Full text of "Breviora"

HARVARD UNIVERSITY 

Library of the 

Museum of 

Comparative Zoology 



B R E V I R A 



useuiii of Comparative Zoology 



HAR\ ARD UNIVERSITY 



Numbers 380^09 



1971-1973 



CAMBRIDGE, MASS., U.S.A. 

1974 



Edited by Penelope Naumann 



BREMORA 
Museum of Comparative Zoology 

CONTENTS 

Numbers 380-409 



1971 

No. 380. The Taxonomic Position of the Genus Petria Semenow 
(Coleoptera: Allecuhdae). By John F. Lawrence. 
9 pp. June 15. 

No. 381. Tetraploidy in a Hybrid Lizard of the Genus Cnemi- 
dophorus (^Teiidae). B\' \\'illiani B. Xeaves. 25 pp. 
December 20. 



1972 

No. 382. Osteology and Description of Thrattidion noctivagus, 
a Minute, New Freshwater Clupeid Fish from Cam- 
eroon, with a Discussion of Pellonulin Relationships. 
By Tyson R. Roberts. 25 pp. February 25. 

No. 383. Two New Genera of Bembidiine Carabid Beetles from 
Australia and South America with Notes on their 
Phylogenetic and Zoogeographic Significance (Co- 
leoptera). B) Terr\ L. Erwin. 19 pp. February 25. 

No. 384. An Attempt to Determine the Systematic Position of 
EUopostoyna ynegaloyyiycter, an Enigmatic Fresh- 
water Fish from Borneo. By Tyson R. Roberts. 
16 pp. February 25. 

No. 385. The Chafiares (Argentina) Triassic Reptile Fauna. 
XII. The Postcranial Skeleton of the Thecodont 
Chayiaresuchus. By .Alfred Sherwood Romer. 21 pp. 
Februarys 25. 



No. 386. A Review of the Australian Species of Elaphrosyron 
and Telostegus, with Notes on Other Genera (Hy- 
menoptera, PompiHdae). By Howard E. Evans. 
18 pp. February 25. 

No. 387. Growth Changes in Amphilimna olivacea (Lyman) 
and the Systematic Status of Amphitarsus spinifer 
Schoener. By Lowell P. Thomas and Amy Schoener. 
9 pp. March 15. 

No. 388. Intense Low-Frequency Sounds from an Antarctic 
Minke Whale, Balaenoptera acutorostrata. By Wil- 
liam E. Schevill and William A. W'atkins. 8 pp. 
April. 

No. 389. The Chahares (Argentina) Triassic Reptile Fauna. 
XI IL An Early Ornithosuchid Pseudosuchian, Gra- 
cilisuchus stipanicicorum, Gen. et Sp. Nov. By Al- 
fred Sherwood Romer. 24 pp. August 1 1 . 

No. 390. The Chanares (Argentina) Triassic Reptile Fauna. 

XIV. Lewisuchus admixtus, Gen. et Sp. Nov., a 
Further Thecodont from The Chaiiares Beds. By 
Alfred Sherwood Romer. 1 3 pp. August 1 1 . 

No. 391. The Relationship of Island Area and Isolation to Color 
Polymorphism in Liguus fasciatus (Pulmonata, Bul- 
imulidae). By Michael A. Rex. 15 pp. August 11. 

No. 392. A Fossil Pelomedusid Turtle from Puerto Rico. By 
Roger Conant Wood. 13 pp. August 11. 

No. 393. Minute Jamaican Prosobranch Gastropods: Stoastoma 
and its Congeners. By Kenneth J. Boss. 13 pp. 
x\ugust 1 1 . 

No. 394. The Chahares (Argentina) Triassic Reptile Fauna. 

XV. Further Remains of the Thecodonts Lagerpe- 
ton and Lagosuchus. By Alfred Sherwood Romer. 
7 pp. August 11. 



No. 395. The Chanares (Argentina) Triassic Reptile Fauna. 
XVI. Thecodont Classification. By Alfred Sher- 
wood Romer. 24 pp. November 6. 

No. 396. The Chaiiares (Argentina) Triassic Reptile Fauna. 
XVn. The Chanares Comphodonts. By Alfred 
Sherwood Romer. 9 pp. November 6. 

No. 397. Xyloredo, a New Teredinid-like Abyssal Wood-borer 
(Mollusca, Pholadidae, Xylophagainae) . By Ruth 
D. Turner. 1 9 pp. November 6. 

1973 

No. 398. Pseudobeaconia, a Perleidiform Fish from the Triassic 
Santa Clara Formation, Argentina. By Peter Hutch- 
inson. 24 pp. March 6. 

No. 399. The Evolution of Mammalian from Reptilian Denti- 
tions. By J. W. Osborn and A. W. Crompton. 
18 pp. March 6. 

No. 400 The Carolina Salt Marsh Snake : a Distinct Form of 
Matrix sipedon. By Roger Conant and James D. 
Lazell, Jr. 13 pp. March 6. 

No. 401. The Chanares (Argentina) Triassic Reptile Fauna. 
XVm. Probelesodon minor, a New Species of Car- 
nivorous Cynodont; Family Probainognathidae Nov. 
By Alfred Sherwood Romer. 4 pp. March 6. 

No. 402. Speciation in the Genus Ochthoeca (Aves: Tyran- 
nidae). By John W. Fitzpatrick. 13 pp. June. 

No. 403. A New Late Paleocene Phenacodont (Mammalia: 
Condylarthra) from Western Colorado. By Br) an 
Patterson and Robert M. West. 7 pp. June. 

No. 404. New Studies on a Montane Lizard of Jamaica, Anolis 
reconditus. By Robert Hicks. 23 pp. June. 



No. 405. A Fossil Trionychid Turtle from South America. By 
Roger Conant Wood and Bryan Patterson. 10 pp. 
September 20. 

No. 406. Two New Lygosomine Skinks from New Guinea with 
Comments on the Loss of the External Ear in Ly- 
gosomines and Observations on Previously Described 
Species. By Allen E. Greer, Jr. 25 pp. September 20. 

No. 407. The Chafiares (Argentina) Triassic Reptile Fauna. 
XIX. Postcranial Materials of the Cynodonts Pro- 
belesodon and Probainognathus. By Alfred Sher- 
wood Romer and Arnold D. Lewis. 26 pp. Septem- 
ber 20. 

No. 408. A Taxonomic Comparison of the American Upogebia 
(Decapoda, Thalassinidea), Including Two New 
Species from the Caribbean. By Da\'id Thistle. 
23 pp. September 20. 

No. 409. The Classification of the Cotingidae (Aves). By D. W. 
Snow. 27 pp. September 20. 



BREVIORA 

Museum of Comparative Zoology 

INDEX OF AUTHORS 

Numbers 380-409 

1971-1973 

No. 

Boss, Kenneth J 393 

CoNANT, Roger 400' 

Crompton, a. W 399 

Erwin, Terry L. 383 

Evans, Howard E. 386 

Fitzpatrick, John W 402 

Greer, Allen E 406 

Hicks, Robert 404 

Hutchinson, Peter 398 

Lawrence, John F 380 

Lazell, James D 400 

Lewis, Arnold D 407 

Neaves, William B 381 

OsBORN, J. W 399 

Patterson, Bryan 403, 405 



Rex, Michael A. . 391 

Roberts, Tyson R 382, 384 

RoMER, Alfred Sherwood 

385, 389, 390, 394, 395, 396, 401, 407 

Schevill, William E 388 

ScHOENER, Amy 387 

Snow, D. W 409 

Thistle, David 408 

Thomas, Lowell P 387 

Turner, Ruth D, 397 

Watkins, Willl\m E 388 

West, Robert M 403 

Wood, Roger Conant 392, 405 



B R E V I O'^'teA 

Museiiiii of Comparative Zoology 

Cambridge. Mass. 15 June, 1971 Number 380 

THE TAXONOMIC POSITION OF THE GENUS 
PETRIA SEMENOW (COLEOPTERA: ALLECULIDAE) 

John F. Lawrence^ 

Abstract. The beetle genus Petria Semenow includes a few species in- 
habiting the Kara-Kum Desert. U.S.S.R. They are known only from males, 
which are soft-bodied, with short elytra, long antennae, and large eyes. 
The genus is usually placed in a separate family, the Petriidae, within the 
section Heteromera, but some Russian workers have included it within the 
family Alleculidae. A detailed study of Petria antcnnata Semenow provides 
abundant evidence supporting the inclusion of Petria within the subfamily 
Omophlinae of the Alleculidae. Brief descriptions and figures are given for 
the prothorax, metendosternite, hindwing, tarsal claw, abdomen, and male 
genitalia, and comparisons are made with members of the Alleculidae and 
other heteromeran families. Speculations are made concerning the habits 
and habitat of the larva and female based on knowledge of related omoph- 
lines and analogous types of desert-inhabiting Coleoptera. 

The genus Petria Semenow includes a few species of peculiar 
looking beetles that inhabit the Kara-Kum, a desert east of the 
Caspian Sea (Turkmen S.S.R. ). The group is known only from 
males, which are attracted to lights at night. They are relatively 
small (3.0-4.25 mm), soft-bodied, and lightly pigmented insects, 
with shortened elytra, long antennae, and large, prominent eyes 
{see illustrations in Jacobson, 1913, 1915; Ogloblin and Znoiko, 
1950; and Znoiko, 1936). The genus has been placed in a sep- 
arate family, the Petriidae. which is considered in most general 
texts to be of uncertain phylogenetic position within the section 
Heteromera (Crowson. 1955). 

Although Semenow ( 1 893. 1 896) made Petria the type of a new 
family, he considered the genus to be related to members of the 

' Museum of Comparative Zoology, Cambridge, Mass. 02138 



2 BREVIORA No. 380 

Allcculidae and particularly the omophline genus Steneryx Reitter, 
also from Central Asia. Most later authors recognized the family 
Petriidae, but Znoiko (1936) presented evidence for the inclusion 
of Petria within the alleculid subfamily Omophlinae and pointed 
out a transition in general form, eye size, wing venation, tarsal 
claws, and antennal structure, among the species of Petria, Stene- 
ryx, and a third genus, Cnecosochara Reitter, also known from 
males only. In spite of Znoiko's conclusions, the Petriidae is still 
treated as a heteromeran family of doubtful affinities, and Crowson 
(1955) made no attempt to place it, since specimens for dissection 
were not available to him. 

The major problem in classifying Petria on the basis of super- 
ficial characters is that, while it closely resembles several Allcculi- 
dae, it has been thought to lack the main diagnostic features of that 
family, namely, the closed procoxal cavities, connate basal ab- 
dominal sternites (3-5), and pectinate tarsal claws. In order to 
obtain material for dissection, I wrote to Leningrad, and, through 
the kindness and cooperation of Dr. G. S. Medvedev of the Zoologi- 
cal Institute, Academy of Sciences, U.S.S.R., received three speci- 
mens from Semenow's series of P. antennata, collected at Utsh- 
adzhi in May of 1889. Sincere thanks are due to Dr. Medvedev 
and the Zoological Institute. I am also grateful to R. A. Crowson 
for his continual inspiration, comments, and criticisms, and to 
P. J. Darlington, Jr., and T. F. Hlavac for their useful suggestions 
and critical review of the manuscript. Although specimens of 
Steneryx and Cnecosochara were not available for study, dissec- 
tions were made of Cteniopus flavus (Scopoli) and examinations 
were made of several alleculids and other Heteromera. 

A study of the above material revealed the following features of 
Petria antennata that shed some light on its phylogenetic relation- 
ships: 

1 ) The procoxae are subconical, projecting, and almost con- 
tiguous, the intercoxal process of the prosternum being laminate 
(Fig. 1 ), but each coxa bears a relatively large articular region (a), 
which is concealed by the coxal cowling (c), so that the articula- 
tion with the pleuron is internalized (Fig. 2). 

2) The pleuro-coxal mechanism of the prothorax is of the tene- 
brionoid type, with the endopleuron fused to the notal wall and 
the trochantin apparently absent. 



1971 GENUS PETRIA 3 

3) The procoxal cavities are closed internally and open exter- 
nally or posteriorly (Fig. 1). 

4) The mesocoxal cavities are not closed outwardly by the meso- 
and metasterna and are thus contiguous laterally with the mes- 
epimera. 

5) The metendosternite is of the tenebrionoid type, with a 
narrow stalk, no laminae, and the anterior tendons out near the 
apices of the lateral arms (Fig. 3). 

6) The tibial spurs are neither serrate nor pubescent. 

7 ) The tarsal claws each have two or three toothlike projections 
(Fig. 5). 

8 ) The hindwing is similar to that of most tenebrionoids, with 
four well-developed anal veins (in the main group), an anal or 
wedge cell (w) present, and a short stalk of Rs extending basad of 
the radial cell (not shown in Znoiko, 1936) (Fig. 4). 

9) Abdominal sternites 3, 4, and 5 are connate. This is not 
easily observed without sectioning, since all the segments are 
broadly overlapping (Figs. 6 and 7). 

10) Abdominal sternite 7 has a pair of posterolateral gland 
openings (Fig. 9) similar to those found in Cteniopus {see Ken- 
dall, 1968). ^ 

1 1 ) Abdominal sternite 8 is developed into a pair of claspers 
(Figs. 6 and 8) that are similar to those of alleculids {see Camp- 
bell, 1966; Champion, 1888; Kaszab, 1969; McDonald, 1960; and 
Oglobhn and Znoiko, 1950). 

12) The aedeagus is of the normal (not inverted) heteromeroid 
type, with a long basal piece (b), short tegmen (apical or cap 
piece) (t), and the median lobe (m) membranous except for a 
ventral strut (or two fused struts) at the apex (Fig. 10). This is 
a typical alleculid aedeagus as illustrated in Campbell (1966. 
1968); McDonald (1960); Marshall (1970a, 1970b); and Sharp 
and Muir (1912). 

Except for the lack of an external coxal closure, the condition 
of the prothorax in Petria is strongly suggestive of a relationship to 
the tenebrionoid complex of the Heteromera (Tenebrionidae, 
Lagriidae, Alleculidae, and Nilionidae). The internalization and 
concealment of the pleural articulation, fusion of the endopleuron 
to the notum, and reduction of the trochantin are characteristic of 



4 BREVIORA No. 380 

the tenebrionoid families and a few related groups, such as the 
Colydiidae, Zopheridae, Monommidae, Prostomidae, and Daco- 
deridae; a similar condition also occurs in certain Clavicornia 
(Propalticidae, cerylonoid complex) (Crowson, 1955; Hlavac, per- 
sonal communication; Watt, 1967). The internal closure of the 
procoxal cavities occurs in most Heteromera, but is absent in the 
Mycetophagidac, Ciidae, Pterogeniidae, Tetratomidae, Perimylopi- 
dae, Zopheridae, Monommidae, Pythidae, and Pyrochroidae. The 
development of conical, projecting procoxae and a laminate inter- 
coxal process has taken place in various members of the tene- 
brionoid association (Lagrio, Mycetochara, Cteniopus), but 
externally open procoxal cavities are unknown in this large group, 
except in a few sub-Antarctic forms that are doubtfully included 
or have been removed (Crowson, 1955; Watt, 1967). 

The laterally open mesocoxal cavity is a fairly common feature 
in the Heteromera, but in the families Prostomidae, Zopheridae, 
Monommidae, Elacatidae, Mycteridae, Inopeplidae, and Salpingi- 
dae the cavity is closed by the meeting of the meso- and metasterna 
(Crowson, 1955; 1967). 

The metendosternite of Petria is also of the tenebrionoid type, 
with no laminae and with laterally placed tendons, and differs 
from that found in most groups of Heteromera. A similar struc- 
ture may be found, however, in certain Melandryidae and Scrap- 
tiidae, which differ from Petria by having serrate or pubescent tibial 
spurs and free abdominal sternites (Crowson, 1938, 1944, 1955, 
1966). 

The wing venation also points to the tenebrionoid complex, be- 
ing almost identical with that of certain omophline Alleculidae and 
a number of Tenebrionidae; in most Melandryidae, the venation is 
simpler with the anal cell absent, but in Melcmdrya it is essentially 
the same as that of Petria (Bernet-Kempers, 1923; Crowson, 1955, 
1966; Forbes, 1922; Znoiko, 1936). 

The fusion of the basal three abdominal sternites and the 
presence of glands on the seventh sternite are probably the strong- 
est pieces of evidence for the association of Petria with the 
Tenebrionidae and their relatives. The fusion of sternites is not 
uncommon in the Heteromera and has probably occurred several 
times, but the connation of the first three appears to be restricted 
to the tenebrionoids, certain Colydiidae, {Mrymechixemis, Pyc- 
nomerus, Anchomma) and Meryx (Merycidae). Two sternites 



1971 GENUS PETRIA 5 

are united in the Pteiogeniidae, Prostomidae, Cononotidae, Myc- 
teridae, Lo^rioida (Anthicidae), and a few other genera of doubtful 
affinities, while four are actually or apparently fused in the ma- 
jority of Colydiidae, Zopheridae, Monommidae, and Dacoderidae 
(Crowson, 1955, 1967; Watt, 1967). The situation is complicated 
by the fact that fusions are not always visible from the surface (as 
in Petria) or intersegmental membranes are concealed (as in 
Dacoderiis, see Watt, 1967). The apparent fusion of three basal 
abdominal sternites, however, is a consistent feature of the tene- 
brionoid families and is almost always correlated with several other 
adult and larval characters. 

The presence of defense glands on the seventh abdominal ster- 
nite is known, according to Kendall (1968), only in the families 
Lagriidae, Alleculidae, and Tenebrionidae. Although the actual 
glandular reservoirs were not seen in my material, openings and 
ducts, similar to those of Cteniopiis sulphuripes (Linnaeus) and 
C. ftuvus (Scopoli), were present at the posterior angles of the 
sternite. 

The presence of small teeth on the tarsal claws, the modification 
of the eighth sternite into abdominal claspers, and the form of the 
aedeagus all argue for the inclusion of Petria in the family Alle- 
culidae, especially when coupled with the several tenebrionoid 
features mentioned above. Male claspers may be found in other 
groups, such as the Oedemeridae and Cephaloidae, but the struc- 
ture of these organs is entirely different and the other tenebrionoid 
characters are lacking (Arnett, 1951; 1953). 

