya Basin, study of the region to the west of Ischigualasto, down the La Peiia, was undertaken by Ortiz on behalf of the Yacimientos Petroliferos Fiscales. His work was done in 1964, and his sections and maps were promptly circulated, although publication of his work was not made until 1968. As his sections show, the actual 1 Resumes of de la Mota's findings are given by Frenguelli (1948: 197- 208) and by Groeber and Stipanicic (1953: 93-95). 1971 TMF CHANARES FORMATION 5 sequence of the "Rhaetic" beds here, below the Ischigualasto, dif- fers considerably from that imagined by Frenguelli and studied by him in the Cerro Sola region, and the strata are considerably thicker than had been believed. The upper part of the type Los Rastros Formation, to about 250 meters, consists mainly of grey sandstones with intervening shales and only a few coal seams. Below this, but also included by Frenguelli — entirely or at least for their most part — in the type section of the Los Rastros For- mation are about 400 meters of beds in which dark carbonaceous shales and coal seams predominate, with intervals of olive-green shales. Below the coal beds, again, are about 450 meters of con- glomeratic sandstones, light in color, in which coals are little developed. Still farther down to the west appear the red sand- stones of "Paganzo IIL" It is clear that Frenguelli's nomenclature of the Los Rastros and Ischichuca beds in the two regions, that of the La Pena and the Ischichuca quebrada, is conflicting. The beds assigned to the Los Rastros in the Ischichuca region include only the upper part of the type Los Rastros; the middle, coal-bearing section of the beds which were included in the type Los Rastros in the La Pena region, form to the north practically the entire Ischichuca Forma- tion. The lowest portion of the "Rhaetic" beds in the La Peiia area has no counterpart in the north, unless it be the thin series of conglomerates that were mentioned above as transitional from "Paganzo III" to Ischichuca. As Ortiz points out, the entire series of beds from "Paganzo III" up to the Ischigualasto forms a single cycle of deposition, to the whole of which the formation name Los Rastros should properly be applied. Further, since the sup- posed Ischichuca Formation is merely a portion of the Los Ras- tros — the lower portion by Frenguelli's definition, the middle portion under Ortiz's suggestion — the term "Ischichuca" should be abandoned, as simply a partial synonym of Los Rastros. The Triassic sequence in the Talampaya-Ischigualasto basin as interpreted by Ortiz is given in column 5 of our table. Our own interpretation, given in column 7 and mentioned earlier, is in most regards comparable. The identity of the Los Colorados and Ischigualasto formations is perfectly clear; we have given names, as the Tarjados and Talampaya formations, to the red sandstones and underlying finer sandstone beds that have previously been referred to under the vague generic term "Paganzo III." We have, with Ortiz, agreed that the coal-bearing sandstones and shales of 6 BREVIORA No. 377 the "Rhaetic" should be considered as a single Los Rastros For- mation rather than be subdivided in two in a confusing fashion. Our one difference has been in distinguishing the lowest part of this cycle as a separate Chanares Formation because of its dis- tinctive nature and, especially, because of the paleontological importance of this unit. Bonaparte, independently of Ortiz, studied the lower western portion of the beds which Ortiz has described and, like Ortiz and in contrast to Frenguelli's hypothetical interpretation, finds that the lower part of the Los Rastros complex consists of light colored shales without coal seams, and thus is somewhat comparable to our Chanares beds. He is, further, to be congratulated for finding in them fossils which are presumably comparable to those from the Chanares. Quite probably the lowest section of the Los Ras- tros of Ortiz and the Chanares beds are homologous (although the Los Rastros beds contain conglomerates, quite in contrast with the even bedding of the Chanares). How should the matter be treated? It may be advocated either that the Chanares be con- sidered as a lower member of the Los Rastros or that the lower segment of the Los Rastros in the La Peiia area be considered as a separate Chanares Formation. Bonaparte, however, would go further and revive the term "Ischichuca" for these beds. Essentially, his argument is that since these beds in the La Pena region were termed Ischichuca by Frenguelli (who had never seen them and was ignorant of their nature), the name should be retained. To so argue, however, is to violate the basic principles of stratigraphic nomenclature. One should refer to the type section of the Ischichuca, 70 km to the north. The type Ischichuca consists of the main coal-bearing section of the "Rhaetic." As Ortiz has shown, the supposed "Ischichuca" in the La Pena region is at a higher stratigraphic level than the beds with which we are concerned and is, by Fren- guelli's definition, equivalent to part of the Los Rastros. No two sets of beds could differ more in nature than the blackish "coal measures" of the Ischichuca and the clear white ash of the Chaiiares. Both the nature of the beds and their stratigraphic position distinguish clearly the type "Ischichuca" from the lower Los Rastros and Chanares beds. The use of "Ischichuca" for the latter strata cannot be defended. 1971 THE CHANARES FORMATION CO e o c <» {£> ^ CD -3 — o o C3 n (1) o o c "O -o 0) c k_ < u o V) v> o (/> o tr a o o D o Q. E C3 H o o a. o u o o CE D o o a. a B a. CQ x: c C _o u. a. u. C 5 o 8 BREVIORA No. 377 REFERENCES CITED BoDENBENDER, G. 1911. Constitucion geologica de la parte meridional de la Rioja y regiones limitrofes (Republica Argentina). Bol. Acad. Nac. Cienc. Cordoba, 19: 1-220. Bonaparte, J. F. 1967. Comentario sobre la "Formacion Chanares" de la cuenca Triasica de Ischigualasto-Villa Union. (San Juan-La Rioja). Acta Geol. Lilloana, 9: 115-119. 1969. Datos sobre la evolucion paleoecologica en las forma- ciones Triasicas de Ischigualasto-Villa Union (San Juan-La Rioja). Acta Geol. Lilloana, 10: 189-206. Frenguelli, J. 1948. Estratigrafia y edad del llamado Retico en la Argentina. Gaea, An. Soc. Argentina Est. Geogr., 8: 159-309. Groeber, P.F.C, and p. N. Stipanicic. 1953. Geografia de la Republica Argentina, II (Primera Parte): Triasico. Buenos Aires. Pp. 13-141. Heim, a. 1949. Estudio geologico del carbon "Retico" y del Valle de la Pena (provincias de San Juan y La Rioja). Bol. Dir. Gen. Ind. Min., 69: 1-31. Huene, F. 1931. Die fossilen Fahrten im Rhat von Ischigualasto in Nordwest-Argentinien. Palaeobiologica, 4: 99-112. OrtIz, a. 1968. Los denominados Estratos de Ischichuca como seccion media de Formacion Los Rastros. Actas III, Jorn. Geol. Argentina, 1: 333-339. RoMER, A. S., AND J. A. Jensen. 1966. The Chanares (Argentina) Trias- sic reptile fauna. II. Sketch of the geology of the Rio Chanares-Rio Gualo region. Breviora, No. 252: 1-20. BREVIORA Musenim of Compsirative Zoology Cambridge, Mass. 15 June, 1971 Number 378 THE CHANARES (ARGENTINA) TRIASSIC REPTILE FAUNA X. TWO NEW BUT INCOMPLETELY KNOWN LONG-LIMBED PSEUDOSUCHIANS. Alfred Sherwood Romer Abstract. Two types of hind limbs, with which incomplete remains of other skeletal parts are associated, are described as new genera and species, Lagerpeton chanarensis and Lagosiichus talampayensis. Both have long and slender legs, with the tibia longer than the femur, and a long slender foot. They differ, however, in foot construction, for in Lagerpeton digit II is short and digit IV the longest of the series, whereas in Lago- siichus metatarsals II-IV are subequal in length and digit III the longest in the foot. INTRODUCTION In the pseudosuchian material collected by the La Plata-Harvard expedition of 1964-1965, there are several forms represented by nearly complete skeletal material. In addition, however, in this collection and among specimens collected later by Sr. Bonaparte of the Instituto Lillo of Tucuman, there are less complete remains of further pseudosuchians. Two such forms are represented by materials including hind limbs of unusual and advanced char- acter; these will be described here. Lagerpeton chanarensis gen. et sp. nov. Holotype. La Plata Museum No. 64-XI-14-10 (field number 64), a hind leg, collected from the Chanares Formation in La Rioja Province, Argentina, about AVi miles east of the mouth of Rio Chaiiares. Combined generic and specific diagnosis. A pseudosuchian; hind limb very long and slender; femur with articular head sharply set off from shaft; tibia and fibula longer than femur; astragalus BREVIORA No. 378 and calcaneum fused and applied closely to tibia and fibula; toe IV longest of the hind leg digits; toe II much shorter than III or IV; toe V represented by short metatarsal only. Description. The type specimen (Fig. 1 ) was found quite isolated, not articulated with or accompanied by other skeletal materials. The femur is long and slender, with a length of 77 mm. It has the typical sigmoidal archosaur shape. The well-ossified curved articular area of the head is pronounced and set off at a sharp angle from the shaft. A marked angulation of the posterior Figure 1. Right hind leg of Lagerpeton, holotype. Left, external view; center, extensor surface of foot; right, internal view. X 1/2. 