The arguments presented above, when added to the comparisons 
of Znoiko (1936), leave little doubt that Petria belongs in the 
family Alleculidae and is related to Steneryx, Cnecosochara, and 
their allies in the subfamily Omophhnae. It represents an adaptive 
extreme characterized by the loosely built and lightly sclerotized 
body, shortened elytra, large eyes, loss of the combs on the tarsal 
claws, loss of the external coxal closure, and sexual wing dimorph- 
ism. The last feature is not actually known in Petria but may be 
postulated by analogy with other groups known from males only 
(see below). 

Semenow (1893) speculated that the females of Petria, since 
they were never collected along with males, might be wingless 
parasites of Hymenoptera or Orthoptera, as is the case in certain 
Meloidae and Rhipiphoridae. In other omophlines, however, such 



6 BREVIORA No. 380 

as Omophlus, Podonta, and Cteniopus, the larvae are soil inhabi- 
tants, feeding on various roots and tubers, while the adults are 
usually found on flowers and fruits (Aguilar, 1962; Kaszab, 1969). 
A more reasonable hypothesis, then, would be that Petria larvae 
are root feeders and that the wingless females live either in the 
soil or at the bases of shrubs or grasses. This particular type of 
wing dimorphism and habitat selection is known in at least two 
other xerophilous beetles, the males of which resemble those of 
Petria, namely Vesperus (Cerambycidae) and Anorus (Dascilli- 
dae). Species of Vesperus inhabit the Mediterranean Region; the 
wingless female deposits her eggs above the ground, and the fast 
moving, triungulinlike larvae enter the soil and are transformed 
into radically dift'erent grubs, which feed on roots (Balachowsky, 
1962). Anorus species occur in the deserts of southwestern North 
America and Chile; the larvae are unknown (but those of the re- 
lated Dascillus are root feeders), and the only known female is 
wingless and subterranean (Blaisdell, 1934). It is likely that the 
females and larvae of Petria, when they are discovered, will have 
similar habits. 

LITERATURE CITED 

Aguilar, J. d". 1962. Famille des Alleculidae, pp. 368-373. In A. S. Bala- 
chowsky (Ed.), Entomologie appliquee a I'Agriculture. Traite. Tome I. 
Coleopteres. Premier Volume. Paris, Masson, XXVII + 564 pp. 

Arnett, R. H., .jr. 1951. A revision of the Nearctic Oedemeridae (Coleop- 

tera). American Midland Nat., 45: 257-391. 
1953. A review of the beetle family Cephaloidae. Proc. U. S. 

Nat. Mus., 103: 155-161. 

Balachowsky, A. S. 1962 Famille des Cerambycidae, pp. 394-434. In 
A. S. Balachowsky (Ed.), Entomologie appliquee a I'Agriculture. Traite. 
Tome I. Coleopteres. Premier Volume. Paris, Masson. 

Bernet-Kempers, K. 1923. Abbildungen von Fliigelgeader der Coleop- 
teren. Ent. Mitt., 12: 71-115. 

Blaisdell, F. E. 1934. Rare North American Coleoptera. Trans. American 
Ent. Soc, 60: 317-326. 

Campbell, J. M. 1966. A revision of the genus Lohopoda (Coleoptera: 

Alleculidae) in North America and the West Indies. Illinois Biol. Mon., 

37: 1-203. 
1968. A revision of the Mexican and Central American 

species of Isomira (Coleoptera: Alleculidae). Canadian Ent., 100: 

449-469. 



1971 GENUS PETRIA 7 

Champion. G. C. 1888. Fam. Cistelidae, pp. 384-465, pis. 17-21. In F. D. 

Godman and O. Salvin (Eds.), Biologia Centrali-Americana, Insecta, 

Coleoptera. Vol. 4. Part 1. Heteromera (part). London, Porter. 
Crowson, R. a. 1938. The metendosternite in Coleoptera: a comparative 

study. Trans. Roy. Ent. Soc. London. 87: 397-415. 

1944. Further studies on the metendosternite in Coleoptera. 

Trans. Roy. Ent. Soc. London, 94: 273-310. 
1955. The Natural Classification of the Families of Coleop- 



tera. London. Lloyd. 187 pp. 
1966. Observations on the constitution and subfamilies of the 



family Melandryidae. Eos, 41: 507-513. 
1967. The natural classification of the families of Coleoptera. 



Addenda and corrigenda. Ent. Mon. Mag., 103: 209-214. 
Forbes, W. T. M. 1922. The wing-venation of the Coleoptera. Ann. Ent. 

Soc. America, 15: 328-352, pis. 29-35. 
Jacobson, G. G. 1913. Zhuki Rossii i zapadnoi Evropi. Vip. X. Leningrad, 

Devrient. Pp. 721-864, pis. 76-83. 
1915. Zhuki Rossii i zapadnoi Evropi. Vip. XI. Leningrad, 

Devrient. Pp. 865-1024. 
Kaszab, Z. 1969. Famille: AUecuhdae, pp. 215-229. In H. Freude, K. 

Harde, and G. Lohse (Eds.). Die Kafer Mitteleuropas. Band 8. Kre- 

feld, Goecke and Evers. 388 pp. 
Kendall, D. A. 1968. The structure of the defense glands in Alleculidae 

and Lagriidae (Coleoptera). Trans Roy. Ent. Soc. London, 120: 

139-156. 
McDonald, J. M. 1960. Morphology of the exoskeleton of Capnochroa 

jiiliginosa (Melsheimer) (Alleculidae). Coleopt. Bull., 14: 97-120. 
Marshall, J. D. 1970a. Isomira Mulsant in America north of Mexico 

(Coleoptera, Alleculidae): Redescriptions, new synonymies, and tax- 

onomic notes on eastern North American species. Ent. News, 81 : 

41-49. 
1970b. Isomira Mulsant in America north of Mexico (Co- 
leoptera: Alleculidae): Species of eastern North America concluded. 

Coleopt. Bull., 24: 88-95. 
Ogloblin, D. a., and D. V. Znoiko. 1950. Piltseyedi (Sem. Alleculidae). 

Ch. 2, Podsem. Omophlinae. Fauna S.S.S.R. Zhestkokriliye. Tom. 18, 

Vip. 8 (Nov. Ser., No. 44). Moscow and Leningrad. Akademia Nauk 

S.S.S.R. 135 pp. 
Semenow, a. 1893. De Coleopterorum familia nova. Melanges Biol., 13: 

359-366; Bull. Acad. Imp. Sci. St.-Petersbourg (N. S.), 3(35) : 607-615. 
1896. Revisio specierum generis Petria Sem. Annuaire Mus. 

Zool. Acad. Imp. Sci. St.-Petersbourg, 1: 25-30. 



8 BREVIORA No. 380 

Sharp, D., and F. Muir. 1912. The comparative anatomy of the male 
genital tube in Coleoptera. Trans. Ent. Soc. London, 1912: 477-642, 
pis. 42-78. 

Watt, J. C. 1967. The families Perimylopidae and Dacoderidae (Coleop- 
tera, Heteromera). Proc. Roy. Ent. Soc. London, Ser. B, 36: 109-118. 

Znoiko, D. V. 1936 Ueber die systematische Stellung der Familie Petriidae 
(Coleoptera). Trudi Zool. Inst. Akad. Nauk S.S.S.R., 3: 67-77. 



1971 



GENUS PETRIA 












Figures 1-10. Petria antennata Semenow. male (1 line = 0.125 mm unless 
otherwise indicated). Fig. 1. Prothorax, ventral view, with left coxa re- 
moved. Fig. 2. Procoxa, showing internal articular area (a). Fig. 3. Meten- 
dosternite, dorsal view. Fig. 4. Hindwing (0.50 mm). Fig. 5. Tarsal claw 
(0.038 mm). Fig. 6. Abdomen (0.25 mm). Fig. 7. Cross section of ab- 
dominal sternites 3-6 (0.076 mm). Fig. 8. Abdominal claspers and asso- 
ciated structures, ventral view. Fig. 9. Abdominal sternite 7, showing 
glandular ducts and openings. Fig. 10. Aedeagus, lateral view. 



M^S. COMP. 200L 
LIBRARY 

MAR 1 8 1985 

B IR E V„,|^§ R A 

Museum of Comparative Zooloi^y 

Cambridge, Mass. 20 Dec-ember. 1971 Number 381 

TETRAPLOIDY IN A HYBRID LIZARD OF THE GENUS 
CNEMIDOPHORUS (TEIIDAE) 

William B. Neaves^ 



Abstract. An apparent hybrid between triploid. parthenogenetic 
Cneiuidophorits exsan(;iiis and diploid, sexual C. inonuitiis is shown to 
possess a tetraploid chromosom; complement. Evidence suggests that 
this tetraploid karyotype resulted from the fusion of a haploid sperm 
pronucleus of C. inornatus with an egg pronucleus carrying the unreduced 
triploid chromosome complement of C. cxsauiniis. Observations of cap- 
tive C. inornatus males show their propensity to engage in mating be- 
havior with C. exsangnis parthenogenones. These findings are discussed 
with regard to the origin and genetic mechanism of parthenogenesis in 
Cnciuidophorus. 

INTRODUCTION 

The hypotiiesis that parthenogenetic species of Cnefuiclophorus 
arose from the offspring of interspecific hybridizations (Lowe and 
Wright, 1966) has been supported by evidence from studies of 
lactate dehydrogenase (Neaves and Gerald, 1968) and adenosine 
deaminase (Neaves, 1969). A particular aspect of that hypothesis, 
namely, that triploid parthenogenones originated in crosses be- 
tween males of a sexual species and females of a diploid, par- 
thenogenetic species, has received support from reports that such 
hybridizations occur in nature. Taylor and Medica (1966) 
described apparent hybrids between Cnemidophorus inornatus, 
a sexual species, and C. neomexicanus, a diploid parthenogenone. 



' Department of Anatomy, Harvard Medical School, Boston, Massa- 
chuseUs 02115 



2 BREVIORA No. 381 

Wright and Lowe ( 1967a) verified the occurrence of C. inornatus 
X C. neomexicanus hybrids, conveniently designated C. ''per- 
plexus," at several localities in New Mexico where the two parental 
species are sympatric. 

The question arises as to whether or not hybridization in 
Cnemidophorus ends with the attainment of triploidy. Although 
triploid parthenogenones are common (of seven parthenogenetic 
species in the sexlineatus group, five are triploid), no tetraploid 
species are known in the genus. Lowe and Wright (1966) men- 
tion a suspected tetraploid individual that they considered to 
represent a hybrid of C. inornatus and C. uniparens, the latter a 
triploid parthenogenone, but no evidence, karyotypic or otherwise, 
was presented to verify their suspicion. My report describes in 
detail a documented instance of tetraploidy in a Cnemidophorus 
hybrid and seeks to explain why there are as yet no tetraploid 
species in nature. Observations of interspecific mating behavior 
involving male C. inornatus and females of various parthenogenetic 
species are included in this report. Finally, these findings are 
discussed with regard to the origin and genetic mechanism of 
parthenogenesis in Cnemidophorus. 

MATERIALS AND METHODS 

Animals used in these studies were captured alive in their 
native habitats in Colorado, New Mexico, and Texas during the 
summers of 1967 and 1968. Soon after capture, these lizards 
were shipped to Boston and Cambridge. Upon arrival in the 
laboratories, they were housed either in a rectangular pen 12 
feet by 14 feet or in oval pens 3.5 feet by 5 feet. The pens had 
sand floors with tiles and sheets of cardboard scattered about for 
shelter. Sun lamps installed over the pens were automatically 
controlled to provide a twelve-hour daily photoperiod. A constant 
supply of open water and available food, in the form of mealworms 
and crickets, was provided. 

Aspects of lizard behavior reported here were observed and 
photographed among the captive inhabitants of the large rec- 
tangular pen described above. Most incidents were recorded dur- 
ing the winter of 1968-1969 when the pen held approximately 
150 Cnemidophorus distributed among five sexual and five par- 
thenogenetic species. 



1971 TETRAPLOID CNEMIDOPHORUS 3 

The tetraploid karyotype illustrated in this report was obtained 
from a phcnotypically aberrant lizard captured at Alamogordo, 
New Mexico, in August, 1967. This lizard was maintained alive 
until May, 1968, when it was used for karyotypic study. Mitotic 
figures were obtained from tissue cultures of 1 mm cubes of heart 
tissue according to the following procedure. Small Falcon flasks 
were loaded with 5.0 ml of a mixture containing 80 percent Eagle's 
Minimum Essential Medium (Microbiological Associates) and 
20 percent fetal calf serum. With the flask in an upright position, 
tissue explants were placed against the dry wafl of the flask. The 
flask was slowly lowered to its side so that the explants were 
covered by the medium but not dislodged from the plastic surface. 
The flasks were incubated at 33" C. without agitation. The cul- 
tures were examined regularly with an inverted phase contrast 
microscope in order to follow cell proliferation. On the ninth day 
of incubation, when mitotic activity was intense, 0.25 microgram 
of Velban (Grand Island Biological Co.) was added to each flask. 
Twenty-four hours later, the medium was discarded and 5.0 ml 
of 0.1 percent trypsin in 0.8 percent sodium citrate was added 
to each flask. FoUowing a twenty-minute incubation at 37 C, 
the flasks were shaken vigorously to free the dissociating cells 
from one another and from the plastic. From this point, the 
cultures were processed according to a modification of the method 
outlined by Moorhead et al. (1960). The solution from each 
flask was decanted into a centrifuge tube and spun at 1000 g for 
five minutes. The cell pellet was fixed in two changes of 
ethanol: acetic acid (3:1) and was suspended in 45 percent acetic 
acid. Drops of this suspension were placed on cold, wet slides 
and flame dried. The chromosomes on these slides were stained 
with 1 percent toluidine blue in 1 percent borax and were photo- 
graphed with an Olympus photomicroscope. 

THE DISCOVERY OF THE TETRAPLOiD 

In the course of a collecting trip to New Mexico in August, 
1967, I visited an exceptionally dense Cnemidophonis population 
within the city limits of Alamogordo in Otero County. The 
population was largely restricted to a weed bed approximately 20 
meters wide and over 1000 meters long, bounded on the west 
by the roadbed of the Southern Pacific Railroad and on the east 



4 BREVIORA No. 381 

by the Alamogordo City Park. Two Cnemidophorus species were 
found in the area described. C. exsanguis, a triploid parthe- 
nogenone, was most abundant, occurring at a density of approxi- 
mately 50 animals per acre; C. inornatus, a sexual species, oc- 
curred at a density of about 10 animals per acre. No lizards were 
seen in the park itself, which had a cover of closely mown grass. 
West of the roadbed, the cover consisted of mesquite shrub and 
cactus on an eroded surface virtually devoid of weeds or grass. 
Here both C. inornatus and C. tigris, a second sexual species, 
were abundant. In two days of collecting west of the roadbed, 
only a single C. exsanguis was seen. No C. tigris were found east 
of the roadbed in the weeds where C. exsanguis was so abundant. 

Forty-three C. exsanguis and eleven C. inornatus were captured 
in the Alamogordo weed bed in 1967. In addition, two aberrant 
Cnemidophorus were taken. One of these, MCZ* 100080, was 
the size of C. inornatus, with partially fused paravertebral stripes, 
a white ventral surface, and a rusty tint on its dorsum, causing 
it to resemble C. exsanguis. The day after its capture, this animal 
died. Its abdomen was opened to expose the gonads, which ap- 
peared rudimentary. The nature of this specimen remains a 
mystery. A second aberrant specimen, MCZ 101991 (Plate 1), 
resembled a typical C. exsanguis except that its paravertebral 
stripes were fused and its ventral surface and tail were suffused 
with a brilliant blue characteristic of the same surfaces in C. 
inornatus males. This exsanguis-like animal was maintained alive 
for almost nine months before it was sacrificed for a study of its 
chromosomes. 

It wai immediately suspected that MCZ 101991 might represent 
a cross between the sympatric species of the weed bed, C. 
inornatus and C. exsanguis. This lizard clearly possessed attributes 
characteristic of both suspected parental species. A decision was 
reached to allow the animal to remain alive as long as possible 
so that it 5 behavior might be observed, but at the same time, an 
assessment of its ploidy was desired. Accordingly, blood smears 
were prepared from the exsanguis-Vike animal and from other 
Cnemidopliorus known to be either diploid or triploid on the 



'•' Museum of Comparative Zoology 



1971 TETRAPLOID CNEMIDOPHORUS 5 

basis of karyotypic evidence (Lowe and Wright. 1966; Wright 
and Lowe, 1967b). These were sent to N. B. Atkin for dcnsi- 
tometric estimation of DNA in the nucleated erythrocytes. Dr. 
Atkin reported (in litt.) that karyotypically triploid species (3n = 
69±1) possessed about 50 percent more DNA per nucleus than 
karyotypically diploid species (2n = 46) and that the aberrant 
specimen (MCZ 101991 ) possessed about 20 percent more DNA 
per nucleus than triploids such as C. e.\sanguis. The details of 
erythrocyte DNA analysis in Cnemidophonis will be reported else- 
where. Although a true tetraploid should have demonstrated 
approximately 30 percent more DNA per nucleus than a triploid, 
the results of Atkin's survey encouraged the suspicion that the 
aberrant specimen represented a hybrid carrying some genetic 
elements characteristic of C. inonuitus in addition to the basic 
triploid complement characteristic of C. exsangiiis. 

In December 1967, Professor C. H. Lowe examined the aber- 
rant lizard and declared it unlike any Cnemidophonis he had ever 
seen. Lowe agreed that it might represent a tetraploid hybrid 
between C. inonuitus and C. exsangiiis. Meanwhile, this lizard 
thrived in captivity and proved itself a vigorous competitor for 
food; it commonly robbed typical C. exsangiiis of mealworms. 

In January 1968, the aberrant lizard laid two eggs, thus con- 
firming the suspicion that it was female and suggesting that its 
ovaries were functional. Attempts were made to provide suitable 
conditions for ovoposition, but she did not take advantage of them, 
and the eggs were dessicated by the time 1 discovered them in my 
daily check. She produced no more eggs during her captivity, 
which extended through May, 1968. 