1971 LAGERPETON AND LAGOSUCHUS 3 margin of the shaft indicates the point of insertion of muscula- ture, presumably the iliofemoralis. On the medial surface below the head, there is a well-developed fourth trochanter in the shape of a pronounced ridge, presumably for the insertion of the caudifemoralis. The epipodials are even more elongate and slender than the femur; the tibia measures 92 mm in length, and is thus about 120 per cent the length of the femur. The tibia is well expanded proximally for a broad double articulation with the distal end of the femur. The fibula is incomplete in the type, and I have in my figure restored the proximal end from a specimen, discussed below, in the Instituto Lillo collection. It is, as always, a slender strap of bone expanded at either end; proximally, it articulates with the lateral surface of the distal end of the femur. The proximal tarsals are of unusual construction for a theco- dont. Astragalus and calcaneum are completely fused in the type specimen; in their combined outline they conform to the area of the distal end of tibia plus fibula, and this astragalo-calcaneum is closely applied to these two elements, although not fused with them. Further contributing to the close union of the astragalo- calcaneum with the epipodials is a triangular flange of bone extend- ing upward from this element posteriorly over the lower ends of tibia and fibula. There is here no indication of the crocodiloid calcaneal tuber frequently found in pseudosuchians. The usual two distal tarsals are present. A more medial and larger element is present on the ventral surface, above the third metatarsal; the lateral element caps metatarsal IV. Like the main limb elements, the foot is long and slender, the toes apparently lying close together in life. Toe V is represented only by a short, pointed metapodial; the other toes are complete, with the primitive phalangeal formula of 2, 3, 4, 5. The terminal phalanges of toes I-IV are sharp claws, somewhat curved. Most unusual for an archosaur, the toes are primitive in one regard, namely that toe IV is the longest of the series. Digit I is, as often, short; digit II is also shortened, its metapodial having but about half the length of those of its lateral neighbors. The total lengths of the digits^ from toe I outward, are 21, 49, 69, 85, and 18 mm. Metapodial lengths are 8, 24, 45, 48, and 18 mm. In the Instituto Lillo collections is a specimen that includes a hind Ice similar to that of the type in size, bone proportions, and 4 BREVIORA No. 378 structure. The femur is 75 mm in length, the tibia (not perfectly preserved) about 90 mm. The femur is closely comparable to that of the type with its sharply set off head and prominent trochanteric ridge. Toe I is not preserved, but toes II-IV resemble closely those of the type, with metapodial lengths of 23, 42, and 47 mm, and total toe lengths of 48, 74, and 87 mm. With this specimen is a pelvis which is apparently nearly com- plete, but in its present state of preparation is seen only from its ventral aspect. The ischia are well developed and extend far backward with a long symphysis. Below and somewhat to the rear of the level of the acetabulum their broad external surfaces are convex in section, giving this region a "swollen" appearance. Lagosuchus talampayensis gen. et sp. nov. Holotype. La Plata Museum No. 64-XI-14-11 (in part). Remains included in a slab collected from the Chaiiares Forma- tion in La Rioja Province, Argentina, about 2 km north of the mouth of Rio Chaiiares. Combined generic and specific diagnosis. A pseudosuchian; hind limb similar in many regards to that of Lagerpeton; limb long and slender; femur with articular head sharply set off from shaft; tibia and fibula longer than femur; astragalus and calcaneum fused and applied closely to tibia and fibula. Digit I short, digit V represented by short metatarsal only; digits II-IV subequal in metatarsal length, but digit III longest of the series. Description. A slab from the Chaiiares exhibits on one surface the greater part of the skeleton of a small ornithosuchid which I hope to describe in a later number of this series. On the other surface of the slab are scattered materials pertaining to a still smaller pseudosuchian. Best preserved are nearly complete and articulated hind limbs (Fig. 2). Although considerably smaller, these limbs in their proportions resemble those of Lagerpeton, and I at first assumed that the specimen was an immature individual of that genus. The foot structure, however, shows that we are dealing with a discrete form of smaller size. As in Lagerpeton, the limbs are long and slender; the head of the femur is set off sharply from the shaft and the trochanteric ridge is well developed. As in Lagerpeton the lower leg is longer than the femur, the two femora, as preserved, having lengths of 38 and 39 mm, the associated tibiae 47 and approximately 48 mm. 1971 LAGERPETON AND LAGOSUCHUS Figure 2. Hind foot of Lagosuchus, in ventral view. Composite of type and a specimen in the Instituto Lillo. 1 1/2 X size of holotype. The proximal tarsal region is poorly preserved in the type. Two well-developed distal tarsals are present. Of the right foot, only metapodials, incomplete distally, are visible. The left foot is well preserved as regards the metatarsals and a portion of the pha- langes. The digits are elongate and slender; metapodials II-IV are, as preserved, parallel to one another and closely apposed. Digit I, with a metapodial length of 15 mm, is short; metapodials II-IV have lengths of 23, 26, and 25 mm; metapodial V, broad at its head, tapers, as preserved, to a point at 11 mm. The two pha- langes appropriate to digit I are present, as are single proximal pha- langes articulated with metatarsals II and III, and several disarticu- lated phalanges (one a clawed ungual). Scattered about the slab are disconnected series of vertebrae with average lengths of centra of 7.5 mm in the best preserved 6 RREVIORA No. 378 region. There are, further, remains of pectoral limbs. In the better preserved of these the humerus has a length of 26 mm, the radius 16 mm. As one might expect from the nature of the long hind legs, the front limbs thus appear to be much reduced in size, the combined length of the long bones of the "arm" being less than half that of the corresponding hind leg elements. A specimen in the Instituto Lillo collection includes much of the posterior part of a skeleton that is closely comparable to the type of Lagosiichus except for somewhat larger size. Much of both hind legs is preserved in articulation. Right and left femora measure 55 and 56 mm in length, the tibiae 72 and 70 mm. As in Lagerpeton, astragalus and calcaneum are united and closely applied to tibia and fibula. Here, however, the line of suture between the two proximal tarsals is still visible, and a small spur of bone (not seen in Lagerpeton) projects backward from the fibular edge of the astragalo-calcaneum. There is no proximal extension of the astragalo-calcaneum of the sort seen in Lagerpeton. The two distal tarsals are present lying above the metatarsal heads. The feet are incompletely preserved, but except for somewhat larger size, they compare well with those of the type. On the right foot metatarsals II and IV have lengths of 34 and 38 mm, respec- tively; lengths of metatarsals I-IV on the left foot are: 23, 35, 39, and 38 mm. No phalanges are present on metatarsal I, but two phalanges, with lengths of 9 and 7 mm are present on the second digit, three phalanges with lengths of 12, 8, and 5 mm on digit III, and four phalanges with lengths of 6, 5, 4, and 4 mm on digit IV. Metatarsal V, broad at the base and tapering distally, is incomplete, with a length as preserved of 10 mm. In Figure 2 the foot is restored as a composite of the type and the Instituto Lillo specimen just described, with allowance made for the difference in size of the two specimens. The more distal phalanges are restored on digits II-IV. As restored, digital lengths of the Instituto Lillo specimen are, from digit I outward: 36, 56, 69, and 62 mm. Here, in contrast to Lagerpeton, there has de- veloped the "typical" archosaur foot, with toe I relatively short, toe V reduced, and toe III the longest of the II-III-IV series. Associated are remains of the pelvic girdle. The acetabula (into which the femoral heads were inserted when the specimen was recovered) are small and deep, strongly overhung dorsally by 1971 LAGERPETON AND LAGOSUCHUS 7 the iliac rim. The acetabular construction plus the nature of the femoral head indicate rather surely that the femur moved in a fore-and-aft plane close to the body. The ilium (Fig. 3) is of peculiar structure. Above the acetabulum it constricts to a rela- tively narrow neck. Here there projects anterolaterally a short but stout, blunt-ended process of a sort otherwise unknown to me; it is possible that it afforded an origin for an iliofemoralis externus muscle as in the case of a somewhat similar process in some ornithischians (cf. Romer, 1927) and Poposaurus (Colbert, Figure 3. Right ilium and head of femur of Lagosiichiis: external process on ilium restored from left side. From a specimen in the Institute Lillo. X 4/3. 1961). Above the "neck," the ihum expands to form a short blade. In all "normal" thecodonts the iliac blade is a simple single vertical structure; here, however, there lies, internal to the normal external blade, a broad trough, comparable to that seen in ophia- codonts, presumably for accommodation of dorsal axial muscles. Shallow posteriorly, this trough deepens and broadens anteriorly. Medial to this trough is a second iliac blade, tilted somewhat medially, to the inner surface of which the sacral ribs pre- sumably articulated. Lateral and medial blades join anteriorly, closing the dorsal trough at this end. I know of such an iliac "trough" structure in only one other archosaur. In Hesperosuchus, Colbert (1952) found in the pelvic region an element which should have been an ihum but which, because of its pecuHar nature, he concluded could not be that element. Its main peculiarity is its possession of a dorsal "trough" comparable to that of Lagositchus. Parts of pubis and ischium are present in this specimen. It is 8 BREVIORA No. 378 clear that both pubis and ischium take part in the acetabulum, but in default of a better preserved specimen I refrain from dis- cussion of this portion of the girdle. Found close to the limbs and pelvic remains were two series of vertebrae. One appears to include sacrals, followed by nine prox- imal caudals. The mean length of these caudal centra is approx- imately 7 mm. A first chevron is present between caudals three and four; its length is 13.5 mm; those following decrease in length, the fourth of the series measuring 10 mm. A second series of vertebrae includes 30 segments, apparently running to the tip of the tail. The first dozen, poorly preserved, appear to have a mean length of centra of about 7 mm; more posteriorly, the length increases to about 1 cm per segment. Chevrons, gradually decreas- ing in length to a nubbin, are present, as preserved, to a position between the tenth and eleventh from the end of the series. A further Instituto Lillo specimen that may pertain to Lago- suchus includes an imperfect pelvis and much of the hind legs. As in both genera described in the present paper, the tibia is longer than the femur, with measurements of the right leg elements of 48 and 42 mm, respectively. As in Lagosuchus, metatarsals II-IV are subequal in length, with measurements of 24, 28, and 27 mm for these metapodials in the left foot. Possibly associated is a fragment of maxilla or dentary bearing small teeth spaced a bit more than a millimeter apart. A number of posterior dorsal verte- brae are present, with an average central length of 5.5 mm and a height of 7.5 to 8 mm. DISCUSSION So distinctive are the hind legs which are the major preserved portions of the two forms described above that formal systematic description of them seems justified despite the paucity of further associated anatomical structures. Greater length of tibia than of femur is generally regarded as associated with speed. Greater length of lower leg than thigh is present in relatively few cases among even presumably bipedal archosaurs — a few thecodonts, such as Scleromochlus, small coelurosaurs, and to a slight degree in some camosaurs. The sharply inturned head of the femur indi- cates that the hind legs were carried close to and beneath the trunk and the suggestion that we are dealing with a biped is in- creased by the shortness of the front legs of Lagosuchus. 1971 LAGERPETON AND LAGOSUCHUS 9 It is surprising to find so advanced a type of limb at such an early stage of the Triassic (probably Anisian in terms of the marine series). In the preceding Cynognathiis stage of the "Eotriassic," thecodonts more advanced than protcrosuchians such as Chasma- tosaurus and Erythrosuchus were represented only by Euparkeria, as recently ably described by Ewer (1965). Euparkeria was ad- vancing toward a bipedal habitus, but was still relatively primitive. One would have expected that, in the Anisian, pseudosuchians would not have advanced far beyond this level, and that forms with limbs of such an advanced sort as seen in the two types just discussed should have been characteristic only of a much later stage of the Triassic. Obviously, as these forms indicate, this assumption is incorrect. It would seem that advance and radiation among thecodonts occurred rapidly in early Triassic times; beliefs to the contrary were due to our lack of knowledge of middle Triassic faunas; study of South American forms is now bringing such faunas into the picture. As was first clearly brought out by Krebs (1963), two distinct types of tarsal joints developed among thecodonts. In one, re- tained by crocodilians, the main joint between lower leg and foot lay between astragalus and calcaneum. The second type of joint is that found in dinosaurs, in which the proximal tarsals were functionally combined with the lower leg, the distal tarsals united with the foot. Both forms here described are of the second type. However, the presence of a small spur on the calcaneal region of Lagosuchus suggests the possibility that a transition from one type to the other may have been possible. Fusion of the two proximal elements is a condition rare even in "advanced" dinosaurs. The development of a proximal flange from the astragalo-calcaneum, seen in Lagerpeton, is a condition found in various theropods, but in these forms the flange is developed on the extensor rather than the flexor surface of the lower leg. With the marked expansion of our knowledge of thecodonts that is currently taking place, classification of the group can be at best but a provisional matter, and an attempt to place the two genera just discussed in a "solid" systematic position is unjustified. It is certain that the two have no association with the series of forms which appear to have crocodilian relationships, and the tar- sal construction suggests that the two are connected in some fashion with a radiation leading toward the dinosaurs and, par- ticularly, toward the coelurosaurian group of the Saurischia. 10 BREVIORA No. 378 Whether either of the two, however, can be regarded as close to the direct line leading to such dinosaurs is doubtful. Lagerpeton is, on the one hand, advanced in tarsal construction and, on the other, primitive or specialized in the matter of relative length of digit IV. Lagosuchus is more orthodox in digital construction, but precocious in astragalo-calcanear fusion, and specialized in iliac construction. There is a classic story of the man who "mounted his horse and galloped off in all directions." The history of thecodonts, we are coming to realize, seems to have followed this pattern. We have as yet no clues as to the course followed toward birds or pterosaurs. Various thecodonts seem to have trended toward a crocodilian type of organization. The ornithischian pattern is so distinctive that at the present we can reasonably consider the Ornithischia as definitely monophyletic in origin. This is not the case with the Saurischia. The presence in the later Triassic of apparent sauropod ancestors of seemingly primitive quadrupedal nature suggests a polyphyletic origin of that order. What of the "prosauropods" of the late Triassic, the coelurosaurs, the advanced theropods of the Jurassic and Cretaceous? Quite possibly the Saurischia are a polyphyletic group that took origin from a "mixed grill" of thecodonts — a varied assemblage amongst which the two forms here described are to be included. I am indebted to the National Science Foundation, grants GB-2454 and GB-8171, for aid in the collection, preparation, and publication of the La Plata-Harvard materials, and I am grate- ful to Sr. Bonaparte for the privilege of studying the Chan ares thecodont material which he has collected. REFERENCES CITED Colbert, E. H. 1952. A pseudosuchian reptile from Arizona. Bull. American Mus. Nat. Hist.. 99: 561-592. . 