When the aberrant lizard was eventually sacrificed for karyo- 
typic analysis, the results (Plates 2 and 3 and Table 1 ) confirmed 
its suspected status as a tetraploid hybrid. The modal chromo- 
some number, based on examination of eight apparently complete 
chromosome spreads, is 9L The karyotype in Plate 2 is from the 
single spread showing 92 chromosomes. Considering the large 
number of chromosomes in the karyotype, the small size of many, 
and the probability of overlap, it is not surprising that fewer than 
92 chromosomes are evident in most spreads. The tetraploid 
karyotype can be divided into four apparently identical haploid 
complements, each closely resembling the haploid chromosome 



6 BREVIORA No. 381 

complement of C. inornatus (Lowe and Wright, 1966). Three 
inornatiis-WkQ haploid sets constitute the typical triploid karyotype 
of C. exscmguis, which is similar to that of another triploid parthe- 
nogenone, C. uniparens (Lowe and Wright, 1966). The fourth 
haploid chromosome complement seen in the aberrant lizard's 
karyotype may have been contributed by C. inornatus itself, 
through an interspecific mating with C. exscmguis. 

OBSERVATIONS OF CAPTIVE SPECIMENS: 
INTERSPECIFIC MATING 

Observations of captive Cnemidophorus during autumn and 
winter of 1968-1969 suggest that C. inornatus males may be 
unusually disposed to interspecific mating with parthenogenetic 
females. Dozens of courtship encounters, including attempted 
copulation, were recorded between C. inornatus males and parthe- 
nogenetic females from the species C. exsanguis and C. tesselatus 
(Plate 4, figs. 5-9). 

The sequence of sexual behavior involved pursuit of a subject, 
often a parthenogenetic female but sometimes an individual of a 
sexual species, including other C. inornatus, for five to twenty 
seconds at a speed depending on the movement of the subject, the 
male apparently trying to stay near the tip of the subject's tail. 
Suddenly, the C. inornatus male would close with the subject, 
mount its back, and grasp in his jaws a fold of skin on the back 
of the subject's neck. If the subject acquiesced, the male would 
immediately execute sinuous movements while pressing its abdo- 
men against the subject's dorsum. With the subject still quiescent, 
the male would loosen its grip on the subject's skin and caress the 
back of the subject's head and neck with his jaw while continuing 
the sinuous body movements. The male seemed particularly in- 
tent on rubbing his ventral pelvis and tail base against the dorsal 
pelvis of the subject. Meeting no resistance, the male would 
maneuver his vent into the area between the subject's rear limb 
and tail base. When the subject was a large parthenogenetic 
female, this maneuver required the male to creep back from his 
original position in which his jaw touched the subject's head and 
neck (Plate 4, figs. 5-7). While struggling to approximate his 
vent to the subject's, the male would constantly stroke the sub- 
ject's dorsal pelvis and tail base with his rear hmb. At times, the 



1971 TETRAPLOID CNEMIDOPHORUS 7 

male"s hemipenis was visibly erect, but in spite of observed close 
apposition of tlie participants' vents, unequivocal intromission 
was not seen. 

It did not appear as though C. inornatus males met with any 
real cooperation, beyond mere quiescence, in their courtship 
efforts, but C. exsangiiis and C. tesselatiis at least permitted the 
sequence of events to proceed as far as described above. Other 
C. inoniatiis, particularly other males, made real elTorts to escape 
the attentions of an ardent but misguided suitor. Perhaps as a 
result, C. inornatus males exhibited a definite preference for the 
larger, parthenogenetic females over their conspecific females. 
Attempted intromission was observed only between C. inornatus 
males and parthenogenetic females. 

DISCUSSION 

The existence of a tetraploid lizard in a mixed population con- 
taining triploid parthenogenones and diploid, sexual lizards argues 
that tetraploid hybrids are synthesized in nature, although the 
genus Cnemidopliorus apparently contains no tetraploid species.^ 
That the tetraploid lizard described here is a hybrid and not an 
autopolyploid is suggested by its possession of phenotypic traits 
characteristic of both suspected parental species, by the propensity 
of the suspected male parent, C. inornatus, to engage in inter- 
specific mating behavior with parthenogenetic females such as C. 
exsanguis, and by the occurrence in nature of another interspecific 
hybrid, C. ''perplexus," which is considered to be the result of 
crosses between sexual and parthenogenetic lizards (Wright and 
Lowe, 1967a). 

Two questions are raised by the existence of the tetraploid 
hybrid. First is the question of the cytological events in gameto- 
genesis and fertihzation that permit a tetraploid to arise from a 
mating between a sexual male and a triploid, parthenogenetic 
female. This question, which is essentially concerned with genetic 
mechanisms operative in parthenogenesis, will be considered at 
some length in order to provide a background for the second 
question. The second question concerns the absence of tetraploid 
species in spite of the synthesis of hybrid tetraploids in nature. 

The first question has a trivial aspect, namely cytological events 
in the sexual male. Nothing more is required of the male lizard 



1 See also the discussion of two tetraploid lizards in a mixed population 
of C. sunorae and C. tigris (Lowe et a!., 1970). 



8 BREVIORA No. 381 

than the production of typical sperm that carry a haploid chromo- 
some complement and that are capable of delivering this genome 
to the female pronucleus in a mature egg. A more complicated 
aspect of the question involves the cytological events that lead 
to the production of a mature egg whose pronucleus contains the 
unreduced somatic chromosome complement characteristic of the 
mother. This, in essence, is the question of the mechanism of 
parthenogenesis in Cnemidophorus, and it deserves detailed con- 
sideration. 

In Cnemidophorus, with the possible exception of C. lemniscatus 
(Vanzolini, 1970; Hall, 1970), interspecific hybridization is im- 
plicated in the origin of parthenogenesis. Oogenesis in parthe- 
nogcnones arising from interspecific hybridization must solve two 
major problems: 1 ) the maneuvering of potentially, and sometimes 
obviously, nonhomologous chromosome sets at meiosis, and 2 ) the 
production of a mature egg possessing the somatic chromosome 
number characteristic of the mother. Several solutions have been 
proposed: 1 ) mitotic egg production replaces the normal meiotic 
mechanism (Lowe and Wright, 1966), 2) failure of either first or 
second polar body formation, or fusion of one of these bodies 
with the egg nucleus after its formation (Beatty, 1957), 3) sup- 
pression of the first cleavage division (Beatty, 1957), or 4) endo- 
mitosis in the oogonium prior to the onset of meiosis (MacGregor 
and Uzzell, 1964). 

Solutions 2 and 3 cannot compensate for difticulties in bivalent 
formation that arise when nonhomologous chromosome sets are 
present at the onset of meiosis, nor can they compensate for 
situations in which more than two homologous chromosome sets 
enter meiosis. Since some diploid parthenogenones, such as the 
four Lacerta parthenogenones studied by Darevsky (1966), may 
be derived from closely related sexual species possessing fairly 
homologous karyotypes, solutions 2 and 3 cannot be ignored. In 
fact, Darevsky's ( 1966) study of parthenogenetic Lacerta suggests 
that the second meiotic division i> incomplete, with a diploid 
restitution nucleus resulting from the fusion of the daughter pro- 
nuclei during anaphase. Solutions 2 and 3 can, however, be dis- 
counted in all self-perpetuating triploids, such as C. exsanguis, 
and in diploids containing grossly nonhomologous haploid chromo- 
some complements, such as C. neomexicanus. If only solution 2 
or 3 were operative in these cases, bivalent formation should fail, 



1971 TETRAPLOID CNEMIDOPHORUS 9 

owing to the absence of homologues, or, in those triploids that 
appear to have homologous haploid chromosome sets, owing to a 
tendency toward trivalent formation. 

Solution 1, mitotic egg production, overcomes difficulties in 
bivalent formation by avoiding the issue completely. This solution 
suffers from a lack of supporting evidence; no instance is known 
in which the germ line of a higher organism can facultatively 
abandon the meiotic theme. Furthermore, solution 1 is rendered 
dubious in Cnouidoplionis, the group for which it was originally 
proposed (Lowe and Wright, 1966), by evidence (personal ob- 
servations) that the nuclei of oocytes approximately 1 mm in 
diameter from ovaries of parthenogenetic species such as C. 
e.\Hini>uis, C. neonie.xicanus, and C tesselatus contain bivalent 
lampbrush chromosomes characteristic of meiotic diplotene. 
Hence, at least some components of typical meiosis occur in the 
ovaries of parthenogenetic Cnemidophonis. 

Solution 4, pre-meiotic endomitosis in the oogonia, has been 
shown to work in one group of parthenogenetic vertebrates, the 
ambystomatid salamanders (MacGregor and Uzzell, 1964), and 
appears likely in triploid poecilid fish (Schultz, 1967). Since pre- 
meiotic endomitosis can solve the problems of increased ploidy 
and nonhomologous chromosome complements, it is the most 
probable mechanism operative in parthenogenetic Cnemidophonis 
as well.^ This probability justifies discussion of its known features 
and its genetic implications. 

While studying preparations of diplotene lampbrush chromo- 
somes, MacGregor and Uzzell ( 1964) found 42 bivalents in each 
germinal vehicle from triploid parthenogenones (3n = 42), but 
in germinal vesicles from closely related, diploid sexual species 
(2n == 28), they found the expected 14 bivalents. This finding 
was explained by the postulation that endomitosis precedes 
meiosis in parthenogenetic oocytes, so that such oocytes enter 
meiosis with a hexaploid (6n = 84) chromosomal constitution. 
Meiotic DNA replication would then give an oocyte that is 
dodecaploid with respect to DNA, and meiosis would operate on 



' A recent paper by Orlando Cuellar verifies the occurrence of this 
mechanism in the triploid parthenogenone Cnemidophonis uniparens. 



10 BREVIORA No. 381 

42 tetrads (bivalents) to yield a pronucleus with 42 chromosomes, 
which is the somatic complement for triploid ambystomatids. 

MacGregor and Uzzell (1964) suggest that only sister chromo- 
somes resulting from the endomitotic duplication associate to 
form bivalents. The genetic status of all parthenogenones that 
may rely on pre-meiotic endomitosis hinges on the question of 
bivalent formation, making a critical assessment of the MacGregor- 
Uzzell proposition necessary. The method of bivalent formation 
will determine whether the parthenogenetic lineage will retain the 
heterozygosity inherent in its hybrid origin or will experience an 
ultimate tendency toward homozygosity. While this question is 
not directly relevant to the issue of tctraploid hybrids, it is im- 
portant to an understanding of parthenogenesis in Cnemidophorits. 

Should sister chromosomes pair exclusively, a most conserva- 
tive pattern of inheritance would result; each offspring would, 
neglecting mutation, emerge with an exact copy of its mother's 
genome. Should homologous chromosomes also pair, independent 
assortment at first meiotic metcphase would cause the offspring to 
deviate from the mother's genotype, owing to the loss of alterna- 
tive alleles in the first polar body. In triploid parthenogenones 
relying on pre-meiotic endomitosis, the probability of losing both 
representatives (sisters) of a single chromosome from a homol- 
ogous set of three in a single generation can be calculated (p = 
0.066), assuming random pairing of sister and homologous chro- 
mosomes, no crossing-over, and independent assortment. Similarly, 
after only two generations, p = 0.0066 that two chromosomes 
from a homologous set of three will be lost, leaving complete 
homozygosity at all loci on that chromosome. 

A diploid parthenogenone in which sister and homologous 
chromosomes pair randomly experiences an even stronger tendency 
toward homozygosity. The probability is 0.33 that one of a set 
of two homologous chromosomes will be lost in a single generation 
if crossing-over does not occur. In either diploidy or triploidy, 
crossing-over will only randomize the occurrence of homozygosity 
with respect to all loci on a single chromosome and will not delay 
the trend of the entire genome toward homozygosity. The ultimate 
consequence of participation of homologous chromosomes in bi- 
valent formation is homozygosity at all loci in the genome. 

The most sensitive test devised to assess the genetic status of 



1971 TETRAPI.OID CNEMIDOPHORUS 11 

parthenogenones, namely tissue grafts to determine histocompati- 
bility (Kallman, 1962; Maslin, 1967), does not discriminate 
between uniform clonal heierozygosity, which would be preserved 
in the case of exclusive pairing of sisters, and established clonal 
homozygosity resulting from independent assortment of bivalents 
formed from random pairing of both sister and homologous 
chromosomes. Only the transient period of developing homo- 
zygosity characteristic of the latter situation would be revealed as 
frequent failure of parent-to-offspring grafts and sibling-to-sibling 
grafts. 

In order to judge if pairing is restricted to sisters, as suggested 
by MacGregor and Uzzell (1964), one is left with the task of 
examining directly the composition of bivalents, or of inferring 
the degree of homo- or heterozygosity that a given parthenogenetic 
clone might possess. The former possibility, direct determination 
of bivalent composition, is smple enough in principle. The admin- 
istration of '^H-thymidine to a parthenogenone at the synthetic 
phase of mitosis immediately preceding endomitosis would result 
in the presence of radioactive label in one member of each sister 
pair arising from endomitotic duplication. Autoradiography of 
lampbrush bivalents would then show label in half of each bi- 
valent, should strict sister pairing be the rule. Random pairing 
of sisters and homologues would result in some bivalents un- 
labeled, some half labeled, and some wholly labeled. Crossing- 
over would not complicate interpretation. Although such an ex- 
periment would clearly resolve the question of bivalent composi- 
tion, practical problems, such as finding the proper time in the 
animal's life cycle for •''H-thymidine administration, make this a 
difficult exercise. 

Judgement of the degree of homo- or heterozygosity in a par- 
thenogenetic clone, and hence, inference of the composition of 
meiotic bivalents, can be based on studies of phenotypic variation. 
While some good studies of phenotypic variation in partheno- 
genones and in their sexual relatives have been performed (Zweifel, 
1965), results are not easily interpreted in favor of either homo- 
or heterozygosity. For example, groups of C. tesselatus from a 
given locality were found to exhibit a range of variation in many 
characters that approximated half that seen in local populations 
of the sexual lizard, C. tigris (Zweifel, 1965). Does the relatively 



12 BREVIORA No. 381 

smaller variation seen in parthenogenones reflect the existence of 
homozygous clones, or does it indicate the importance of recombi- 
nation in freeing variation inherent in heterozygous genomes? 
Complicating this question is a fundamental ignorance of the 
genetic regulation of most phenotypic expression. One could 
resort to the doctrine of superior fitness in heterozygotes and 
argue that parthenogenones that compete successfully with their 
sexual counterparts must necessarily be heterozygous. This begs 
an interesting question and ignores an opportunity to test notions 
of fitness that have become a foundation of evolutionary theory. 
Furthermore, such reasoning is contradicted by Darevsky's (1966) 
observations on the cytology of parthenogenesis in Lacerta. 
Darevsky maintains that failure of second meiotic division is the 
parthenogenetic mechanism operative in these lizards. Under this 
mechanism, crossing-over will temporarily maintain some hetero- 
zygosity, particularly at loci far from the centromere. However, 
the ultimate tendency is toward complete homozygosity of the 
genome (Beatty, 1957). Hence, if Darevsky's observations are 
correct, one must expect the competitively successful Lacerta 
parthenogenones to exhibit a high degree of homozygosity relative 
to their sexual counterparts. 

Assuming that pre-meiotic endomitosis facilitates partheno- 
genesis in Cnemidophorus, a reliable indication that pairing may 
be strictly limted to sister chromosomes comes from studies that 
have deduced genotypes for certain enzymes. Parthenogenetic 
Ciieiuiclophonis exhibit a striking incidence of heterozygosity at 
genetic loci determining phenotypes for lactate dehydrogenase, 
adenosine deaminase, phosphogluconate dehydrogenase, and 
NADP-dependent malate dehydrogenase (sec Table 1 in Neaves, 
1 969 ) . Most impressive are the genotypes for adenosine deami- 
nase; every parthenogenetic species studied showed heterozygosity 
at the ada locus. This is clear evidence in favor of fixed hetero- 
zygosity in parthenogenetic Cnemidophorus, and hence, in favor 
of strict sister pairing at meiosis. 

With the question of parthenogenetic mechanisms aired, it ap- 
pears that pre-meiotic endomitosis provides a basis for under- 
standing how a triploid, parthenogenetic C. exsanguis could pro- 
duce an egg whose pronucleus carried an unreduced somatic 
chromosome complement, and how the union of this pronucleus 



1971 TETRAPLOID CNEMIDOPHORUS 13 

with a haploid pronucleus from male C. inornatus could rcsull in 
an offspring carrying the chromosomes seen in Plates 2 and 3. A 
precedent for these events is established in the occurrence of such 
a fertile union of pronuclei from parthenogenetic females and 
sexual males in the genus Poeciiia (Rasch et al., 1965). 

There remains the question of the absence of tetraploid species 
of Cneniiclophoriis in spite of the existence of tetraploid hybrids 
in nature. Probing this question requires some indulgence in 
speculation, which the novelty of the subject will hopefully excuse. 

The most likely reason for the absence of tetraploid species may 
be the failure of tetraploid hybrids to reproduce parthenogeneti- 
cally. There is no evidence that the tetraploid lizard described 
in this study was parthenogenetically competent. The fact that it 
laid fully yolked eggs does not imply that these eggs either, 1 ) 
carried an unreduced chromosome complement, or 2) were 
capable of undergoing spontaneous embryonic development. 
These are two basic criteria that must be met if an interspecific 
hybrid is to achieve the reproductive success characteristic of 
existing parthenogenetic species in the genus Cneinidophorus. If 
one prefers to assume that tetraploid hybrids can reproduce 
parthenogenetically, then one must account for their absence as 
species on the grounds that no suitable ecological niche is avail- 
able to them or that they cannot successfully compete with other 
species for a mutually suitable niche. However, the similarity of 
the known tetraploid hybrid to other Cne/nidophorus suggests that 
it might compete with them for a currently available niche, and 
the behavior of the tetraploid lizard in captivity suggests that it 
could be successful in this regard. It seems that one must attempt 
instead to justify reproductive failure. 

One possibility is that tetraploidy is incompatible with the 
mechanism of parthenogenesis operative in Cneinidophorus. How- 
ever, the suspected mechanism, which is pre-meiotic endomitosis, 
has the important virtue of theoretically permitting any karyotype, 
regardless of ploidy, to function normally at meiosis. Given the 
suspected mechanism of parthenogenesis, tetraploidy itself should 
not be a barrier to reproduction. 