1961. The Triassic reptile, Poposaiints. Fieldiana. Geol- ogy, 14: 59-78. Ewer, R. F. 1965. The anatomy of the thecodont reptile Eiiparkeria capensis Broom. Philos. Trans. Roy. Soc. London, Ser. B, 248: 379-435. Krebs, B. 1963. Bau imd Funktion des Tarsus eines Pseudosuchiers aus der Trias des Monte San Giorgio (Kanton Tessin, Schweiz). Palaont. Z., 37: 88-95. Romer, a. S. 1927. The development of the thigh musculature of tlic chick. Jour. Morph. Physiol., 43: 347-385. BREVIORA Museum of Comparative Zoology CAMBRrocE, Mass. 15 June, 1971 Number 379 THE CHANARES (ARGENTINA) TRIASSIC REPTILE FAUNA XI. TWO NEW LONG-SNOUTED THECODONTS, CHANARESUCHUS AND GUALOSUCHUS Alfred Sherwood Romer Abstract. A description is given of the skull and jaws of two new thecodonts, Chanaresuchiis bonapartei and Gualosuchus reigi. from the early Middle Triassic Chanares Formation of Argentina. The two forms here described plus Cerritosaurus and Proterochampsa are closely related and may be included in the single family Proterochampsidae. Their structure is in many regards so primitive that this family should be in- cluded in the thecodont suborder Proterosuchia. Suggested relationships of members of the family to crocodilians and phytosaurs are discussed. INTRODUCTION Commonest of all thecodont remains so far discovered in the Chaiiares Formation are those of Chanaresuchus bonapartei and Gualosuchus reigi. Considerable material, particularly of Cha- naresuchus, was found by the La Plata-Harvard expedition of 1964-65, and further specimens, which Sr. Bonaparte has been kind enough to let me study, are in the collections of the Instituto Lillo of Tucuman. In the present paper I will confine myself to description of the skulls and jaws of these two closely related forms; I hope shortly to give an account of the postcranial skel- eton of Chanaresuchus. I gratefully acknowledge grants from the National Science Foundation that covered most of the expenses of collection and preparation of the materials, and for aid in publication. Chanaresuchus bonapartei, gen. et sp. nov. Holotype. La Plata Museum 1964-XI-14-12 (field no. 47), a skull and jaws together with postcranial remains. Collected 2 BREVIORA No. 379 from the Chanares Formation, in La Rioja Province, about half a mile southeast of a volcanic plug in the valley of the south fork of the Chanares River. Combined generic and specific characters. A thecodont of modest size (largest known skull about 260 mm in length). Skull long and low; broad posteriorly; slithke external nares placed close together dorsally some distance back of tip of snout; antorbital vacuity small; postfrontal absent; no pineal opening; parietals swing sharply outward posteriorly, above superior tem- poral fenestrae, toward meeting with squamosals; suspensorium far back of occiput, and lateral fenestra hence elongate antero- posteriorly. Very elongate choanae partially covered below by a secondary palate. Pterygoid and palatine toothed. Basal articu- lation of braincase and palate movable. A long if narrow inter- pterygoid vacuity, exposing a slender parasphenoid rostrum. The generic name refers to the Chanares River and Forma- tion. The specific name is given in honor of Sr. Jose Bonaparte, able collector and describer of Triassic reptiles. Description (Figs. 1-3). Considerable skull material of Chanaresuchus is available. Most notable in the La Plata-Harvard collection, in addition to the holotype, are MCZ 4037 (field no. 154), which includes a large skull and jaws, and MCZ 4039 (field no. 81), the well-preserved left half of a small skull. Skull lengths of these three specimens (measured to the quadrate) are, respectively, 211 mm, 260 mm, and 155 mm. MCZ 4036 con- sists of a slab containing, as well as considerable postcranial material, two rather poor skulls of about the size of the holo- type. The dermal bones of the skull are ornamented, particularly on the skull table, with ridges radiating from centers of ossification. The degree of sculpturing appears to vary with size and presumed age; it is little developed on MCZ 4039, somewhat more prom- inent on the type, and the largest skull, MCZ 4037, is quite rugose. The skull is long, low, slender-snouted, broad posteriorly. Snout elongation is clearly shown by the length of premaxillae, nasals, and frontals, and by proportionate measurements. If, for example, the anteroposterior diameter of the orbit be taken as a base, the facial length, anterior to the orbit, is about AVi times the length of the orbit itself, whereas in other thecodonts 1971 CHANARESUCHUS AND GUALOSUCHUS 3 (phytosaurs excepted) this measurement is typically no more than two or three times the orbital diameter. There is, further, a considerable degree of postorbital length. From the post- orbital bar to the tip of the quadrate, the distance is about 21/2 times the orbital diameter, whereas in pseudosuchians this dis- tance is almost invariably considerably less, and is sometimes not even equal to orbital diameter. It is difficult to select a standard against which skull height may be objectively measured, but it may be noted that even in the postfacial region of the skull, where flattening is less pronounced than in the snout, the skull height is not sufficient to give space for the eyes in the lateral wall and the orbits are, in consequence, incised into the skuU roof. The tip of the snout is sUghtly decurved. Dorsal and lateral surfaces of the skull are clearly marked off from one another. The dorsal surface is nearly flat, although rising along the length of the snout to then become horizontal in plane along the length of the skull table. On either side a well-marked ridge, separating lateral and dorsal surfaces, develops at the level of the nares and continues back to the anterodorsal angle of the orbit, where there is a modestly developed lateral projection. A similar pro- jecting angle is present at the posterior margin of the orbit, and a clear-cut angle separating roof from "cheek" continues along the bar separating the temporal fenestrae. The external nares are elongate slitlike structures, dorsally placed and close together, separated only by a narrow bony bar. In contrast to the subterminal position seen in many thecodonts, they are placed well back of the snout tip; grooves leading into the narial openings are present anteriorly and posteriorly. Deep to the external nares, and separated from the narial margins at all points, is seen a sheet of bone within which is found, on either side, an opening of considerably smaller size than the narial opening. At first glance one would assume that these openings are the choanae; however, as discussed below, they are apparently anterior palatal foramina for the vomeronasal organs, and this bony sheet forms a short secondary palate. This sec- ondary palate is separated from the bone rimming the narial margin for only a short distance on the anterior and lateral margins; medially, however, the bony sheets of the two sides appear to meet one another below and free from the bar sepa- rating the two nares, and the opening posteriorly between the BREVIORA No. 379 narial margin and the secondary palate leads into a short naso- palatine duct. The antorbital fenestra is a small triangular opening, the apex of the triangle lying anteriorly, at about one-third the distance from orbit to snout tip, the posterior base separated from the orbit by a narrow bony bar. The orbits are large (as, presumably were the eyes) and are subcircular in shape; on the lateral sur- face they occupy nearly the whole height of the skull, leaving but a narrow bar of bone between them and the lower skull margin. Dorsally their semicircular margins cut far into the sides of the skull table. The upper rim is slightly elevated; as mentioned, there are pronounced projections anteriorly and pos- teriorly at the junction of dorsal and lateral portions of the orbital boundaries. There is no parietal foramen. The superior temporal fenestra is well developed, facing directly dorsally, and is triangular in Figure 1. Chanaresuchus bonapartei, skull in dorsal view. This and Figs. 2-5 are based on the holotype, with additions from other specimens. Abbreviations for Figs. 1-7: an, angular; ar, articular; bo, basioccipital; c, coronoid; d, dentary; ec, ectopterygoid; /, frontal; ;, jugal; /, lacrimal; m, maxilla; n, nasal; op, opisthotic; p, parietal; part, prearticular; pi, pala- tine; pm, premaxilla; pa, postorbital; pp, postparietal; prf, prefrontal; ps, parasphenoid; pt, pterygoid; q, quadrate; qj, quadratojugal; san, sur- angular; sac, supraoccipital; sp, splenial; sq, squamosal; v, vomer. 1971 CHANARESUCHUS AND GUALOSUCHUS 5 shape, with an anterior base and a posterior apex. The lateral temporal opening is, as noted above, elongate anteroposteriorly, in contrast to its much shorter span in most archosaurs; its pos- terior margin is somewhat convex in outline, suggestive of the initiation of the V-shape of this border in many archosaurs. In strong contrast with most archosaurs, the suspensorial region slants far back ventrally, the jaw articulation lying some distance back of the level of the occiput. The back margin of the sus- pensorial region is somewhat concave; this, plus a slight dorsal spur, shows the initiation of the type of otic notch characteristic of many more advanced archosaurs. The occiput slants diago- nally downward and backward from the posterior margin of the narrow medial portion of the skull table. Of individual dermal roofing elements, the premaxillae are elongate, in conformity with general snout elongation. They form most of the boundaries of the external nares. The pre- maxilla sends a long and well-developed process backward ex- ternal to the naris, excluding the maxilla from the narial border and terminating in a slender tip between maxilla and nasal. Medially, conjoined slender processes from the two premaxillae form a narrow bar separating the nares and extending back some distance beyond tht narial region to taper between and below the anterior tips of nasal processes that meet them. There is no evidence of a separate septomaxilla, reported as existing in this region in phytosaurs. Like the premaxillae, the nasals are elon- gate They enter into the posterior border of the external nares for a short distance and extend far backward to meet the frontals in a zig-zag transverse suture; laterally there is a long line of suture with the maxillae and a short contact with the prefrontals. The frontals, as indicated by the development of their radiating surface ridges, are somewhat elongated anteriorly. Laterally they have a long contact with the prefrontals anteriorly and a short posterior contact with the postorbitals; they enter broadly into the dorsal rims of the orbits. The parietals are short anteropos- teriorly, as regards their development on the skull table. They have a transverse suture with the frontals in the middle portion of the roof; lateral to this, the line of suture with the frontals retreats sharply posteriorly, then turns forward again, allowing the parietals contact with the postorbitals. This lateral extension of the parietal forms the anterior border of the superior temporal 6 BREVIORA No. 379 fenestra. The posterior end of the skull roof is narrow; on either side, however, each parietal sends, posterolaterally, a long process to meet and overlap the medial surface of the squamosal behind the superior temporal fenestra. Narrow dorsally but broadening below, the medial surface of this process forms the upper lateral boundary of the occipital plate and laterally forms the medial boundary of the superior temporal fenestra. This boundary is sharply marked off dorsally; more ventrally, how- ever, the parietal slants outward to meet, obviously, the prootic area of the otic region of the braincase. Laterally, the premaxilla continues back some distance along the margin of the upper jaw before being succeeded by the elon- gate maxilla. The latter element gradually increases in depth, posteriorly, to occupy the entire height of the side of the snout anterior to the antorbital fenestra. The maxilla extends upward beyond the ridge separating lateral and dorsal skull surfaces to form a lateral strip of the dorsal surface for, roughly, the length of the nasal. When the antorbital opening is reached, the maxilla extends but a short distance backward above this opening. Below, however, it forms the fenestral border for most of its length, to be finally succeeded by the jugal. Posteriorly the maxilla tapers downward in depth to terminate at about the level of the posterior border of the orbit. The lacrimal forms most of the bar sepa- rating orbit and antorbital fenestra and (although the line of suture is none too clear) appears to extend forward above this fenestra for most of its length on the lateral skull surface. The prefrontal forms a triangular area on the skull roof, bounded medially by nasal and frontal, posterolaterally by the orbital margin and laterally by the ridge separating dorsal and lateral margins of the snout. It forms a prominent projection over the anterodorsal margin of the orbit and sends a process. Figure 2. Chanaresiichus bonapartei, type skull in side view. X 4/9. 1971 CHANARESUCHUS AND GUALOSUCHUS 7 reinforcing the lacrimal, part way down the bar separating orbit and antorbital fenestra. There is no postfrontal clement. The postorbital is well developed. Its center of ossification is a prom- inent elevation at the posterodorsal corner of the orbit, whence it extends in three directions. Medially, it extends over the skull roof surface to meet the frontal anteriorly and, more posteriorly, the parietal. This process forms anteriorly part of the orbital rim, and posteriorly a small portion of the boundary of the upper temporal fenestra. A posterior flange extends backward to form much of the upper border of the lateral temporal fenestra, grad- ually tapering backward to give place to the squamosal. Ven- trally the postorbital extends downward to form the upper part of the bar separating orbit and lateral temporal fenestra. The jugal, in normal fashion, forms the lower margin of the orbit and much of the lower margin of the lateral temporal fenestra. Anteriorly it reaches the posteroventral margin of the antorbital fenestra, and it forms the lower part of the postorbital bar. The squamosal, in a fashion somewhat analogous to the post- orbital, sends out three processes from a center located dorso- posterior to the lateral temporal fenestra. An anteromedial process runs forward to join laterally the process of the parietal, which forms the medial wall of the superior temporal fenestra. A long and strong process runs anteriorly to meet the postorbital in a long diagonal suture and to form, in conjunction with that element, the bar lying between the lateral margin of the upper temporal fenestra and the upper boundary of the lateral temporal fenestra. A broad but thin process runs ventrally to form the posterior boundary of the lateral temporal fenestra; this meets the quadratojugal ventrally and posteriorly has a long suture with the quadrate. The squamosal, as in thecodonts generally, sends a short spur backward above the head of the quadrate. The quadratojugal forms the posterior part of the bar below the lateral temporal fenestra and then curves sharply upward to meet the squamosal and to form the lower part of the posterior mar- gin of this fenestra. Posteriorly, at the junction of its two limbs, the quadratojugal is thickened and is in intimate contact with the quadrate. As noted above, the occipital surface of the skull slants back- ward as well as downward. The posterior rami of the parietals form vertical walls that bound the occiput on either side. At the 8 BREVIORA No. 379 apex of the occipital surface is a small triangular postparietal, which faces backward as much as upward and plays no part in the dorsal surface of the skull. Below this is a large supraoc- cipital, essentially triangular except for truncation of its narrow anterodorsal extremity. On either side this plate articulates with the proximal part of the posterior processes of the parietals. Dorsally the bone has a prominent medial ridge; ventrally this element forms the upper margin of the foramen magnum. The lateral margins of the foramen are formed by the exoccipitals, which are fused with the opisthotics to form long and slender paroccipital processes extending laterally to abut against the inner surfaces of the squamosals dorsally; this articulation is a loose one, with the presence of kineticism. Possibly a small post- temporal fenestra may have been present above the paroccipital process; if present, however, it cannot have been more than a slender slit. The exoccipital forms a dorsolateral fraction of the occipital condyle; the condyle is, however, mainly formed by the basioccipital. The condyle is circular in outline and essentially convex with, however, a slight median excavation, presumably for the anterior termination of the notochord. I regret that the condition of the material is such that I have been unable to make out the vagus and hypoglossal foramina nor the lateral surface of the braincase in the otic region. I have not found a stapes, and can say nothing regarding the possible presence of epiptery- goid or laterosphenoid. In correlation with snout elongation, the anterior part of the palate is much modified. Medially a pair of stout processes formed by the premaxillae extend backward in the midline; be- hind them, the central area is continued by narrow paired vomers, bearing a row of small teeth. Posteriorly the vomers diverge somewhat to accommodate between them the anterior tips of the pterygoids. At their posterior ends the vomers are in contact with the palatines, the anterior ends of which have a concave border, meeting the vomers medially and the elongate maxillae laterally. In "typical" tetrapods the choanae are situated well forward, as rounded or oval openings, bounded anteriorly by the pre- maxillae, laterally by the maxillae, posteriorly by the palatines and medially by the vomers. If we look for the equivalent of the normal choanae here, we find a pair of long and narrow areas 1971 CHANARESUCHUS AND GUALOSUCHUS Figure 3. Chanaresuchiis bonapartei, type skull in palatal view. X 4/9. reaching about half the length of the skull, from the premaxillary region to the position of the "incised" anterior ends of the palatines. This elongation of the original choanal region is ob- viously related to snout elongation. From the palatines for some distance forward there are open choanae. But, farther forward, there is a sheet of bone apparently pertaining to the maxillae, which extends medially across the choanal areas from the max- illae to gain a contact with the posterior portions of the stout ventral processes of the premaxillae and the edges of the vomers. More anteriorly this sheet is broken by oval openings, already mentioned in connection with the skull roof. It is reasonable to believe that these openings are anterior palatine foramina for access to the mouth cavity of vomeronasal organs (Jacobson's organs), and that the transverse sheet of bone between these openings and the true choanae is a definite, if short, secondary palate formed for facilitation of underwater breathing in this long-snouted animal. Much more normal and primitive in construction is the pos- terior portion of the palatal surface. Anterior to the occipital condyle, there are projecting basisphenoidal tubera, connected by an incised transverse fine, concave posteriorly, which pre- sumably marked the anterior limit of subvertebral musculature. 