The most attractive possibility is that only a small proportion 
of interspecific hybrids meets the basic requirements of partheno- 
genetic reproduction. In other words, the genetically determined 



14 BREVIORA No. 381 

compensatory events, such as pre-meiotic endomitosis and spon- 
taneous embryonic development, which are presumably needed 
for parthenogenesis, may be frequently absent in F, hybrids. 
This possibility is particularly attractive, owing to the apparent 
genetic uniformity within existing parthenogenetic species of 
Cnemidophorus. 

MasUn ( 1967) has demonstrated a pattern of histocompatibility 
in C. tesselatus that suggests that all diploid members of the 
species, even when taken from localities hundreds of miles apart, 
are genetically identical. Similarly, all triploid C. tesselatus are 
reciprocally histocompatible, and what is more, can accept tissue 
grafts from the diploid ^ but cannot reciprocate. Biochemical 
evidence (Neave^, 1969) suggests that diploid C. tesselatus arose 
from interspecific hybridization between C. tigris and C septemvit- 
tatus, two sexual species, and that triploid C. tesselatus resulted 
from the addition of a haploid genome from C. sexUneatus, a third 
sexual species, to the diploid C. tesselatus genome. Coupled with 
this evidence, Maslin's (1967) findings suggest that all existing 
populations of C. tesselatus arose from the offspring of a single 
hybrid individual representing a cross between C. septeinvitiatus 
and C. tigris and that all triploid C. tesselatus are derived from a 
single hybrid lizard representing a cross between C. sexUneatus 
and diploid C. tesselatus. The genetic uniformity in C. tesselatus 
could not exist if the species contained offspring of more than one 
parthenogenetic hybrid, since each individual hybrid resulting 
from a C. septemvittatus x C. tigris cross will carry a unique 
recombinant genotype. 

The genetic uniformity of C. tesselatus points to either one or 
both of two possibilities, namely that interspecific hybrids are rare 
or that parthenogenetic competence in an interspecific hybrid is 
rare. The first possibility cannot be ruled out in the case of C. 
tesselatus, as no evidence is available that might suggest the fre- 
quency at which hybrids between either C. septemvittatus and C. 
tigris or C sexUneatus and diploid C. tesselatus occur in nature. 
The first possibility can be eliminated in the case of C. "perplexus," 
the hybrid between C. inornatus and C. neomexicanus. C. "per- 
plexus" was first collected in New Mexico in 1841 (Maslin et ai, 
1958), and since 1962, at least six of these hybrids have been 
captured at sites where both C. inornatus and C. neomexicanus 



1971 TETRAPIOID CNEMIDOPHORUS 15 

are sympatric (Wright and Lowe, 1967a). This is a case where 
hybrids are rather common in nature and where they have oc- 
curred for at least 1 30 years without developing a parthenogenetic 
species. This case supports the view that genetic uniformity in 
existing parthenogenetic species and the absence of tetraploid 
species are both due to the rarity with which parthenogenetic 
competence is achieved in interspecific hybrids. 

The establishment of a parthenogenetic species in Cnemido- 
phonis may require a lengthy period of experimentation in which 
thousands or more individual hybrids are synthesized before a re- 
productively successful hybrid gene combination occurs. Never- 
theless, the result may still be termed saltatory speciation, in 
that the divergence of the new species from its progenitors is 
instantaneous, deriving as it does from a single, reproductively fit 
individual, rather than from cumulative changes in a population 
over long periods of time. It seems that both C. "perplexus" and 
the tetraploid hybrid illustrated here could represent previews of 
species that might eventually become established in the New 
Mexico deserts, if a gene combination facilitating parthenogenetic 
reproduction ultimately occurs in one of these hybrids. 

SUMMARY 

A tetraploid lizard resembling C. e.\.\anguis but bearing traits 
characteristic of C. itioniatus is considered to have resulted from 
a hybrid mating in which a haploid sperm pronucleus of C. 
inornatus fused with an egg pronucleus carrying the unreduced 
somatic chromosome complement of C. exsanguis, a triploid par- 
thenogenone. Production of such an egg by C. exsanguis may 
have relied on endomitosis in the oogonium, followed by normal 
meiosis operating on bivalents composed of paired sister chromo- 
somes. This modification of oogenesis is compatible with all 
known aspects of parthenogenesis in Cnemidupliorus, including 
the existence of apparently fixed heterozygosity within partheno- 
genetic species. 

Most cases of parthenogenesis in Cnemidophorus began with 
the synthesis of interspecific hybrids. However, several considera- 
tions suggest that many hybrids may be generated before a 
parthenogenetically competent individual, capable of giving rise 
to a species, is produced. Among these suggestive considerations 



16 BREVIORA No. 381 

are: 1) the synthesis of many hybrid individuals in nature, i.e., 
C. "perplexus," without the appearance of a corresponding 
species, and 2) the apparent genetic uniformity of parthenogenetic 
impedes, which indicates their origin from a single hybrid individual. 
Thus, the appearance of a new parthenogenetic species in Cnenii- 
dophorus may be preceded by a period of hybridization during 
which large numbers of reproductively incompetent prototypes 
of the new species are generated. 

ACKNOWLEDGMENTS 

This work was supported by Anatomical Sciences Training 
Grant GM 406 from the National Institutes of Health. The col- 
lection of animals was supported by the Evolutionary Biology 
Committee of Harvard University and by NSF Grant GB 3167. 
Drs. David Hamilton, Don Fawcett, and Ernest Williams provided 
generous assistance throughout the course of this study. William 
P. Hall offered many valuable comments on the manuscript. 

LITERATURE CITED 

Beatty, R. a. 1957. Parthenogenesis and Polyploidy in Mammalian 
Development. London, Cambridge University Press. 

CuELLAR, O. 1971. Reproduction and the mechanism of meiotic restitu- 
tion in the parthenogenetic lizard Cnemidophorous iiniparens. J. 
Morphology, 133(2): 139-165. 

Darevsky, I. S. 1966. Natural parthenogenesis in a polymorphic group 
of Caucasian rock lizards related to Lacerta sa.xicola Eversmann. 
J. Ohio Herp. Soc, 5: 115-152. 

Hall, W. P. 1970. Three probable cases of parthenogenesis in lizards 
(Agamidae, Chamaeleontidae, Geivkonidae). Experientia, 26: 1271- 
1273. 

Kallman, K. D. 1962. Gynogenesis in the teleost, Mollienesia formosa 
(Girard), with a discussion of the detection of parthenogenesis in 
vertebrates by tissue transplantation. J. Genetics, 58: 7-21. 

Lowe, C. H., and J. W. Wright. 1966. Evolution of parthenogenetic 
species of Cnemidophoriis (whiptail lizards) in western North 
America. J. Arizona Acad. Sci., 4: 81-87. 

Cole, C. J., and R. L. Bezy. 1970. Natural hybridization between the 
teiid lizards Cnemidophoriis sonorae (parthenogenetic) and Cnemi- 
dophoriis tigris (bisexual). Syst. Zool., 19: 114-127. 

MacGregor, H. C, and T. M. Uzzell. 1964. Gynogenesis in sala- 
manders related to Ambystoma jeffersonianitm. Science, 143: 1043. 



Errata for Breviora 381 by William B. Neaves, lines 35-37, 
page 16: 

r 1 Cole and R. L. Bezy. 1970. Natural 
genetic) and Cnemidophorus tigris (bisexuai;. y 



1971 TETRAPLOID CNEMIDOPHORUS 17 

Maslin, T. p. 1967. Skin grafting in the bisexual teiid lizard Ciicmi- 

dopJwnis sexUiicatus and in the unisexual C. tcssrldtus. J. Hxp. Zool., 

166: 137-150. 
, R. G. BiiDiJMAN, AND C. H. LowT. 1958. The status o'i 

the lizard Cucniidophonis perplcxus Baird and Girard (leiidae). Proc. 

U.S. Nat. Mus., 10«: 331-345. 

MOORHHAD, P. S.. P. C. NOWEIL, W. J. MtLLMAN. B. M. BaTTIPS, AND 

D. A. HuNGERFORD. 1960. Chromosome preparations of leucocytes 

cultured from human peripheral blood. Exptl. Cell Res.. 20: 613-616. 
N EAVES, W. B. 1969. Adenosine deaminase phenotypes among sexual 

and parthenogenetic lizards in the genus Cnemidoplioriis (Teiida;). 

J. Exp. Zool.. 171: 175-184. 
, AND P. S. Gerald. 1968. Lactate dehydrogenase isozymes 

in parthenogenetic teiid lizards {Cnciuidophonis) . Science. 160: 

1004-1005. 
Rasch, E. M., R. M. Darnell, K. D. Kallman, and P. Abramoff. 1965. 

Cytophotometric evidence for triploidy in hybrids of the gynogenetic 

fish. Puccilia fonuosa. J. Exp. Zool., 160: 155-170. 
ScHULTZ, R. J. 1967. Gynogenesis and triploidy in the viviparous fish 

Poeciliopsis. Science, 157: 1564-1567. 
Taylor, H. L., and P. A. Medica. 1966. Natural hybridization of the 

bisexual teiid lizard Cnemidophunts inornatus and the unisexual 

Cnemidophorus perple.xiis in southern New Mexico. Univ. Colorado 

Studies, ser. biol., 22: 1-9. 
Vanzolini, p. E. 1970. Unisexual Cnenudophonis Iciuniscatiis in the 

Amazonas Valley: a preliminary note (Sauria, Teiidae). Papeis 

Avulsos Zool., S. Paulo. 23: 63-68. 
Wright, J. VV., AND C. H. Lowe. 1967a. Hybridization in nature be- 
tween parthenogenetic and bisexual species of the whiptail lizards 

(genus CneiuidopJwnis). American Mus. Novitates, No. 2286: 1-36. 
, AND 1967b. Evolution of the alloploid par- 

thenospecies Cncniidophonis tesselatus (Say). Mamm. Chrom. 

Newslett., 8: 95-96. 
Zweifel, R. G. 1965. Variation in and distribution of the unisexual 

lizard Chemidcphoriis tesselaliis. American Mus. Novitates, No. 2235: 

1 49. 

Table 1 
chromosome number 

87 88 89 90 91 92 

frequency 2 113 1 

Table 1. Eight apparently complete chromosome spreads from MCZ 
101991, the tetraploid hybrid Cnemidophorus. were studied. The fre- 
quency with which various chromosome numbers were observed in these 
spreads is indicated in this table. 



18 



BREVIORA 



No. 381 




1971 TETRAPLOID CNEMIDOPHORUS 19 



Plate I 

Figure 1. A tetraploid hybrid (MCZ 101991) is shown between its 
suspected parental species. C. inonialiis (IN) and C. exsaiii^iiis (EX). 
Unlike typical C. exsaiti^iiis, the hybrid has fused paravertebral stripes and 
blue on its tail and ventral surfaces. C. inoruatiis also has vivid blue on 
these surfaces. 



971 TFTRAPIOID CNI-MIDC^PUORUS 21 



Plate II 

Figure 2. An apparently complete set of 92 chromosomes from the 
tetraploid hybrid (MCZ 101991) is shown. The technique for obtaining 
chromosome spreads is described in the text. 



22 



BREVIORA 



No. 381 



oo 



f 

i 

I 

I 
I 



IT 
i 

I 

t 

I 
t 






1 
€ 

I 






CO 



■c 



^ 



fcO 



'^'«» t:^ "^^ 



971 TETRAPI.Oll) CNEMIDOPHORUS 23 



4-" 

CD 



13 


^ 


c 




a 


-» 


OJ 




o 


*». 


(U 




C 


x> 




V5 










!^ 


V5 




C 








(D 








J3 




c 




CJ 


G 


(U 

E 




> 






rt 


<4-i 


0) 




j: 


O 


"H, 




V5 


c 


E 




E 

o 


E 


o 




n 


JD 


•a 




o 








E 

o 


D. 


^ 




E 


"5. 


. 


;. 


o 


CS 


c" 


j: 


o 


x: 


o 


CJ 










u 


j: 


."^ 


1> 


r- 


u 


V) 


sz 


C 




o 


H 


o 


i; 


D. 


c 


o 
E 


^c 


(U 


& 


o 


vo 


E 


o 


u. 


ON 








o 


ri 


•o 


x: 


C 
(U 


rl 


■_o 


£/J 


o 


tjj 


a. 




c 


£ 


J= 


" 


,-« 


^ 


a; 


C 


1) 

N 


. ™ 


x: 


^ 








c/: 


v5 




<u 




It 


Cfi 


O 


(L) 


E 


_c 


E 


o 

1/5 


S 


hJ 


o 

in 


o 


E 


^^ 


O 


E 


C/1 


E 


o 


c/5 




o 


4= 


'f-H 


OJ 


j: 


o 


J= 


CJ 




CJ 


yi 




E 
o 




!£ 


o 


U. 






« 


'-W-M 


C/5 


u« 








O 


't« 


T3 


c 
E 


4-. 


03 
Xi 










a 


E 
o 


o 

U4 


c 
o 


1) 


•a 


■a 


C/1 


O 


"H. 


sz 
■ '— 


y5 




i: 


Z 


sz 


















< 


^ 




o 


. 


1 


o 


_c 


rf^ 


X 


C/J 


■o 


1J 


U. 




."2 


^ 


o 


■5 


> 


£/j 


^4- 


o 








"O 


LJU 


o 


CJ 


J2 






24 



BREVIORA 



No. 381 




f t . 



/ 






8 



'..**•. 
•*«/.?> 



V -■»*;&■' 





9 



1971 TETRAPLOID CNEMIDOPHORUS 25 



Plate IV 

Figure 4. Courtship behavior between two C. inornutus. The male 
strokes the subject's neck and head with his jaw. 

Figure 5. Courtship behavior between a male C. inoniatiis and a 
parthenogenetic C. e.xsangiiis. The male executes sinuous movements 
against the dorsum of the female. 

Figure 6. As in Fig. 5. The male retreats along the female's dorsum, 
still executing sinuous movements while approaching the copuiatory 
position. 

Figure 7. As in Fig. 5. The male wedges his tail between the female's 
hind limbs and tail and attempts copulation. 

Figure 8. Courtship behavior between a male C. inornatus and a 
parthenogenetic C lesselatus. The male strokes the female's neck and 
head with his jaw. 

Figure 9. As in Fig. 8. The male executes sinuous movements against 
the dorsum of the female. 



>*. ■ 



^t 



^^ f^l^'S. CGMP. ZOOL 

UC.r.ARY 

B R E V I O "ft™l 

Muiseiiiim of Comparsitive Zoology 

Cambridge, Mass. 25 February, 1972 Number 382 

OSTEOLOGY AND DESCRIPTION OF THRATTIDION 

NOCTIVAGUS, A MINUTE, NEW FRESHWATER 

CLUPEID FISH FROM CAMEROON, WITH 

A DISCUSSION OF PELLONULIN 

RELATIONSHIPS 

Tyson R. Roberts^ 



Abstract. Thrattidion iwctivagiis. a new genus and species of Pellonu- 
linae, is described from the Sanaga River in Cameroon, and an account 
given of its osteology. Although it agrees with Congothrissa in having no 
supramaxillary bone, the two appear not to be closely related. The ecology 
of African freshwater Pellonulinae is reviewed. Feeding habits range from 
planktophagous to insectivorous and piscivorous, with piscivores feeding 
largely on other species of Pellonulinae. Most or all of the species undergo 
diel migrations, moving upwards or shorewards at nightfall in mixed ag- 
gregations with other small fishes, chiefly translucent schilbeid catfishes 
and cyprinids of the subfamily Bariliinae. 

Sierrathrissa Audenaerde (1969) is evidently the larva of either Pellonula 
or Cynothrissa. Congothrissa Poll (1964) is referred to the Pellonulinae. 
Relationships of round herrings (Dussumieriidae) that seem close to 
Pellonulinae are also discussed; Laevisciitella should be placed in the 
Pellonulinae. 

INTRODUCTION 

During fieldwork on the Sanaga River in Cameroon, I collected 
a minute clupeid which is clearly a member of the PellonuHnac but 
defies placement in any described genus. In lacking a supra- 
maxillary bone in the upper jaw it difl:ers from all other known 
clupeoids excepting Congothrissa (Poll, 1964) from the Congo 
basin. Consequently the distinctions between Congothrissa and 



1 Museum of Comparative Zoology, Cambridge, Mass. 02138 



2 BREVIORA No. 382 

other clupeids arc not so great as they at first appeared, and it 
may now be referred with some confidence to the PellonuHnae. 
Nevertheless these two minute clupeids are not more closely related 
to each other than to other forms, and their relationships within 
the PellonuHnae are unclear. The description of this new form 
from Cameroon is accompanied by an account of its osteology in 
the hope that this will contribute to its eventual placement in a 
phylctic classification. 

Acknowledgments. I wish to thank Mr. Thomason Newcomb 
for providing a congenial lodging place during my stay at Edea, 
and Prof. George S. Myers for reading the manuscript. The field- 
work was assisted financially by the Office of Environmental 
Sciences (contract no. RC 717634). 

THRATTIDION, new genus 

Type species. Thrattidion noctivagus, new species. 

Diagnosis. Minute (largest specimen 21.4 mm), largely trans- 
lucent Pellonulinae whh direct development. Four to seven keel- 
less prepelvic abdominal scutes and three to five (usually four) 
strongly keeled postpelvic abdominal scutes. All scutes with well- 
developed lateral (ascending) arms; pelvic scute with bifurcate or 
trifurcate lateral arms in larger specimens, 

Supramaxillary bones absent. Maxillary toothless. Branchi- 
ostegal rays 5,5 or 5,4. Lower limb of first gill arch with 10-11 
and upper limb with four short, widely spaced gill rakers. 

Anal fin considerably longer than dorsal, with 23-25 rays. 
Origin of dorsal fin about one-third of distance between verticals 
through pelvic and anal fin origins. Origin of anal fin on a 
vertical through base of eighth or ninth dorsal fin ray. Pectoral 
rays usually ten. 

Squamation restricted to a patch of about a dozen scales just 
behind head, a double row of scales on either side of prepelvic 
abdominal scutes, and a single row of scales on either side of 
postpelvic abdominal scutes. Vertebrae 45-47, about equally 
divided between abdominal and caudal (preural centrum 1 counted 
as last vertebra). Epurals three. 