10 BREVIORA No. 379 Lines extending posteromedially from the tubera suggest an ex- tension of parasphenoidal dermal ossification backward over part of the basioccipital area. Anterior to the tubera, the combined basisphenoid-parasphenoid contracts somewhat in width, then expands again to the projecting spherical basal articular proc- esses, on which it is obvious the pterygoids had considerable freedom of movement. Paired foramina for the carotid arteries are present between these processes. A slender parasphenoid rostrum extends far forward along the midline of the interptery- goid vacuity. Of the elements of the posterior part of the palate, the palatine occupies an area between the posterior end of the choana anteriorly and the palatine fenestra posteriorly, and be- tween the maxilla laterally and the pterygoid medially. It bears ventrally an anteroposterior row of small teeth. The pterygoids are highly developed. Their slim anterior ends extend far forward between the vomers. A short distance back of their anterior termination the two pterygoids diverge slightly, so that a long but narrow interpterygoid vacuity is developed. Along the medial border each pterygoid carries a long row of small teeth; this series terminates just above the region of the basal articula- tion with the braincase. A second row of small teeth is present on a distinct ridge that slants diagonally backward and medially from a point near the posterior end of the palatine to terminate somewhat short of the basal articular region. The palatal fenestra is of considerable extent, bounded laterally by the jugal and incised medially into the posterior part of the palatine and the adjacent portion of the pterygoid. Back of the fenestra the palatal ramus of the pterygoid expands widely laterally. Part of this expansion is presumably formed by an ectopterygoid, but I have not been able to detect a pterygoid-ectopterygoid suture. Anteriorly the lateral margin of this expansion is sutured to the jugal; posteriorly there is a constriction in width, and the bone slants ventrally to form a stout transverse pterygoid flange, broadening distally. The palatal portion of the pterygoid terminates medially and pos- teriorly in a short spur beneath which is the socket for reception of the articular process of the basisphenoid. Lateral to this area, there arises a typical quadrate flange of the pterygoid, of modest height, which posteriorly meets the quadrate. 97: CHANARESUCHUS AND GUALOSUCHUS The quadrate bone is well developed. Its posterior end is a thickened articular area, widened transversely and convex at both internal and external ends; it appears that the quadratojugal (as often) takes part to some degree in the lateral condyle. The main shaft of the quadrate extends upward, to terminate in a recess on the under surface of the squamosal at and close to its posterior spur. This ascending ramus of the quadrate is broad ventrally, gradually contracting in width dorsally, and has a concave posterior margin. It faces as much posteriorly as later- ally, at an angle to the adjacent areas of the quadratojugal and descending ramus of the squamosal. As generally, a foramen is present on this surface between quadrate and quadratojugal. The ascending ramus presents a broad, forward-slanting, medial sur- face that is covered anteriorly by the quadrate ramus of the pterygoid. In correlation with skull length, the jaw is long and slender anteriorly (Figs. 4, 5). The symphysis is poorly represented in available material but was obviously weak and formed mainly — perhaps entirely — by the dentary. For much of the length of the muzzle the dentary forms almost the entire outer surface of the ramus — a surface that slants markedly inward below, rather than being directed vertically downward. Posteriorly the bone forks, the two branches enclosing between them the anterior end of the long external mandibular fenestra. The upper branch ex- tends along the upper margin of the ramus to the end of the tooth Figure 4. Cluijuircsiicliiis bonapartei, jaw of type in lateroventral view. X 4/9. row, where it is replaced by the surangular; the lower branch extends backward below the fenestra for some distance, applied to the outer surface of the angular. The splenial is exposed at the lower edge of the external surface. The posterior portion of the external surface is made up almost entirely by the surangular and angular. The former bone runs backward along the curving 12 BREVIORA No. 379 upper margin of the ramus, whence a dorsally facing area of the surangular extends inward above the lateral border of the man- dibular fossa. The angular extends backward below the lateral mandibular fenestra, the two elements meeting at the posterior end of the fenestra, whence a ridge, with which the suture be- tween the two elements is associated, runs posteriorly. The suture is indistinct posteriorly but the conjoined angular and surangular extend backward nearly to the posterior end of the jaw, sheathing the articular laterally. On the inner surface the splenial lines the jaw for nearly its full height for most of the length of the tooth row. Beyond this point its upper margin slants gradually downward to the termination of the bone well posteriorly. It is succeeded posteriorly on the inner surface by the prearticular, which, narrow anteriorly, follows the splenial downward and backward. The prearticular forms the inner bor- der of the mandibular fossa, the surangular the outer border. A coronoid is present, but seen only as a disarticulated element. I have restored it, in Figure 5, with some doubt, in its probable position. It appears to have extended forward as a thin sliver Figure 5. Chanaresuchus bonapartei, jaw of type in medial view. X 4/9. of bone between the dentary and splenial and more posteriorly appears to have been applied to the inner surface of the surangu- lar. Posteriorly, the surangular thickens on its inner surface to form the back margin of the mandibular fossa. The prearticular fuses posteriorly with the articular. This stout element appears to have been but loosely attached to the surangular and angular, since it has separated from them in the two available specimens with jaws in the La Plata-Harvard collection. The articular oc- cupies the full height of the inner surface of the jaw at its pos- terior end; it is braced anteriorly by the medial extension of the surangular mentioned above, and is nearly completely covered externally by the thin posterior extension of the conjoined angular 1971 CHANARESUCHUS AND GUALOSUCHUS 13 plus surangular. There is little development of a retroarticular process, but the bone extends somewhat ventroposteriorly. The articular surface of the articular, as of the quadrate, is trans- versely broadened, with median and lateral concavities corre- sponding to the pair of convexities on the quadrate. There is a pronounced process developed on the medial surface just below the level of the articular surface. The marginal dentition is but imperfectly preserved in avail- able specimens. The teeth are of a common archosaurian type, somewhat compressed mediolaterally, hence with an oval section, sharp-pointed, and backwardly curved distally. The insertion is protothecodont. There appear to have been six teeth, the last small, on the elongate premaxilla, about 18 on the maxilla, and about the same number on the dentary. As in many reptiles, there appears to have been a high degree of replacement of teeth in essentially alternating fashion, so that for much of the jaw length alternate teeth are well developed, those between barely erupted or represented by empty sockets. GuALOSUCHUS REiGi, gen. et sp. nov. Holotype. La Plata Museum 1964-XI-14-13 (field no. 75), including most of the right half of skull and jaws and a limited amount of postcranial material. Collected from the Chanares Formation, from the valley of the north fork of the Chanares River, La Rioja Province, about five miles east of the point where this stream debouches into the Talampaya plain. Combined generic and specific characters. Similar to Cha- naresuchus in nearly all regards, but larger; posterior portion of skull deeper but with a narrower skull table than in Chanare- suchiis; orbit taller than in that genus and less incised into skull roof; parietals extend in paired fashion farther back on skull table than in Chanaresuchus and diverge less sharply posteriorly toward the squamosals; superior temporal openings proportion- ately narrow and more elongate. The generic name, by analogy with Chanaresuchus, refers to the Gualo River, which, with the Chanares, drains most of the known area of exposure of the Chanares Formation. The specific name is in honor of Sr. Osvaldo A. Reig, an active student of archosaur evolution. 14 BREVIORA No. 379 Gualosuchus is represented in the Harvard-La Plata collections only by the holotype, which includes the dermal bones of the right side of the skull roof, the pterygoids, right palatine, most of the right lower jaw and a number of postcranial elements. Further materials, including a second skull, are present in the Instituto Lillo collections. The length of the holotype skull (Figs. 6, 7), measured to the quadrate, is about 325 mm; the Instituto Lillo skull, presumably that of a young individual, is much smaller. Quite probably the holotype represents a "mature" speci- men; it is nearly a quarter larger than the largest known skull of Chaiuiresuchus. The skull of the holotype is highly rugose, pre- sumably in correlation with large size; this feature has made identification of sutures difficult. In every major structural feature the Gualosuchus skull closely resembles that of Chanaresuchus. In consequence, detailed de- scription is unnecessary; mention need be made only of points in which the two genera differ. As in Chanaresuchus, the skull is long and low; the proportions of snout length to total skull length are much the same in the two genera, and both have the same anteroposterior elongation of the lateral temporal fenestra. The skull of Gualosuchus, however, is much less depressed pos- teriorly than that of Chanaresuchus, and less broad. In Cha- naresuchus, for example, the height of the skull at the orbit is little more than 15 per cent the skull length; in Gualosuchus about 22 per cent, and the comparable figures at midlength of the lateral temporal fenestra are 17 per cent and 24 per cent. In relation to greater depth at the orbit, this opening, which is subcircular in lateral view in Chanaresuchus, is taller and sub- quadrate in shape in Gualosuchus and is much less incised into the skull roof. Part of the contrast in depth is due to the greater depth of the maxilla and jugal beneath the antorbital vacuity, orbit, and the anterior part of the lateral fenestra. The differ- ences in breadth between the two genera relate mainly to differ- ences in width of the skuU table. In Chanaresuchus the width across the postorbital-squamosal bars bordering the skull table on either side, is nearly a third the measurement of skull length; in the Gualosuchus type this width is but a quarter the skull length. This difference in table proportions results in contrasts in the pattern of the posterior part of the skull roof. In Chanaresuchus the superior temporal fenestrae are relatively short and broad 97: CHANARESUCHUS AND GUALOSUCHUS 15 and slant outward posteriorly; in Gualosuchus these openings are relatively long and narrow and lie on a direcdy anteroposterior line. In Chanaresuchus the two parietals are united on the skull table for only a short distance before they diverge sharply to extend outward and backward to meet the squamosals; in Gualo- suchus, in contrast, the parietals extend backward in contact with one another for a considerable distance before diverging, at a lesser angle, toward the squamosals. Figure 6. Gualosuchus reigi, holotype skull in dorsal view. X 1/3. Figure 7. Gualosuchus reigi, holotype skull in lateral view. X 1/3. 16 BREVIORA No. 379 The disarticulated pterygoids and right palatine are very simi- lar in construction to those of Chanaresuchus. Most of the right ramus of the lower jaw is preserved in the type. It had been somewhat weathered before recovery, but in all observable fea- tures it agrees well with the Chanaresuchus jaw. DISCUSSION Chanaresuchus and Gualosuchus are closely related; they ap- pear to be almost identical in all structural features and as far as known differ only in size and in skull proportions, the Gualo- suchus skull being relatively narrower and taller posteriorly. We need not search far to find relatives. It is obvious that Cerrito- saurus described by Price (1946) from the slightly later Santa Maria beds of Brazil is a close relative, as shown by similar skull proportions, slitlike dorsally-placed nostrils, loss of post- frontal element, absence of the parietal foramen, comparable shape of the lateral temporal fenestra, and posterior situation of the jaw articulation. Possibly some of the thecodont material from the Manda beds may pertain to a related type, but this material is too fragmentary to allow positive determination. Quite surely, however, a further South American relative is Proterochampsa (Reig, 1959; Sill, 1967). A number of points of resemblance to Chanaresuchus and Gualosuchus can be seen in the figures of Reig and Sill, and further unpublished observa- tions and study of the Proterochampsa specimens in the fight of our better knowledge of the Chanares forms strongly suggest that the relationship is close indeed. The Proterochampsa skufis are even more flattened than in Chanaresuchus (although this may be due in part to post-mortem crushing) . The skull is sculptured as regards its dermal roofing elements in very rugose fashion, presumably in correlation with the fact that its size is consider- ably greater than that of either of the Chanares forms. This rugose condition makes for difficulty and doubt in the determina- tion of sutures. The skull proportions, with a long slender snout and a broad posterior region, are identical with those in the earlier genera. The outlines of the external nares are imperfectly preserved, but the structure here is apparently the same as in Chanaresuchus and Gualosuchus. The antorbital fenestrae are smafi, as in those genera; the orbit, subcircular in shape as in Chanaresuchus, is strongly incised into the skull table; because 1971 CHANARESUCHUS AND GUALOSUCHUS 17 of the great flattening of the skull, these openings face nearly directly dorsally rather than laterally. The pattern of the posterior part of the skull table, with a pineal opening absent, and the posterolateral extensions of the parietals swinging broadly out- ward, closely resembles that of the Chafiares genera. As in those forms, the lateral temporal fenestra is large and long anteropos- teriorly; the jaw articulation is far back of the occiput, and, as in the earlier genera, there is but a slight projection of the squamosal above the incipient archosaur otic notch. The lower jaw structure, as far as can be made out, is similar to that of the forms here described. The palate is poorly seen, but recent study indicates that the posterior portion of it was quite similar to that of Chanaresuchus and Gualosiichus. As in those genera, the basal articulation was movable and an interpterygoid vacuity present, in which, as in the genera here described, there projected forward a slender parasphenoidal rostrum. A row of denticles was present, as in the Chanares forms, on the palatine as well as denticle rows on the pterygoid. Again, as in these forms, the anterior ends of the palatines were notched for the posterior margins of the choanae. Forward of this point little can be made out regarding palatal structure. Both Reig and Sill restore this area with a small choana and a long secondary palate. This is, however, uncertain and the situation here may well have been much the same as in the Chaiiares forms. In sum, Proterochampsa in all observable features appears to be very similar in cranial structure to Chanaresuchus, Gualo- suchus and Cerhtosaurus; the differences between them are cer- tainly not more than of generic value, and all four may be reasonably grouped within the single family Proterochampsidae, erected by Sill (1967) for the reception of Proterochampsa. The general structure of these four genera is strongly suggestive of amphibious habits, not improbably paralleling those of the later phytosaurs and crocodilians. The postcranial skeleton ap- pears to have the general proportions of crocodilians (although without diagnostic crocodihan characteristics). The slender snout, flattened skull, the trend for a dorsal facing of the orbits and the dorsal position of the nostrils are all suggestive of water- dwelling habits. 