Remarks. If the largest specimens in my material are fully 
grown, Thrattidion is the smallest clupcoid yet described. Next 
smallest is Congothrissa, with specimens recorded up to 27.7 mm 



1972 AFRICAN CLUPEID FISH 3 

(here and throughout this paper lengths referred to are standard 
lengths). Almost all African freshwater clupcids have nine or 
more prepelvic scutes and at least seven postpelvic scutes; in 
these forms the prepelvic as well as postpelvic scutes are invariably 
serrated or keeled. The only exceptions from freshwater in Africa 
are 1 ) Congothrissa, which lacks abdominal scutes entirely except 
for a pelvic scute; 2) Laevisciitella, with seven to eight keel-less 
prepelvic scutes and five to six keeled postpelvic scutes; and 3) 
Gilchristella, with six to nine unkeeled prepelvic scutes and no 
postpelvic scutes. The postpelvic scutes in Laevisciitella are 
unique in lacking lateral arms as adults (lateral arms are, how- 
ever, absent in the postpelvic scutes of larval Pelloniila). The 
lateral arms of the pelvic scutes are usually simple; they are 
bifurcate in Congothrissa (Poll, 1964, fig. 5 on p. 14). 

All previously described clupeoids have one or two supra- 
maxillary bones except Congothrissa. Pellonulinae characteris- 
tically have only one supramaxillary bone (Regan, 1917); its 
presence has been noted in alevins of Poecilothrissa moeruensis, 
Stolothrissa tanganicae and Microthrissa stappersii so small that 
the abdominal scutes are as yet unformed (Poll, 1964: 19). The 
supramaxillary is reduced in size in some Pellonulinae. The 
maxillary bone bears teeth in all other African freshwater clupeoids 
except Congothrissa and Potaniothrissa. Most African freshwater 
clupeids have 6,6 or more branchiostegal rays; Poecilothrissa, 
Microthrissa and Potamothrissa have as few as five; Congothrissa 
has 3,4 (here as elsewhere in this paper, the count on the left side 
is given first). All other African clupeoids have 16 or more gill 
rakers on the first gill arch, excepting Congothrissa, which has only 
seven on the lower limb and four on the upper limb of the first 
arch. 

All other African Pellonulinae have 23 or fewer anal rays, 
except Microthrissa, which has up to 25. Congothrissa has only 
15. In Congothrissa the dorsal fin origin is distinctly anterior to a 
vertical through origin of pelvic fins. In almost all African fresh- 
water clupeids the anal fin origin is far behind a vertical through 
base of last ray of anal fin, the exceptions being larval Pelloniila, 
in the metamorphosis of which the dorsal fin migrates forward, 
and Microthrissa and Gilchristella, in which the anal fin origin is 
on or only slightly posterior to a vertical through the last ray or 



BREVIORA 



No. 382 



two of the dorsal fin. In larval Pellonula the anal fin origin may 
be as far forward as posterior third of dorsal fin. All other 
African freshwater clupeoids have 1 1 or more pectoral rays. 

In all other African clupeoids the body is completely scaled in 
adults; the highest vertebral count previously recorded is 44, with 
abdominal vertebrae usually more numerous than caudal (Poll, 
1964, table 1). Congothrissa has 21+20 vertebrae. The number 
of epurals is unrecorded for most Pellonulinae; Congothrissa and 
larval Pellonula have two epurals. 

Etymology. Thrattidion, Greek, neuter diminutive of Thrassa 
(Thratta), feminine, a small, herringlike fish. 

Thrattidion noctivagus, new species 
Figures 1 and 2; Table 1 

Holotype. MCZ 48161, 21.4 mm, Sanaga River at Edea, 

Cameroon, 4 May 1971. 
Paratypes. MCZ 48162, 175 specimens, 9.8-20.7 mm (of which 

11 specimens 10.4-20.7 mm cleared and stained), Sanaga 

River at Edea, 17 April and 4 May 1971. 

The specimens are in excellent condition. They were fixed in 
formahn immediately after capture and placed in Nalgene plastic 
bottles containing only small specimens and filled to the brim to 
prevent damage by shaking. Consequently, the fins are in perfect 
condition on almost all specimens, and the squamation completely 
intact. After three months in formalin they were soaked in water 
for two days, in 40 percent ethyl alcohol for two more days, then 
transferred to 70 percent ethyl alcohol for permanent preservation. 

General characteristics. Snout short and rounded. Jaws equal. 
Body relatively compressed, depth 3.55-5.0, increasing with size. 




Figure 1. Thrattidion noctivagus, 19.4-mm paratype. MCZ 48162. 



1972 AFRICAN CLUPEID FISH 5 

Eye large, 8.5-9.9, its center on or only slightly above longitudinal 
midline of body (center of eye distinctly above midline in most 
Pellonulinae, markedly so in OdaxotJirissa and Cynothrissa) . 
Adipose eyelid well developed. Dorsal fm origin distinctly pos- 
terior to a vertical through pelvic origin, slightly closer to end of 
hypural fin than to snout tip. Anal fin long, its origin on a vertical 
through base of eighth or ninth dorsal fin ray. 

Pelvic, dorsal, and anal origins on verticals through 15th, 19th, 
and 25th vertebrae. Pelvic fin relatively large, only slightly 
smaller than pectoral, its origin closer to pectoral fin origin than 
to anal fin origin. Dorsal and anal fin margins gently falcate, 
dorsal height about 1.5 times anal height. Caudal fin deeply 
forked, upper and lower caudal lobes identical in size and shape. 

For proportional measurements see Table 1. 

Fin counts (Table 1). Dorsal fin with 14-16 rays, first three 
or four simple, last divided to base. Anal fin with 23-25 rays, 
first four or five simple, last divided to base. Length of first dorsal 
and anal rays variable, minute in some specimens. Pectoral fins 
usually with ten rays. Pelvic fins with either seven or eight rays. 
Principal caudal rays invariably 10+9, upper procurrent rays 
12-13 and lower procurrent rays eight to ten. 

Squamation (Fig. 1). Scales largely absent, restricted to a 
double row of four to nine scales superficial to lateral arms on 
each side of prepelvic scutes, a single row of four or five scales 
superficial to lateral arms on either side of postpelvic scutes, and 
about a dozen scales in four or five short rows immediately behind 
head. The latter scales are deeply embedded in tough tissue in 
which the trunk portion of the cephalic laterosensory system 
ramifies, and can only be fully discerned in cleared and stained 
specimens. A 20.7-mm specimen has 12 scales in four rows 
(three in the uppermost row, five in the second row, one in the 
third row, and three in the lowermost row); a 19.1-mm specimen 
11 scales in five rows (2-4-1-3-1); and an 18.6-mm specimen 
11 scales in five rows (1-2-4-2-2). Most of the scales in the 
longest row are perforated by laterosensory canals. Larval 
Pellonula of 36-38 mm have the body naked except for two or 
three scales immediately behind the head and complete squamation 
on the caudal peduncle. 



BREVIORA 



No. 382 



•11 



-^ o V 

~ d r^ ** "* 

i -r ;. f -r 

CT- O- O M *n 

H CD « in ^ 

o- o o- 



B - * 









~o to -o 



|l 



S s t ; 



1972 



AFRICAN CLUPEID FISH 



Cephalic laterosensory system (Fig. 2). Cephalic latcrosensory 
system with supraorbital, infraorbital, mandibulo-preopcrcular, 
pterotic, opercular, occipital and lateral canals (cf. Wohlfahrt, 
1937). A supratemporal branch arises from the supraorbital; in 
some specimens it appears to enter the occipital canal, but in 
others it definitely stops short of reaching the occipital canal. 
Figure 2 is based on a 15.1 -mm specimen that had a drop or two 
of Ehrlich's hematoxyUn placed on its head. Of several specimens 
thus treated, its canals showed up the best. The mandibular 
portion of the mandibulo-preopercular canal could not be dis- 
tinguished on this specimen, but it appeared in others. The main 
canals appear to join in the manner illustrated, but the figure 
should be used with caution. The canals did not take up stain 
evenly, and comparisons with less well-stained specimens did not 
ehminate my reservations about the interpretation. I could find 
no connections between the laterosensory systems of the two 
sides of the head. In particular the ethmoidal commissure is 
absent. 

Dentition. Premaxillary with four to seven small conical teeth 
soUdly implanted, and two to four replacement teeth in the gum. 



supraorbital 



supratemporal 



occipital 




lateral 



infraorbital 



preopercular 



opercular 



Figure 2. Thrattidion noctivagiis, cephalic laterosensory canals of 
15.1 -mm paratype. 



8 BREVIORA No. 382 

Maxillary edentulous. Dentary with a dozen to 15 small conical 
teeth solidly implanted and several teeth in stages of replacement. 
Palatine with one to three teeth, roof of mouth otherwise edentu- 
lous. Fifth ceratobranchial and upper pharyngeal with numerous 
small teeth, gill arches otherwise edentulous. 

Abdominal scutes. Four to seven prepelvic scutes, unkeeled, 
with well-developed lateral arms. Pelvic scute keel-less, with 
bifurcate or trifurcate lateral arms in largest specimens. Post- 
pelvic scutes three to five, usually four, with well-developed keels 
and lateral arms. Specimens as small as 12-14 mm with full 
complement of scutes (Table 1). 

Coloration of preserved specimens. Body opaque white, with- 
out mclanophores, or with only a few melanophores in some 
specimens at base of anal fin rays and on caudal peduncle. No 
lineum argenteum. Peritoneum probably opaque white and body 
otherwise translucent in living specimens. A dense cluster of 
large melanophores over brain. Specimens were collected only 
at dusk or at dawn and immediately preserved, so observations 
were not made on their appearance in life. 

Additional characteristics. See generic diagnosis and oste- 
ological account. 

OSTEOLOGY 

Neurocranium (Figs. 3-5). Anterior and posterior frontal 
fontancUes very large. Anterior margin of anterior frontal fonta- 
nelle formed by diverging posterior arms of mesethmoid. Pos- 
terior frontal fontanelle incompletely divided into right and left 
halves, the median extension of supraoccipital failing to reach 
posteromedial border of frontal bones. Of the types of clupeoid 
posterior frontal fontanelles figured by Whitehead, those of 
Spratelloides and Ehirava (Whitehead, 1963a, fig. 1) resemble 
most closely that of Thrattidion. As in Ehirava, a median exten- 
sion of the frontals forms a distinct wedge in the posterior frontal 
fontanelle. Frontals slightly separated at midline of skull. 
Anterior fontanelle about half as wide as posterior fontanelle. 

Mesethmoid solidly ossified, with well-developed anterolateral 
and descending processes, and diverging posterior processes firmly 
united to anterior ends of frontal bones. Lateral ethmoid with a 
large cartilaginous portion ventrally. 



1972 



AFRICAN CLUPEID FISH 



Pre-epiotic fossa large, an anteroventrally directed process of 
parietal bone only partially separating it from temporal foramen. 
Pre-epiotic fossa bordered mainly by frontal and parietal; temporal 
foramen bordered by parietal, cpiotic and pterotic (Fig. 3). 



orbitosphenoid 
mesethmoid \ 


frontal 


supraoc 


:ipital 


preiuaxillary \ ^^^^r^"! '•■'•'. 


•;-/<i_ " — -^==»5 


, parietal 


W^f{f^=^^ 


^SS 








v\ AJf 


Q>Aw«;v 




=- pterotic 


maxillary / /Ntv 

vomer / / ^^^j-^ 


//iw 




V basioccipital 


cartilage / /~'^^^<>--~/ 


/i%i^^^ 


\ \ 






^ 1, '^.j. ■ - \ V ■ \^ — V" 


exoccipital 


lateral ethmoid / / 


/ ~\ V 




parasphenoid / 


/ sphenotic 


^ sagitta 


pterosphenoid 


basisphenoid 


prootic 
1 1 mm 


1 



Figure 3. Cranium of 18.6-mm Thrattidion noctivagiis, lateral view. 



parasphenoid 
maxillary 
premaxillary 

dentary 



lateral ethmoid 

entopterygoid 



frontal 



parietal 




supraoccipital 



exoccipital 



basioccipital 



mesethmoid 

nasal 



palatine \ \ ^'^'^--i'-'^—V' epiotic 

supraorbital orbitosphenoid pterotic 



( ^ '""^ I 



Figure 4. Cranium of 20.0-mm Thrattidion noctivagiis, dorsal view. 



10 



BREVIORA 



No. 382 



A large foramen bordered by the frontal, pterotic, and prootic 
opens into the recessus lateralis (Fig. 3). It seems likely that all 
of the main branches of the cephalic laterosensory system enter 
the recessus through this one large opening. A laterally directed 
ridge on prootic extending onto anteroventral portion of pterotic, 
both bones apparendy contributing to this ridge posteriorly. 

Intcrcalar bone present, a small, oval bone lying in the interos- 
seous membrane between exoccipital, pterotic, and prootic, its 
edges touching each of these bones (Fig. 5). When the pectoral 
girdle is removed from the skull, the intercalar is invariably 
carried away with the lower limb of the posttemporal bone. 

I am unable to find a passageway through which anterior exten- 
sions of the swim bladder might gain the prootic bulla. A small 
oval opening in the exoccipital bone opens into a passageway 
directed dorsoposteriorly, away from the prootic. This opening 
and its passageway do not have counterparts in the prootic. The 
opening is flush with the exoccipital, not funnel-shaped as one 
would expect of an opening for the swim bladder; it probably 
serves for passage of cranial nerves. A large dorsal foramen of 
the prootic is readily visible. 



parasphenoid pterosphenoid 

orbitosphenoid \ i , . , 

\ basisphenoid 

lateral ethmoid 

vomer 

cartilage 
maxillary 



intercalar 



sagitta 




premaxillary 



nasal 
mesethmoid 
supraorbital 



frontal 
sphenotic 



basioccipital 
exoccipital 
pterotic 



prootic 



1 mm 



Figure 5. Cranium of 18.6-mm Thratddion noctivagus, ventral view. 



1972 



AFRICAN CLUPEID FISH 



11 



Orbitosphenoid, pterosphenoids, and basisphenoid well de- 
veloped, with large lateral foramina. Orbitosphenoid with an 
anteriorly directed process. 

Vomer edentulous, elongate, and flattened. Parasphenoid uni- 
formly slender, dividing where it meets base of neurocranium into 
a pair of posterior processes, the tips of which extend beyond 
occiput to immediately below second or third vertebral centrum 
(Fig. 5). 

Jaws (Figs. 6-7). Premaxillaries separated at midline, each 
bearing four to seven conical teeth. Maxillary relatively slender, 
extending to below anterior third of eye, toothless. Supra- 
maxillaries absent. Dentary with a low coronoid process and a 
single row of about a dozen to 15 conical teeth. Articular with a 
posteriorly directed process posterior to quadrate socket. 

Siispensoriiim (Fig. 7). Palatine bifid at its tip, bearing one 
to three conical teeth. Suspensorium otherwise edentulous. 
Ectopterygoid slender anteriorly, moderately expanded where it 



supraorbital 



dermosphenotic 



antorbital 




opercle 



premaxillary 
dentary 
maxillary 



subopercle 



interopercle 
articular' quadrate symplectic preopercle 



I ^ ""° 1 



Figure 6. Jaws and facial bones of 20.7-mm Thratddion noctivagus, 
lateral view (1-5 = circumorbital bones). 



12 



BREVIORA 



No. 382 



premaxillary 
maxillary 
palatine 




opercle 



dentary 



coronomeckelian 

articular 



angular quadrate j ijiterhyal 

symplectic 



subopercle 

preopercle 
interopercle 



1 mm 



Figure 7. Jaws, suspensorium, and opercular bones of 20.7-mm Thrat- 
tidion noctivagus, medial view. 



attaches to dorsoanterior portion of quadrate bone. Entoptery- 
goid and metapterygoid well developed. Sphenotic and pterotic 
articular surfaces of hyomandibular not strongly differentiated 
from its main body. Hyomandibular with a distinctively shaped 
lamella dorsoanterior to its symplectic process, the ventral margin 
of this lamina separated from symplectic process of hyomandibular 
by a notch. Symplectic bone relatively small, lying midway be- 
tween hyomandibular and quadrate but in contact with neither. 
Quadrate with a posterior process directed towards anterior end 
of symplectic. 

Facial bones (Fig. 6). Circumorbital series with an antorbital 
followed by six infraorbitals, a pattern found in generalized mem- 
bers of many teleostean groups, including Clupeomorpha (Nelson, 
1969). Dermosphenotic greatly reduced. Infraorbitals shallow, 
without extensive ventral laminae. Supraorbital bone large and 
elongate. 

Nasal bones widely separated, moderate in size, bearing a tube 
for laterosensory canal. 



1972 



AFRICAN CLUPEID FISH 



13 



dorsal hypohyal ceratohyal 



epihyal 





ventral hypohyal 



branchiostegal rays 1-5 



1 mm 



Figure 8. A. Urohyal bone (upper: dorsal view; lower: left lateral 
view); and B. hyoid arch (basihyal and interhyal removed), lateral view, 
of 20.7-mm Thraitidion noctivagus. 



ventral hypohyal 
dorsal hypohyal , 
ceratohyal 
epihyal 
interhyal 



pharyngobranchials 1-3 



upper pharyngeal 




basihyal 



basibranchials 1-3 



hypobranchials 1-3 



ceratobranchials 1-5 



epibranchials 1-4 



1 mm 



Figure 9. Hyoid arch (branchiostegal rays removed) and branchial 
arches of left side, of 20.7-mm Thrattidion noctivagus. Dorsal view, upper 
limb of gill arches unfolded and pinned to the side. 



14 BREVIORA No. 382 

Opercular series complete; dorsoanterior corner and posterior 
margin of opcrcle deeply incised. 

Hyoid and branchial arches (Figs. 8-9). The branchial skele- 
ton of Thrattidion differs from that in most clupeids in several 
respects; this is perhaps mainly due to its reduced size. Gill rakers 
and pharyngeal dentition reduced. Branchial skeleton lacking 
elongation of individual parts typical of most living clupeoids (and 
which is at least partly correlated with increased numbers of gill 
rakers and branchiostegal rays). Basihyal well ossified and tooth- 
less (usually cartilaginous and tooth-bearing in Pellonulinae, 
Nelson, 1967: 393). Pharyngeal teeth restricted to upper 
pharyngeal and ceratobranchial 5. Gill rakers few in number, 
widely spaced, and relatively short. The 20.7-mm specimen rep- 
resented in Figure 9 has 15 rakers on the first gill arch, including 
a few that are rudimentary or unossified. Gill rakers progressively 
reduced in size and number on arches 2-5. 

The hyoid arch has paired ventral and dorsal hypohyals, 
ceratohyal, epihyal and interhyal, and a median basihyal. 
Branchiostegal rays 5,5 in eight specimens, 5,4 in two; generally 
one branchiostegal ray attached to epihyal, three to ceratohyal, 
and one branchiostegal ray in between, not attached to either 
bone. 

Individual ossified parts of the branchial arches are median 
basibranchials 1-3, and paired hypobranchials 1-3, cerato- 
branchials 1-5, epibranchials 1-3, suspensory pharyngeals or 
infrapharyngeals 1-3, and upper pharyngeals. 

Pectoral girdle (Fig. 10). Two slender postcleithra, distal end 
of second postcleithrum attached by ligament to distal end of 
first rib. Four proximal radials and about six distal radials (distal 
radials absent in clupeoids other than Chirocentridae, according 
to Whitehead, 1963a and Greenwood, 1968). Other bones in 
pectoral girdle: extrascapular, posttemporal, cleithrum, coracoid, 
scapula, mesocoracoid. 

Pelvic girdle (Fig. 11). Pelvic bones relatively elongate and 
slender. Articular portion of pelvic bone with a deep, elongate 
notch. Three pelvic radials. Pelvic splint present (pelvic splint 
absent in clupeoids, according to Whitehead, 1963a and Green- 
wood, 1968). 



1972 



AFRICAN CLUPEID FISH 



15 



posttemporal 



supracleithrum 
postcleithra 1-2 



\ 

cleilhrum intercalar 




extraseapular 




distal radials 



proximal radials 



mesocoracold 



curacoid 



Figure lU. Pectoral girdle of 2U.7-mm Thratiidion noctivagus, medial 
view. Inset, upper right hand corner: extraseapular bone of same speci- 
men, lateral view. 



pelvic splint 




pelvic radials 



pelvic bone 



Figure 11. Pelvic girdle of 18.6 mm Thraiiidion noctivagus, dorsal 
view. 



Abdominal scutes (Fig. 12). Prepelvic scutes four to seven, 
keel-less, with slender lateral arms. Pelvic scute enlarged, with 
bifurcate or trifurcate lateral arms in larger specimens (specimen 
figured has two branches on one side, three on the other). Post- 
pelvic scutes usually four, invariably with well-developed keels. 
Specimens only 12.1 and 14.3 mm have a full complement of 
pelvic scutes (Table 1). In a 10.4-mm specimen only the pelvic 
scute is visible. In a 12.1 -mm specimen the scutes are repre- 
sented only by the lateral arms, which ossify separately. With 



16 



BREVIORA 



No. 382 



growth they fuse to form a median scute. In the middle of the 
scutes in larger specimens is sometimes a foramen or gap in the 
ossification where the two halves are incompletely fused, as in the 
second three prepelvic scutes in the 20.7-mm specimen figured. 
Vertebral column and associated bones (Fig. 13). In a series 
of ten specimens the vertebral count ranges from 45 to 47; 
abdominal and caudal vertebrae about equal in number. The 
counts and their frequencies are 22+23 (1); 23+22 (3); 23+23 
(3); 23+24 (2); and 24+23 (1). The third vertebra is the first 
to bear ribs. The tips of ribs 4-7 to 2-9 are bound by connective 
tissue to the lateral arms of the prepelvic scutes; rib 9 or 10 to 
the pelvic scute, and ribs 10 or 11 through 15 to the postpelvic 
scutes. 



pelvic bone 



IJelvlc splint 




postpelvic scutes 



prepelvic acutos pelvic scute 



Figure 12. Abdominal scutes of 20.7-mm Thrailiclion noctivagus, ventral 
view. 



There are four distinct sets of intermuscular bones: 1) epaxials 
or epincurals, 2) epicentrals, 3) anterior hypaxials or epipleurals, 
and 4) posterior hypaxials. The epaxials begin at about the 
eighth vertebra and continue to the second preural centrum. The 
epicentrals begin at the first vertebra and continue to the second 
preural centrum. The epicentrals begin at the first vertebra and 
extend to about the first caudal vertebra. The anterior hypaxials 
or epipleurals begin at the second vertebra and extend to the last 
abdominal vertebra. The posterior hypaxials begin at about the 
fourth or fifth caudal vertebra and extend to about the second 
preural centrum. All of the intermuscular bones are relatively 
simple; in particular the epicentrals, epaxials, and posterior 



1972 



AFRICAN CLUPEID FISH 



17 




Figure 13. Axial skeleton of 18.6-mm Thrattidion noctivagits, lateral 
view. 



hypaxials are unbranched. The anterior hypaxials are almost as 
long as the ribs and are expanded near their proximal ends into a 
triangular lamella. 

There are ten to twelve supraneural bones. The first supra- 
neural lies just dorsal to the first neural spine; the last lies over 
the tenth to twelfth neural spine. 

Caudal skeleton (Fig. 14). Caudal skeleton with six hypurals, 
hypural 1 separated from ural centrum 1 by a distinct gap. 
Preural centrum 1, ural centrum 1 and ural centrum 2 fused. 



epurals 1-3 



urostyle 




parhypural 



Figure 14. Caudal skeleton of 20.7-mm Thrattidion noctivagits, lateral 
view (H^-H^ = hypurals; P^ + U^ + U., = fused preural centrum 
1, ural centrum 1 and ural centrum 2; P., = preural centrum 2). 



18 BREVIORA No. 382 

Hypural 2 fused with ural centrum 1. Parhypural (= hemal spine 
of preural centrum 1 ) closely articulated with centrum of preural 
centrum 1, its dorsal margin with a very large, laterally directed 
flange or hypurapophysis. The single uroneural Ues free against 
the ural spine and does not extend anteriorly to the ural centrum 
1. Epurals three. 

Caudal fin lobes almost perfectly symmetrical. Principal caudal 
rays invariably ten in upper lobe and nine in lower lobe. Speci- 
mens 14 mm upwards have 12 or 13 upper and eight to ten lower 
procurrent rays, all of which take up alizarin stain. Hypurals 1 
and 3 expanded, hypural 2 slender. Hypural 1 articulating with 
six principal rays, hypural 2 with two, and hypural 3 with five. 

DISCUSSION 

Ecology oj African Pellonulimie. Relatively little has been 
published on the ecology of African Pellonulinae. Poll (1953) 
and Coulter (1962) gave some information about Limnothrissa 
and Stolothrissa, the two genera endemic to Lake Tanganyika 
(Tanganyika is the only East African lake with endemic clupeids). 
Reynolds (1970; 1971) reported on feeding habits and schoohng 
and migration of Pellonula and Cynothrissa in man-made Volta 
Lake in Ghana. The ecology of Pellonulinae in the Congo basin 
is unknown, excepting brief remarks on food or general habitat by 
Gosse (1963) and Matthes (1964). 

Stolothrissa feeds on zooplankton. Juvenile Limnothrissa feed 
on zooplankton, but adults are macrophagous and evidently feed 
to some extent on larval Stolothrissa. Stolothrissa are apparently 
the only food of the endemic centropomid Luciolates (Poll, 
1953). Some riverine Pellonulinae are insectivorous to a greater 
or lesser extent. Mature Povilla (Ephemeroptera) nymphs were 
the dominant item in the food of Pellonula in Volta Lake, with 
terrestrial insects and aquatic stages of Diptcra contributing sig- 
nificantly to the rest of the food taken by them (Reynolds, 1970, 
table 1 on p. 578). Of the Congo species, Potamothrissa obtusi- 
rostris feeds largely on aquatic insects, Nannothrissa parva on 
phytoplankton (unicellular algae, diatoms) and zooplankton 
(entomostracans), and Odaxothrissa losera on shrimps, Povilla 
nymphs, and small fish (mainly other clupeids) (Matthes, 1964). 



1972 AFRICAN CLUPEID FISH 19 

The West African Cynothrissa (which superficially resemble^ 
Odaxothrissa) is piscivorous, feeding largely on Pellonula in Volta 
Lake (Reynolds, 1970). 

Reynolds (1971) found that Pellonula was among several small 
species in Volta Lake that migrate upwards land shore wards?] 
at night in mixed aggregations, and compared this migration to 
that of Stolothrissa and Limnothrissa in Lake Tanganyika. It 
should be noted that a similar phenomenon occurs in the Pellonu- 
linac of the Congo basin. Poecilotlirissa, Microthrissa royauxi, 
Potamothrissa, and Odaxothrissa evidently migrate upwards or 
shorewards at dusk in aggregation with other small fishes, prin- 
cipally the cyprinids Chelaethiops, Engraulicypris and Leptocypris, 
and the schilbeids Parailia and Pareutropiiis. These aggregations 
apparently do not remain close to the shore all night long, but 
reform there for another short period around dawn. Reynolds 
(1971) discusses the selective advantages of such diel migrations, 
which are evidently primarily improved feeding opportunities, 
coupled with predator avoidance and possibly a metabolic ad- 
vantage in living at lower temperatures and feeding at higher ones. 

Thratiidion was collected only at dusk and dawn, by dipnetting 
in water two to three feet deep flowing swiftly along the low 
retaining wall that extends for about a mile above the hydroelectric 
dam on the Sanaga River at Edea. The only other fishes caught 
with it were small characids, tentatively identified as juvenile 
Alestes and Virilia. Its stomach contents included much un- 
identified material and some terrestrial insects (either Diptera or 
Hymenoptera) (kindly identified by Dr. David Rentz). 

The larval stage of Pellonula. The only other species of 
Clupeidae known or likely to occur in the Sanaga River much 
above its mouth are Pellonula and Cynothrissa. Although adults 
of these genera can be distinguished, they evidently are closely 
related. Distinguishing characters for the juveniles have not been 
reported. Svensson (1933) described larvae of Pellonula from 
the Gambia River. I have collected similar juveniles in the Tano 
River and in the Sanaga River along with adults of Pellonula, and 
am inclined to believe that they are the same species. They are 
radically different from Thrattidion. The larvae and juveniles of 
Cynothrissa are yet to be described; presumably they are similar 
to those of Pellonula. There can be little doubt that Sierrathrissa 



20 BREVioRA No. 382 

(Audcnaerde, 1969), from Sierra Leone, is based on the larvae 
of either Pellonida or Cynothrissa. My specimens of larvae from 
the Tano and Sanaga rivers agree in almost all respects with the 
figure and description of Sierrathrissa except that they have more 
vertebrae (24+17 instead of 19+17) and are incompletely scaled 
(scales present only immediately behind head and on caudal 
peduncle). Audenaerdc reported "approximately 37-38 scales 
in a longitudinal line, about one scale row for each body segment, 
scales hardly visible on posterior part of body in largest speci- 
mens." Svensson (1933: 47-48, fig. 16) describes the larva of 
Pellonida (identified as P. vorax) and its metamorphosis. His 
description of the larva agrees very well with that of Sierrathrissa. 
Concerning the metamorphosis, we may quote Svensson: 

When the fry is about 30 mm. long, an obvious and rapid change takes 
place in its appearance. The transparency disappears, and pigment com- 
mences to develop on the back; the head and the anterior part of the 
body grows thicker and higher, and the dorsal fin advances forwards so 
that its front part becomes situated above, or slightly behind, the vertical 
of the origin of the ventral fins. The gill-rakers increase in number to 25 
(they are 25-30 in the adult fish, according to Boulenger) and the num- 
ber of rays in the pectoral and ventral fins, as well as in the branchiostegal 
membrane, increases to that in the adult. It may be presumed that those 
changes represent a kind of metamorphosis, corresponding to that in 
other Cliipeids, and in Elopids, and Albiilids. 

Svensson found that larval Pellonida have nine to ten pectoral 
rays, seven pelvic rays, rudimentary abdominal scutes, about 14 
gill rakers on lower part of first arch, and only three branchi- 
ostegal rays visible (in specimens stained with alizarin). He did 
not mention the presence of scales. Adult Pellonida have 14 
pectoral rays, eight pelvic rays, 12-14 prepelvic and eight to nine 
postpelvic abdominal scutes, all strongly serrate and with lateral 
arms, 27-32 gill rakers on the first gill arch, and six branchiostegal 
rays. Despite statements to the contrary (Ridewood, 1904; Poll, 
1964, table 1 opp. p. 29), adult as well as larval Pellonida bear 
numerous fine teeth on the maxillary bone (Regan, 1917; personal 
observation). The supramaxillary bone is well developed in both 
adults and larvae. 

Concerning other African Pellonulinae, Limnothrissa and Stolo- 
thrissa undergo profound changes in morphology from eight or 
ten to 25 mm, including considerable forward migration of the 



1972 AFRICAN CLUPEID FISH 21 

dorsal fin (Poll, 1953). Microthrissa miri, from the Niger and 
Chad basins, has a slender larva in the metamorphosis of which 
the dorsal fin position changes relatively little (Daget, 1954: 
66-67, fig. 12; Blache, 1964: 58-59, fig. 21 on p. 418). 
Development is direct or nearly direct in at least some riverine 
Pellonulinae of the Congo basin. Thrattidion has direct develop- 
ment: specimens as small as 12 and 14 mm have the main features 
of the larger specimens, including full or nearly full complements 
of abdominal scutes, scales, fin rays, branchiostegal rays, gill 
rakers, and teeth. Fin placement remains unchanged. The most 
notable change is a regular increase in depth of body from 
4.7-5.0 in 10-1 1-mm specimens to 3.55-4.1 in 19-21-mm 
specimens (Table 1), at which size Thrattidion is deeper bodied 
than any other African freshwater clupeid. 

Systematic status of Congothrissa. In erecting a new family 
for Congothrissa, Poll (1964) stressed its 1) lack of prepclvic 
and postpclvic abdominal scutes; 2) unkeeled pelvic scute [with 
branched lateral arms]; 3) terminal mouth; 4) lack of supra- 
maxillary; 5) short lateral [medial?] branch of supraorbital canal; 
6) relatively large scales; and 7) low branchiostegal ray count. 
Characters 1,4, 6, and 7 are loss or reduction characters such as 
one commonly finds in teleosts of greatly reduced size. Reliance 
on such characters leads to classifications in which the smallest 
representatives of a group are placed in either monotypic or poly- 
phyletic taxonomic categories. Reduction of scutes has occurred 
in many lines of clupeoids. Thrattidion has also lost the supra- 
maxillary. The branchiostegal ray count of Thrattidion (5,5 or 
5,4) is intermediate between that of Congothrissa (3,4) and other 
Pellonulinae. Congothrissa has about 24 scales in a lateral series 
(Poll, 1964, table 1; fig. 1 opp. p. 8). Judging from the figure 
of Congothrissa, it may have 25 scales in a lateral series, since 
there are usually deeply embedded scales immediately behind the 
head that are not externally visible in Pellonulinae. Some speci- 
mens of Poecilothrissa have as few as 28 scales in a lateral series 
(personal observation). An unkeeled pelvic scute with branched 
lateral arms similar to that in Congothrissa is found in Thrattidion. 
I find nothing in the nature of the supraorbital canal of Congo- 
thrissa to distinguish it from other Pellonulinae. A terminal 
mouth occurs in Thrattidion and in some of the Congo Pellonu- 
linae. 



22 BREVIORA No. 382 

In conclusion, Congothrissa does not merit placement in a 
family by itself. It may either be placed in a separate subfamily, 
Congothrissinae, or preferably in the Pellonulinae. Its closest 
relatives are to be sought among the Pellonulinae in the Congo 
basin. 

Systematic status of Thrattidion. Several of the distinguishing 
characteristics of Thrattidion — absence of supramaxillary, tooth- 
less maxillary, reduced squamation and scutellation, and low 
branchiostegal and gill raker counts — are loss or reduction 
characters more or less shared with Congothrissa. On the other 
hand, Thrattidion has more vertebrae (45-47), anal rays 
(23-25), and epurals (three) than are recorded for almost any 
other African Pellonulinae {Microthrissa royauxi has as many as 
24 or 25 anal rays), and is relatively deeper bodied than any of 
the others at comparable size. If Congothrissinae is recognized, 
then a separate subfamily might also be justified for Thrattidion. 
It is preferable to place them both in the Pellonulinae. 

Relationships of Dussumieriidae. Whitehead (1963b) ad- 
vanced a theory with important implications for clupeoid phylo- 
geny, namely, that Dussumieriidae are modern representatives of 
a primitive group of scuteless clupeoid. This theory is strongly 
contraindicated by the following considerations: 

1 ) All Dussumieriidae have at least a pelvic scute that is 
apparently homologous with the prcpelvic and postpelvic scutes 
in other clupeoids. 

2) Several dussumieriids have either prepelvic or postpelvic 
scutes or both: Etrumeus (subfamily Dussumieriinae) has a post- 
pelvic scute immediately behind, and partially overlapping, the 
pelvic scute (Chapman, 1948, fig. 17 of E. micropus on p. 38; 
personal observation); Gilchristella and Ehirava mulabaricus 
(subfamily Spratelloidinae) have prepelvic scutes, unkeeled but 
with lateral arms, and no postpelvic scutes (Whitehead, 1963b; 
and Poll, Whitehead, and Hopson, 1965: 286); and Laeviscutella, 
which Poll, Whitehead, and Hopson placed in the same tribe with 
Gilchristella and Ehirava, has both prepelvic and postpelvic 
scutes. In Laeviscutella the prepelvic scutes have lateral arms 
but are unkeeled, whereas the postpelvic scutes are keeled but 
lack lateral arms, a condition unlike that in any other clupeoid. 

If the absence of prepelvic and postpelvic scutes is a primitive 



1972 AFRICAN CLUPEID FISH 23 

character in those Dussumieriidae hacking them, then such scutes 
must have originated independently in several lines of clupeoids, 
including Indo-Pacific Engraulidae and two subfamilies of Dus- 
sumieriidae. It seems more likely that the ancestors of living 
clupeoids had a more or less full complement of scutes, and that 
they have been reduced or lost to varying degrees many times. 
A strong trend to reduce scutellation is evident in Pellonulinae. 
So far as known, the pelvic scute is retained in all living clupeoids. 
Whitehead (1963b) suggested that the pelvic scute of Dussumi- 
eriidae originated from pelvic splint bones, independently of pre- 
pelvic and postpelvic scutes. Greenwood (1968: 256) rejected 
a pelvic spUnt origin for the pelvic scute, largely on the ground 
that the pelvic scute is morphologically so similar to the other 
abdominal scutes. In Thrattidion the ontogeny of the pelvic scute 
is similar to that of the others, and the pelvic splint bone (usually 
absent in clupeoids) is quite separate. 