18 BREVIORA No. 379 What is the pedigree of these forms? Currently, as regards classification of thecodonts, one tends to sort out a few early and primitive genera as the Proterosuchia, separate off as advanced types the Phytosauria and (in some fashion or other) crocodilian ancestors, and, having done this, "lump" all remaining forms as members of the suborder Pseudosuchia. One's first inclination would be to include the Proterochampsidae in this last general category. Further consideration, however, suggests that the proterochampsids are too primitive structurally to be placed in the Pseudosuchia. In a few regards our forms are advanced or specialized — dorsal position of the nares, loss of the postfrontal and of the parietal foramen, structure of the anterior part of the palate, and an advanced jaw structure. But there are many primitive features — for example, small size of the antorbital opening, posterior position of the suspensorium, long antero- posterior extent of the lateral temporal fenestra, presence of a movable basal articulation, retention of an interpterygoid vacuity and retention of palatal teeth. Typical pseudosuchians are ad- vanced in all these characters. Euparkeria, recently well described by Ewer (1965), which is either reckoned as a primitive pseudo- suchian or as an advanced proterosuchian leading toward the pseudosuchians, is as primitive as the proterochampsids in most of the features listed. But even Euparkeria is more advanced in some features, such as the relatively large antorbital fenestra and, more significantly, shortening of the lateral temporal fenestra and forward movement of the suspensorial region. One is thus tempted to consider a direct origin of the protero- champsids from a proterosuchian ancestor. Charig and Reig (1970) list some 27 structural features that are characteristic of proterosuchians, 16 of which (2-17) pertain to the cranium. In the greater part of these characters, the Proterochampsidae are in agreement with the Proterosuchia. They differ in a few points: (2, part) absence of a postfrontal and of a parietal foramen; (5, part) nonterminal position of external nares; (9) a slight projection of the squamosal back beyond the head of the quadrate (present, however, in Chasmatosaiirus); (13) formation of an incipient secondary palate; (14) the presence or absence of an epipterygoid is unknown. In all other points the Proterochamp- sidae are in full agreement with the Proterosuchia: (2, part) a median postparietal present; (3) well-developed and projecting 1971 CHANARESUCHUS AND GUALOSUCHUS 19 prefrontal; (4) short and broad parietal; (5, part) exclusion of maxilla from naris by premaxilla; (6) moderate size of antorbital vacuity; (7) superior temporal fenestra facing dorsally; (8) no V-shaped lateral temporal fenestra; (10) Uttle development of otic notch; (11) jaw^ articulation well behind condyle; (12) interpterygoid vacuity present; (15) marginal teeth more or less isodont; and, (16) tooth insertion subthecodont. To this long list of primitive characters may be added the presence of a mov- able basipterygoid articulation, and retention of palatal teeth. It seems clear that the Proterochampsidae are of direct protero- suchian derivation and, despite a few advances, may best be classified as a family of the Proterosuchia. Further, one's attention can be immediately directed to Pro- terosuchus [Champsosaurus] as a proterosuchian probably not far removed from the near ancestry of the Proterochampsidae. Reig (1959), Sill (1967), and Walker (1968) have each in turn commented on the similarity of Proterochampsa to ''Clias- matosaurus.'' Apart from the nasal apparatus, little change is needed to transform "Chasmatosaurus" into a proterochampsid — reduction of the downward curvature of the snout, loss of the postfrontal bone, and modification of the posterior part of the lower jaw. Two changes anteriorly are needed — an upward and backward shift of the external nares and, with further elongation of the already slitlike choanae of ''Chasmatosaurus,"" initiation of a secondary palate. The proterochampsids can be reasonably considered to be direct and relatively unmodified descendants of a proterosuchian of "chasmatosaurid" type. Were the Proterochampsidae a sterile line or could they have given rise to more advanced archosaurs of any sort? The two possibilities are the Phytosauria and Crocodilia. Reig and Sill believe Proterochampsa to be an ancestral crocodile; Walker (1968, 1970) denies the crocodilian affinities of Proterochampsa but suggests relationships to phytosaurs, while, on the other hand, he suggests that Cerritosaurus is a crocodile relative. I see little positive evidence to support relationship of any member of the Proterochampsidae to the Crocodilia. As far as I am aware, the postcranial skeleton of Clianaresuchus shows none of the significantly crocodihan features of coracoid, pubis, etc., that are characteristic of Triassic "pre-crocodilians." As regards the skull, Reig calls attention, in addition to the beginning of a 20 BREVIORA No. 379 secondary palate, to the rather crocodihan skull proportions, particularly those of the table. Sill gives a careful and detailed analysis of skull structure, but, apart from the secondary palate, cannot point out any feature in which Proterochampsa approaches crocodilian conditions; at best, it exhibits features that may have been present in the remote ancestors of the Crocodiha, and that are, essentially, those present in generahzed ancestral thecodonts. The posterior portion of the palate, for example, is of an ex- tremely primitive pattern, markedly modified in all nonprotero- suchian thecodonts. Notably primitive is the suspensorial region, with the jaw articulation far to the rear of the occiput and the lateral temporal opening greatly elongated. To attain the croco- dilian condition it seems structurally necessary for this region to pass through the pseudosuchian stage of a short lateral temporal region with a V-shaped posterior boundary, followed by closure of the upper part of the lateral vacuity, a forward shift of the upper end of the quadrate and, finally, downward closure of the squamosal back of the otic notch. There is not, in protero- champsids, the slightest trace of the beginning of this highly important series of structural changes; these forms are not a whit more advanced than the archaic thecodont "'Chasmato- saurusy The posterior part of the proterochampsid jaw, again, is specialized in a noncrocodilian fashion. Finally, the movement of the external nares upward and backward along the skull roof is a structural feature that is not primitive or merely "neutral" in nature, but is in direct contrast to the situation expected in an ancestor of the Crocodilia, in which the nostrils are persistently terminal in position in almost every case. In short, for positive signs of crocodilian relationships of the proterochampsids, we are reduced to the presence of a short secondary palate. It is possible, but difficult, to imagine this structure being expanded and modified to form the elongate secondary palatal structure seen in true crocodilians. It seems more probable, at present, to believe that the development of this structure plus the backward movement of the external nares, represent an attempt, parallel to that of crocodilian ancestors, to improve respiration in a long-snouted amphibious reptile. If we turn from the proterochampsids to a series of later Triassic archosaurs, such as Notochampsa, Erythrochampsa, Protosuchus, and the recently described Orthosuchiis (Nash, 1971 CHANARESUCHUS AND GUALOSUCHUS 21 1968; cf. Walker, 1970), we find a series of forms in which there is little development of a secondary palate, but in which there are numerous positive indications of crocodilian relation- ship, such as the series of crocodilian postcranial characters that arc absent in proterochampsids, progress in the development of the crocodilian type of suspensorial and otic regions, presence of supraorbital bones, fusion of braincase and palate ventrally. There is little indication that these forms are directly derived from such archaic and essentially proterosuchian forms as the proterochampsids; rather, it would seem, the crocodile ancestors advanced from the proterosuchian to the pseudosuchian stage of thecodont development, and then began to specialize in the direction of the Crocodilia. Although Walker (1968), as noted above, suggested that Cerritosaiinis might be related to crocodilian ancestry, he denies this for its relative Proterochampsa, and suggests, in contrast, that this genus might have been ancestral to the phytosaurs. Most of the items listed by him as phytosaurian similarities appear to be of little weight and could be countered by other features wherein Proterochampsa differs from possible "proto-parasuch- ians" (as, for example, in loss of postfrontals in proterochamp- sids). The one seemingly important and suggestive feature is the movement of the nares, as slitlike structures, well back onto the dorsal skull surface. This could well be an initiation of the strong posterior narial trend seen in phytosaurs. But in default of intermediate forms, the gap between such a proterochampsid as Chanaresiichus and a typical phytosaur is so great as to make an assumption of relationship, in the present state of our knowl- edge of thecodonts, little more than an interesting possibility. LITERATURE CITED Charig, a. J., AND O. A. Reig. 1970. The classification of the Protero- suchia. Biol. J. Linn Soc, 2: 125-171. Ewer, R. F. 1965. The anatomy of the thecodont reptile Eiiparkeiia capensis Broom. Phil. Trans. Roy. Soc. London, ser. B, 248: 379- 435. Nash, D. 1968. A crocodile from the Upper Triassic of Lesotho. J. Zool. London, 156: 163-179. Price, L. L 1946. Sobre um novo pseudosuquio do Triassico superior do Rio Grande do Sul. Bol. Serv. Geol. Min. Brasil, 120: 7-38. 22 BREVIORA No. 379 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. Sill, W. D. 1967. Prolerochampsa barrionuevoi and the early evolution of the Crocodilia. Bull. Mus. Comp. Zool., 135: 415-446. Walker, A. D. 1968. Protosuchus, Proterochampsa, and the origin of phytosaurs and crocodiles. Geol. Mag., 105: 1-14. 1970. A revision of the Jurassic reptile Hallopiis victor (Marsh), with remarks on the classification of crocodiles. Phil. Trans. Roy. Soc. London, ser. B, 257: 323-372. ^'^ Harvard MCZ 3 2044 066 302 787 DOES NOT CIRCULATE