While placing Laeviscutella in the dussumieriid tribe Ehiravini 
on account of its reduced scutellation, Poll, Whitehead, and 
Hopson (1965: 286-288) also cited reasons why it might be 
placed in the Clupeidae, and specifically in Pellonulinae. Daget 
and litis (1965: 45) placed it in the Clupeidae because "the only 
known species of Laeviscutella is in reality very close to the forms 
of Pellonula with which it is easily confused." In my opinion 
Laeviscutella should be placed in the Pellonulinae. Other genera 
currently placed in the Ehiravini also may have close relatives in 
the Pellonulinae. Thus the scuted Pellonulops, from Madagascar 
and South Africa (cf. Smith, 1954: 90-91, fig. 110), looks as if 
it might belong to the Pellonulinae and at the same time be closely 
related to Gikivi Stella and Sauvagella. 

Whitehead's main reason for considering the Dussumieriidae a 
primitive type of clupeoid, apart from their lack of scutes, is the 
high number of branchiostegal rays in the subfamily Dussumi- 
eriinae, and a belief that reduction in number of branchiostegal 
rays is a major trend in clupeoid evolution. The notion that the 
number of branchiostegal rays can be reduced but not increased 
in teleostean phylogeny is one that had some currency formerly 
but has lost ground in recent years. Rosen (1964) mentions 
several phyletic fines in Acanthopterygii in which the number of 
branchiostegal rays has increased. Increases — some quite 



24 BREVIORA No. 382 

striking — surely have occurred in various highly specialized lines 
of catfishcs in South America and India (personal observation), 
as well as in eels. Reduction of branchiostegal rays, of course, 
is the more usual trend, very frequently associated with reduction 
in body size, as it surely is in Congothrissa and Thrattidion. The 
high number of branchiostegal rays in Dussumeriinae is probably 
due to specialization rather than retention of a primitive character. 
The more pellonulin-like Spratelloidinae have only six or seven 
branchiostegal rays. 

LITERATURE CITED 

AuDENAERDE, Thys, V. D. 1969. Description of a new genus and species 
of clupeoid fish from Sierra Leone. Rev. Zool. Bot. Afr., 80(3-4): 
385-390. 

Blache, J. 1964. Les poissons du Bassin du Tchad et du Bassin adjacent 
du Mayo-Kebbi. Mem. Off. Rech. Sci. Tech. Outre-mer, Paris: 1-483. 

Chapman, W. M. 1948. The osteology and relationships of the round 
herring Etriiineus inicropus Temminck and Schlegel. Proc. California 
Acad. Sci., 26(2): 25-41. 

Coulter, G. W. 1962. Limnothrissa miodon and Stolothrissa tanganicae 
in Lake Tanganyika. In: Research results. Rep. Jt. Fish. Res. Org. 
N. Rhodesia 1960, 10: 7-30 (not seen). 

Daget, J. 1954. Les poissons du Niger superieur. Mem. Inst. Franc. Afr. 

Noire, No. 36: 1-391. 
, AND A. Iltis. 1965. Poissons de Cote d'lvoire (eaux douces 

et saumatres). Mem. Inst. Franc. Afr. Noire, No. 74, 1-385, 4 pis. 

GossE, J. P. 1963. Le milieu aquatique et lecologie des poissons dans 
la region de Yangambi. Ann. Mus. Roy. Afr. Centr., ser. oct., sci. 
zool., No. 116: 113-270, pis. 1-10. 

Greenwood, P. H. 1968. The osteology and relationships of the Denti- 
cipitidae, a family of clupeomorph fishes. Bull. British Mus. (Nat. 
Hist.), 16(6): 215-273. 

Matthes, H. 1964. Les poissons du lac Tumba et de la region de I'lkela. 

Ann. Mus. Roy. Afr. Centr., ser. oct., sci. zool.. No. 126: 1-204, 1 

map, 1 chart, pis. 1-6. 
Nelson, G. J. 1967. Gill arches of teleostean fishes of the family 

Clupeidae. Copeia, 1967(2): 389-398. 
. 1969. Infraorbital bones and their bearing on the phylogeny 

and geography of osteoglossomorph fishes. American Mus. Novitates, 

No. 2394: 1-37. 



1972 AFRICAN CLUPEID FISH 25 

Poll, M. 1953. Poissons non Cichlidae. Exploration hydrobiologique du 

Lac Tanganika (1946-47), Resiiltats Scientifiques, Vol. 3, part 5a. 

251 pp., 11 pis. Bruxelles, Inst. Royal Sci. Nat. Belgique. 
1964. Une famille dulcicole noiivelle de poissons africains: 

les Congothrissidae. Mem. Acad, roy Sci. Outre-mer, cl. sci. nat. 

med., n.s., 15(2): 1-40, 8 pis. 

Poll, M., P. J. P. Whitehead, and A. J. Hopson. 1965. A new genus 
and species of clupeoid fish from West Africa. Bull. Acad. roy. 
Belgique, cl. sci., ser. 5, 51(3): 277-292. 

Regan, C. T. 1917. A revision of the clupeid fishes of the genus 

Pellonula and of related genera in the rivers of Africa. Ann. Mag. 

Nat. Hist., ser. 8, 19: 198-207. 
Reynolds, J. D. 1970. Biology of the small pelagic fishes in the new 

Volta Lake in Ghana. I. The lake and the fish: feeding habits. 

Hydrobiologia, 35(3-4): 568-603. 
1971. Biology of the small pelagic fishes in the new Volta 

Lake in Ghana. II. Schooling and migrations. Hydrobiologia, 38(1): 

79-91. 

RiDEWooD, W. G. 1904. On the cranial osteology of the clupeoid fishes. 
Proc. Zool. Soc. London, 1904, 2: 448-493. 

Rosen, D. E. 1964. The relationships and taxonomic position of the 
halfbeaks, killifishes, silversides, and their relatives. Bull. American 
Mus. Nat. Hist., 127(5): 217-268. 

Smith, J. L. B. 1954. The sea fishes of Southern Africa (2nd ed.). South 
Africa, Central News Agency, Ltd. 580 pp. 

Svensson, G. S. O. 1933. Fresh water fishes from the Gambia River. 
Kungl. Svenska Vetens. Handl., ser. 3, 12(3): 1-102, 8 pis. 

Whitehead, P. J. P. 1963a. A contribution to the classification of 
clupeoid fishes. Ann. Mag. Nat. Hist., ser. 13, 5: 737-750. 

— 1963b. A revision of the Recent round herrings (Pisces: 

Dussumieriidae). Bull. British Mus. (Nat. Hist.), zool., 10(6): 
307-380. 

Wohlfahrt, T. 1937. Anatomische Untersuchungen iiber die 
Seitenkanale der Sardine (Cliipea pilchardus Walb.). Zeitschr. Morph. 
Okol. Tiere, 33(3): 381-411. 



ML'S. COM?. ZOOU 

Linr.ARY 



M^R 1 8 m 

B R E V I O R A-?. 



^..^ ^SITY 



usemni of Comparative Zoology 

CAMBRrocE, Mass. 25 February, 1972 Number 383 

TWO NEW GENERA OF BEMBIDIINE CARABID BEETLES 

FROM AUSTRALIA AND SOUTH AMERICA WITH NOTES 

ON THEIR PHYLOGENETIC AND ZOOGEOGRAPHIC 

SIGNIFICANCE (COLEOPTERA) ' 

Terry L. Erwin- 

Abstract. Two genera and two species are described as new. A key 
to related species is given and morphological striiciiires are discussed in a 
phylogenetic perspective with other bembidiine groups. An "elytral 
chaetotaxy map," which is based on a broad study of tachyine beetles, is 
included. The distribution of each species is presented by locality records. 
Illustrations of important morphological characters are given as well as 
habitus drawings of two species, each representing the new genera 
described. 

INTRODUCTION 

Darlington (1962) pointed out that the members of the 
hobani group "are certainly Tachys rather than Bembidion by 
current classification. | although | the species of this group arc 
anomalous (primitive?) in some ways and should be specially 
considered by students of bembidiine phylogeny." During the 
course of my current study of tachyine beetles, I examined sev- 
eral specimens of the hobani group and found some remarkable 
characteristics, as well as representatives of two new species. In 
the analysis of these characteristics and their distribution through- 
out the carabid beetles, another species with unusual character- 
istics was uncovered; Jeannel (1962) redescribed and illustrated 



1 The types located in the British Museum (Natural History). London, 
were examined with the support of a grant from the American Philosophical 
Society, Penrose Fund #5795. 

- Division of Coleoptera. Department oi Entomology, Smithsonian 
Institution. Washington, D. C. 20560. 



2 BREVIORA No. 383 

this species, "Plataphus" reicheelliini Csiki. and assigned it to 
a "phylogenctic series" of the Tribe Bembidiini .v. str. On the 
basis of these newly discovered and newly interpreted character- 
istics, and information from my general tachyine study, I here 
propose two new genera to contain the groups of species dis- 
cussed above, and discuss the phylogenetic implications and 
relict distribution pattern exhibited by the extant species of each 
group. I have included in this paper an "elytral chaetotaxy map,'" 
which is the result of the examination of all described and many 
undescribed groups of tachyine beetles. This was done now to 
facilitate the descriptions and discussions of these groups. 

METHODS 

The methods used here are essentially the same as those I 
employed elsewhere (Erwin, 1970) with a few exceptions. 
Measurements were made with a micrometer eyepiece in a Wild 
M5 stereoscopic microscope at a magnification of 50 diameters. 
The scale interval was 0.015 mm. Total length given here is in 
the sense of Lindroth (1961-69), that is, one overall "habitus" 
measurement of the specimen in a near normal pose, rather than 
the summation of head, pronotum, and elytron measurements 
used commonly for variation studies. 

Illustrations are of two types. Most are India Ink on bristol 
board, but those with half-tone qualities were made with an 
"F softness" pencil on bristol board. 

ACKNOWLEDGEMENTS 

I wish to thank P. J. Darlington, Jr. and H. B. Leech for the 
loan of beetles in their charge and C. H. Lindroth for critically 
reading the manuscript and offering many valuable suggestions. 

TAXONOMY 
Tasmanitachoides new genus 

(Figures 1, 5, 6, 8, 10, 11, 12, 13, 15, 16, 19, 21, 22) 

Type Species. BemhkHon liolnirli Blackburn 1901: 123. present designa- 
tion. This genus includes those species Darlington (1962) recognized 
as the hoharti group of Tachys, as well as two new species described 
below. 



!972 



SOUTH TEMPERATE CARABID-BEETLES 




Figure 1. TasmunitachoicU's hoharti (Blackburn), dorsal aspect, male, 
Queenstown. Tasmania. 



4 BREVIORA No. 383 

Diagnosis. Head with two parallel and sulcate frontal furrows 
continuous on clypeus; apical palpal article subulate; ligula 6- 
setose; anterior coxal cavity uniperf orate; anterior tibia with apex 
truncate to suboblique; scutellar stria absent externally; sutural 
stria entire, but not stricdy recurrent (see below); male aedeagus 
with symmetrical basal lobes. 

Description. Size: small beetles, 1.6-2.8 mm (upper limit 
from Darlington). Color: piceous to testaceous, appendages 
rufous to testaceous, or piceous. Microsculpture: wide, irregu- 
larly isodiametric meshes almost longitudinally arranged on 
elytra, more transverse on head and pronotum, very finely im- 
pressed. Head: broad and moderately depressed, with small but 
protruding eyes; frontal furrows sulciform and long, continuous 
on clypeus; clypeus with two setae at each anterior angle; ligula 
6-setose (Fig. 10); mentum not foveate, anterior margin acutely 
and strongly toothed; mouthparts (Figs. 5, 10, 16). Antennal 
article 2:3 ratio more than 1.0; flagellar articles longer than wide, 
filiform (Fig. 15). Eyes and head with very short and sparsely 
spaced setae. Prothorax: pronotum narrowly cordiform with 
sharp hind angles and broadly lobed base, sides barely reflexed, 
carina medial to hind angles; two pairs of lateral setae present, 
the anterior pair far forward even before anterior third, the 
posterior pair just before hind angles; margins not serrate or 
setulose; anterior transverse impression obsolete, the posterior 
impression deeply engraved, interrupted at middle, basal foveae 
deeply impressed. Prosternum with short, sparsely spaced setae. 
Coxal cavities uniperforate-separate-closed. Tibia very shallowly 
oblique apically with outer spine lateral to subapical transverse 
comb. Male with two dilated basal tarsal articles each with sparse 
setae beneath. Claws simple. Pterothorax: elytra narrow and 
elongate, more or less parallel-sided with prominent humeri; sides 
barely reflexed, without subapical plica; margins setulose-serrate 
or not; dorsal surface with deeply impressed punctulate striae, 
at least sutural and base of 5 always present, others various, but 
8 absent except at seta EoSa (Fig. 3); recurrent groove appear- 
ing double because of two elongate and foveate punctures (Fig. 
1 ) that are not continuous with sutural stria. Elytral chaetotaxy 
as in Figure 1. Mesonotum fully winged. Mesocoxae conjunct- 
confluent. Middle tibia of male with lateroapical brush. Claws 
simple. Abdomen: segments II, III, IV, and V with a single 



1972 SOUTH TEMPERATE CARABID-BEETLES 5 

pair of ambulatory setae; segment VI with one pair in male and 
two pairs in female specimens, those in female parallel to hind 
margin of segment; venter with many scattered setae as described 
for eyes and head. Genitalia: female stylus (Figs. 8, 11, 12). 
Male parameres and median lobe (Figs. 19, 21, 22), the latter 
with symmetrical basal lobes. 

Habitat. According to Darlington (1962) these beetles occur 
"in sand or gravel or under stones by rivers or brooks." 

Distribution. The known range extends from Tasmania and 
temperate southeastern Australia north to Katherine in the 
Northern Territory, but no specimens are known from the area 
between Rockhampton and Katherine, a distance of more than 
1500 miles. 

Etymology. From Tasmania, the type locality of the type 
species; from genus Tachys (another group of carabid beetles), 
to which this group was originally assigned; and from Latin, 
aides, meaning similar to. 

Key to species of Tasmanitacho':des new genus 
(modified from Darlington's 1962 key to the hobarti group) 

1. Dorsal elytral striae, except sutural, obliterated. . . liitiis Darlington 
r. Dorsal striae, at least 5, and usually some others, present on 

disc 2 

2(1'). Clypeus distinctly impressed at middle 3 

2'. Clypeus not distinctly impressed at middle 5 

3(2). Color piceous, length ca. 1.75 mm niurninibictgensis Slcane 

3'. Color: forebody darker, rufo-piceous to rufo-testaceous, than elytra, 

rufous to testaceous; length ca. 1.95 mm or longer 4 

4(3'). Small narrow beetles, ca. 1.95-2.0 mm long, 0.65 mm wide across 
elytra; forebody rufo-piceous, elytra testaceous 

arnhemensis n. sp. 

4'. Larger and broader beetles, ca. 2.5-2.8 mm long and 0.9-1.1 mm 

wide; forebody rufo-testaceous, elytra testaceous at apex; male 

genitalia (Figs. 6, 19) fitzroyi Darlington 

5(2'). Larger (2.3-2.7 mm long), almost black beetles with more or less 

well-defined elytral striae 6 

5'. Small ( 1.8 mm or less), pale beetles with weakly impressed elytral 

striae 7 

6(5). Head across eyes subequal to distance across pronotum at widest 
part; elytral striae well impressed (may include Icai Sloane) 

liobarii Blackburn 



6 BREVIORA No. 383 

6'. Head narrower than pronotum; striae, especially 5, shallower 

wattsense Blackburn 

7(5'). Elytral stria 5 well defined to at least middle of elytron; pronotum 

convex, broad anteriorly, and wider than head across eyes; male 

genitalia (Fig. 22) kingi Darlington 

7'. Elytral stria 5 hardly visible, barely impressed except just behind 
humerus; pronotum subconvex, narrower than head across 
eyes katherinei n. sp. 

Tasmanitachoides arnhemensis n. sp. 
(Figure 11 ) 

Type specimens. The holotype female and one female para- 
type, labelled "Australia, N.T., Edith Falls nr. Katherine, 110 m, 
X-25-62," and "Collectors E. S. Ross, D. Q. Cavagnaro" are 
in CAS (California Academy of Sciences, San Francisco). 

Type locality. Edith Falls on the north branch of the Daly 
River, near Katherine, Northern Territory, Australia. 

Diagnosis. Medium-sized beetles, broad, subconvex with rufo- 
piceous forebody and testaceous elytra; clypeus well impressed 
medially; striae 1, 2, and 5 well marked beyond middle. 

Description. Size: 1.95-2.0 mm total length; 0.6-0.7 mm 
wide (2 specimens measured). Microsculpture: almost perfectly 
isodiametric on elytra, moderately impressed, surface shiny. 
Head: broadly transverse, slightly wider across eyes than prono- 
tum; antennae with middle articles slightly longer than wide; 
clypeus well impressed medially; frontal furrows deeply sulcate 
from anterior margin of clypeus to just behind mid-eye level, 
roughly parallel throughout. Prothorax: subcordate, sides sinuate 
just before acute hind angles; base broadly lobed, sinuate later- 
ally inside hind angles; side margins moderately beaded, strongly 
rounded at apical third; anterior margin truncate; surface micro- 
punctulate; basal transverse impression deep and coarse laterally, 
interrupted medially by extended median groove. Pterothorax: 
wings long; elytra elongate and narrow, sides subparallel; 
humerus about square, margins subserrate and microsetulose; 
stria 1 (sutural) deep and entire from base to apex, broadly 
recurrent at apex to foveate puncture Eo8b; stria 2 moderately 
impressed from near base to Ed6c; striae 3 and 4 short, weakly 
impressed from Ed3 to Ed5a; stria 5 sulcate behind recurrent 



1972 SOUTH TEMPERATE CARABID-BEETLES 7 

humeral margin to level of Ed3, less impressed from Ed3 to 
mid-elytron; Ed7b foveate, forming inner pseudo-recurrent groove. 
Abdomen: as described under generic description. Male genitalia 
unknown. Female stylus (Fig. 11). 

Notes. This species is most closely related to the more south- 
ern T. fitzroyi Darlington, but it has members that are consider- 
ably smaller, both in length and width. 

Habitat notes. Exact habitat unrecorded, but taken near Edith 
Falls, so probably as Darlington (1962) and Sloane (1921) re- 
corded for other species, that is, near water's edge. 

Distribution. Known only from the type locality. 

Etymology. From Arnhem Land, the northernmost projection 
of land in Australia; in reference to the general area where the 
types were collected. 

Tasmanitachoides katherinei n. sp. 
(Figure 1 2) 

Type specimen. The holotype female labelled "Australia, 
N.T., Edith Falls nr. Katherine, 110m, X-25-62,'" and "Collec- 
tors E. S. Ross, D. Q. Cavagnaro" is in CAS. 

Type locality. Edith Falls on the north branch of the Daly 
River, near Katherine, Northern Territory, Australia. 

Diagnosis. Small beetles, narrow, fragile, depressed, with rufo- 
testaceous forebody and testaceous elytra; clypeus not impressed; 
striae 2-5 almost obliterated; micropunctulae and microsculpture 
well marked. 

Description. Size: 1.6 mm total length; 0.5 mm wide (type). 
Microsculpture: almost perfectly isodiametric on elytra, moder- 
ately impressed, surface shiny. Head: broadly transverse, slightly 
wider across eyes than pronotum and elytra; antennae with mid- 
dle articles about twice as long as wide; clypeus not impressed 
medially; frontal furrows shallowly sulcate, poorly defined be- 
hind level of anterior edge of eye, slightly convergent posteriorly. 
Prothorax: subcordate; sides sinuate just before slightly obtuse 
hind angles; base broadly lobed, abruptly sinuate laterally inside 
hind angles; side margins weakly beaded, moderately rounded 
at apical third; anterior margin truncate; pronotum barely convex, 
micropunctulate; basal transverse impression deep and short, 



8 BREVIORA No. 383 

laterally interrupted well inside hind angle and at middle by 
extended median groove. Pterothorax: wings long; elytra elongate 
and narrow, sides parallel; humerus almost square; side margins 
subserrate and microsetulose; stria 1 deep and entire from base 
to apex, broadly recurrent at apex to foveate puncture EoSb; 
outer striae nearly obliterated; Ed7b shallowly foveate, forming 
an inner pseudo-recurrent groove. Abdomen: as described for 
genus. Male genitalia unknown. Female stylus (Fig. 12). 

Notes. This species is most closely related to the more south- 
ern T. kingi Darlington, but it has members with much less im- 
pressed striae and a pronotum that is narrower than the head 
across the eyes. 

Habitat notes. See preceding species. 

Distribution. Known only from the type locality. 

Etymology. From Katherine, the name of the town near which 
the type was discovered. 

NOTES ON OTHER SPECIES OF TASMANITACHOIDES 

Darlington (1962) thought that Bembidion wattsense Black- 
burn might be a synonym of T. hobarti. I have examined the 
type in the British Museum and believe that B. wattsense is a 
valid species and should be included in Tasmanitachoides. I have 
not seen specimens of T. leai Sloane and prefer to follow Dar- 
lington's ideas at this time. I have seen the types of all other 
species listed in the key; those described by Darlington are in 
the Museum of Comparative Zoology, Cambridge, Massachusetts, 
and those of Sloane and Blackburn are in the British Museum 
(Natural History). 

Bembid arenas new genus 
(Figures 2, 4, 7, 9, 14, 17, 18, 20, 23) 

Type Species. Bembidion reicheellum Csiki 1929: 162, present designation. 
This species was originally described as Bembidiiini reichei Germain 
(1906: 617). Since reichei was preoccupied (Putzeys, 1846), Csiki 
(1929: 162) proposed reicheellum. Jeanne! (1962: 653) placed the 
species in Plataphiis because of the structure of the basal lobes of the 
aedeagus. However, his drawings (Jeannel 1941: 533) are not ac- 
curate, and, as Lindroth ( 1963) points out, 'The structure of the basal 
orifice of penis, used by Jea. (1941, p. 533) for characterizing his 
'genus' Plataphiis. is not different from that of related groups." 



1972 



SOUTH TEMPERATE CARABID-BEETLES 




Figure 2. Bemhidareiias reicheelliiin (Csiki), dorsal aspect, male, 
vicinity Punta Arenas, Chile. 



10 BREVIORA No. 383 

Diagnosis. Apical palpal article subulate, scutellar stria 
present, anterior coxal cavity uniperforate, anterior tibia with 
apex truncate, but with transverse subapical comb, clypeus 4- 
setose, apex of stria 5 deepened, abdomen pubescent and male 
aedeagus with symmetrical basal lobes. 

Description. Size: 3.6 to 4.2 mm. Color: piceous with 
rufescent tibiae. Microsculpture: moderately wide on elytra, 
almost regularly isodiametric meshes with slight tendency toward 
transverse arrangement throughout. Head: broad and moderately 
depressed, with moderate but prominent eyes; frontal furrows 
double, the inner pair linear, deep and wide from clypeus to just 
behind eyes at posterior edge, and with a slightly raised tubercle 
at middle of each furrow, the outer pair short and deep, enclosing 
the anterior supraorbital setae but not continued on clypeus; 
clypeus with two setae at end of each frontal furrow; ligula 
6-setose (Fig. 9); mentum not foveate, anterior margin with 
strong, truncated tooth; mouth parts (Figs. 4, 9, 17). Antennal 
articles 1, 3-5 of subequal length, 1 more robust; flagellar arti- 
cles and 2 only slightly shorter (Fig. 18). Eyes and gena be- 
neath eyes with short and scattered setae. Prothorax: pronotum 
narrowly cordiform with sharp but slightly obtuse hind angles 
and truncate base, sides slightly reflexed, with deep basal foveae; 
two pairs of lateral setae present, the anterior pair far forward, 
even before anterior third, the posterior pair just anterior to hind 
angles; margin not setulose, anterior transverse impression obso- 
lete, the basal impression deep and wide. Prostemum glabrous. 
Coxal cavities unipcrforate-separate-open. Tibia truncate apically 
but with subapical transverse comb terminated with two spine-like 
setae. Male with two dilated basal tarsal articles each with 
parallel rows of sparse modified setae beneath. Claws simple. 
Pterothorax: elytra moderately narrow, depressed and elongate, 
with slightly arcuate sides and prominent humeri; sides reflexed, 
with small subapical plica; margins not setulose; dorsal surface 
with 5 well-impressed striae, the sixth feebly impressed, seventh 
obsolete, and eighth wefl impressed behind middle only, absent 
before middle; scutellar striae well developed; apex of stria 5 
deeply impressed and connected with sutural stria (forming "re- 
current groove"). Elytral chaetotaxy as in Figure 2. Mesonotum 
fully winged. Mesocoxae conjunct-confluent. Metasternum and 



1972 SOUTH TEMPERATE CARABID-BEETLES 1 1 

metacoxae with scattered pubescence. Abdomen: segments II, 
III, IV, V with a single pair of ambulatory setae; segment VI 
with one pair in male and two pairs in female specimens, those 
of female arranged in a row transversely across segment; all 
segments with short, sparse pubescence. Genitalia: female stylus 
(Fig. 20). Male parameres and median lobe (Fig. 23), the latter 
with symmetrical basal lobes, but no '"brush sclerite" in the in- 
ternal sac as in Bcfubiclion members. 

Notes. I have also seen specimens of a second species of this 
genus from southern South America in the collection of J. Negre. 
Negre plans to describe and figure this species in a paper he is 
preparing on Berubidion of southern South America. 

The presence of short sparse setae covering the venter of all 
segments of the abdomen is also found in the subgenus Triclio- 
plataphus Netolitzky of Bembidion. Setae are also found along 
only the apical margin of each segment in members of the sub- 
genus Blepharoplataphiis Netolitzky. This character state must 
be regarded as convergent in these otherwise quite different 
groups of beetles. 

Habitat. According to Darlington {in litt.) this beetle 
"live(d) in gravel by brooks." 

Distribution. The known range of this group is southern Chile 
(Prov. Magellanes ) in the vicinity of Punta Arenas (53°40' S 
Lat.). 

Etymology. From Bembidion, another group of closely related 
beetles, and Punta Arenas, the area in which these beetles are 
found. 

DISCUSSION 

The unexpected discovery of symmetrical basal lobes on the 
male genitalia of the hobarti group and Bembidarenas led me to 
investigate this and many other characteristics, as well as their 
distribution throughout the trechine-bembidiine complex. Some 
of this investigation has been done on actual specimens, but I 
have also consulted articles by Jeannel (1926, 1932, 1936, 1941, 
1946, 1962), Lindroth (1961, 1963, 1966, 1969), Darhnston 
(1962), Ball (1960), Bell (1967), Valentine (1932), and bene- 
fited through correspondence with T. C. Barr and his unpub- 
lished "Key to Tribes of Subfamily Bembidiinae." 



12 



BREVIORA 



No. 383 



eol 
eo2 




ed 1 
ed2 



ed3 



ed4 



edS 



ed6 



ed7 



ed8 



Figure 3. Diagrammatic elytron showing all known positions of setae 
(O) in tachyine beetles. The Eo series is the elytral "umbilicate" series. 
The Ed series is the elytral disc series. The letters represent the various 
positions in which these setae are found in different groups, while the short 
connecting lines are the hypothetical directions of movement from the 
ancestral condition (A). This drawing is to be used as a map to accompany 
the descriptions. 



1972 



SOUTH TEMPERATE CARABID-BEETLES 



13 









Figures 4-12. Fig. 4. Mandibles of Benibidarenas reicheellum (Csiki), 
dorsal aspect, male, vicinity Punta Arenas, Chile. Fig. 5. Same of Tasinoni- 
tachuides hobarti (Blackburn), male, Queenstown, Tasmania. Fig. 6. Geni- 
talic ring sclerite of Tasmanitachoides fitzroyi (Darlington), dorsal aspect, 
male, Fitzroy River, North Rockhampton, Queensland, Australia. Fig. 7. 
Base of genitalic ring sclerite of Bcnihidarcnas reicheellum (Csiki), dorsal 
aspect, male, vicinity Punta Arenas, Chile. Fig. 8. Genital sclerites of 
Tasmanitachoides hobarti (Blackburn), ventral aspect, female, Queenstown, 
Tasmania. Fig. 9. Labium and left palpus of Bembidareiuis reicheellum 
(Csiki), ventral aspect, male, vicinity Punta Arenas, Chile. Fig. 10. Same 
of Tasmanitachoides hobarti (Blackburn), male, Queenstown, Tasmania. 
Fig. 11. Right stylus of female genitalia of Tasmanitachoides arnhemensis 
n. sp., ventral aspect, Edith Falls, Northern Territory, Australia. Fig. 12. 
Same of Tasmanitachoides katherinei n. sp., same locality. 



14 



BREVIORA 



No. 383 




Figures 13-18. Fig. 13. Left anterior leg of Tasinanitachoides hobaitl 
(Blackburn), lateral aspect, male, Queenstown, Tasmania. Fig. 14. Same 
of Bembidarenas reicheellum (Csiki), vicinity Punta Arenas, Chile. Fig. 15. 
Right antenna of Tasmanitachoides hobarti (Blackburn), lateral aspect, 
male, Queenstown, Tasmania; pubescence and setae not shown. Fig. 16. 
Left maxilla and palpus of Tasinanitachoides hobarti (Blackburn), dorsal 
aspect, male, Queenstown, Tasmania. Fig. 17. Same of Bembidarenas 
rcichceUum (Csiki), vicinity Punta Arenas. Chile. Fig. 18. Same of 
Bembidarenas reicheelhim (Csiki). vicinity Punta Arenas, Chile; see Fig. 15. 



1972 



SOUTH TEMPERATE CARABID-BEETLES 



15 




Figures 19-23. Fig. 19. Male genitalia and apicies of parameres of 
Tasinanitachoides fitzroyi (Darlington). Fitzroy River. North Rockhamp- 
ton. Queensland, Australia. Fig. 20. Female stylus of Bemhidarenas 
reicheelliim (Csiki), ventral aspect, vicinity Punta Arenas, Chile. Fig. 21. 
Male genitalia and apicies of parameres of Tasmaiiitaclioidcs hobarti 
(Blackburn), Queenstown, Tasmania. Fig. 22. Same of Tasinanitachoides 
kingi (Darlington). Queenstown, Tasmania. Fig. 23. Same of Bemhidarenas 
reicheelliim (Csiki), vicinity Punta Arenas, Chile. 



16 BREVIORA No. 383 

Traditionally, the trechine-bembidiine complex has been con- 
sidered as two closely related groups, each at the tribal level 
(higher levels in the Jeannelian "French School"). Trechini, 
Bembidiini, and other more loosely associated groups were con- 
sidered by Jeannel (1941) to constitute the "Styhfera," a 
"groupement naturel" of the "Caraboidea Limbata" or higher 
Carabidae. Jeannel's contention that this "groupement naturel, 
sans cependant avoir etc exactement defini" still holds today. It 
seems no authors before Ball (1960) and Lindroth (1961-69) 
worried much that their classifications were based on one or two 
major character states, as well as one or two minor character- 
istics, or less. Also these character states were investigated on 
only one or two specimens of one or two local species in what 
were in reality unknowingly diverse groups. Even so, the general 
classification of Carabidae has been in relatively good shape for 
the last 30 years and many small studies have contributed much 
to our knowledge at all levels; some of these studies were based 
on new and novel techniques and approaches. It is hoped that 
this paper will stimulate just such studies in the "Stylifera" to 
determine its inner and outer limits and its place among the 
diversity of the entire Family. 

DISCUSSION OF BEMBIDARENAS 

Members of Bembidarenas reicheellum differ from Bembidion 
at least to the extent that Phrypeiis rickseckeri Hayward does, 
and under certain analysis considerably more. Superficially, 
though, B. reicheellum is more conservative than P. rickseckeri 
and thus has not undergone close scrutiny by students of carabid 
phylogeny. The symmetrical basal lobes of the male genitalia in 
B. reicheellum do not occur anywhere in the true Bembidion, but 
are characteristic of many trechines, all patrobines, Anillina, 
apotomines, psydrines, Horologion, deltomerines, and Tasmani- 
tachoides. Nearly symmetrical lobes occur in Phrypeus and the 
subgenus Pseudolimnaeum of Bembidion, but the reduction of 
the right lobe is easily observable after dissection. The distribu- 
tion of symmetrical basal lobes in the Stylifera and other carabid 
groups indicates to me that this condition is primitive (plesio- 
morphic). If so, then reduced right basal lobes and the "basal 
bulb" have arisen several times in diverse groups, and hence 
must be used with caution in phylogenetic analyses. 



1972 SOUTH TEMPERATE CARABID-BEETLES 17 

B. reicheelliim has another major difference in the male geni- 
talia. All known Bembidion species possess a "brush-scleritc" 
in the internal sac. or remnants of one; in some groups there are 
species without the structure but it is clearly a secondary loss 
(Lindroth, personal communication). B. reicheelliim does not 
have this structure at all. The brush sclerite does not occur in 
Phrypeus either, but is found in Asaphidion. This structure and 
many others clearly illustrate the sister group relationship of 
Bembidion and Asaphidion, regardless of the one extra umbilicate 
seta (Eo9) in the latter. 

B. reicheellum has a plurisetose ligula with remnants of para- 
glossal lobes. The plurisetose ligula is found in all trechines, 
Asaphidion, and Tasmanitachoides, each with six to eight setae. 
All other bembidiines, pogonines, patrobines, etc., of the Styli- 
fera have four or less setae, usually two. The long narrow lateral 
lobes, probably remnants of the paraglossae, are characteristic of 
all trechines, but also occur in some Anillina, patrobines, and 
Horologion. Other members of the Stylifera have small rounded 
lobes or no lobes. The distribution of these character states indi- 
cates to me that the multisetiferous ligula is derived (apo- 
morphic), while the produced lateral lobes are primitive (rem- 
nants of the paraglossae). At present, any attempt to interpret 
these two character states into overall phylogenetic analysis leads 
to utter confusion; new data must be accumulated before further 
attempts are made. 

The last characteristic that is now known to be peculiar in 
B. reicheelliim is the presence of two pairs of setae on the clypeus. 
This character state also is now known in Tasmanitachoides and 
a few primitive trechines, e.g., Amblystogenium pacificiim 
Putzeys. 

On the basis of gross overall form (e.g., subulate terminal 
palpal articles, short sutural stria, elytral chaetotaxy, pubescent 
penultimate palpal articles, etc.) and the analysis of character 
states above, I think that B. reicheellum is an early off-shoot of 
the Bembidion lineage which retained some characteristics of the 
mutual "trechine-bembidiine" stock. It seems more primitive in 
several characteristics than Phrypeus and probably separated 
from the main line of evolution before Phrypeus. The habitat of 
gravel by brooks is considered by Darlington {in litt.) to "be a 
very old. stable habitat, and that the running water acts as a 



18 BREvioRA No. 383 

buffer against climatic changes. I think insects in this habitat 
may persist for very long periods." 

DISCUSSION OF TASMANITACHOIDES 

In addition to some of the points discussed above, Tasmani- 
tachoides members have other peculiar characteristics not gener- 
ally found in the Bembidiini. Deeply sulcate frontal furrows, ex- 
tending to about mid-eye level, are elsewhere found only in 
Phrypeus. The small group of lateral papillae of the lacinia are 
unique, but some Anillina and Lymnastis have a single spine 
near this same location. Whether the structures are truly homol- 
ogous is doubtful, however, because of their orientation. 

On the basis of gross overall form (e.g., subulate terminal 
palpal articles, complete sutural stria, elytral chaetotaxy, uniper- 
forate anterior coxal cavity, pubescent penultimate palpal arti- 
cles, etc.) and the analysis of character states above, I think the 
hobarti group is best placed as an early off-shoot of the tachyine 
lineage which gave rise to the Anillina. All true tachyines (and 
Lymnastis-Micratopus, which I regard as true tachyines on the 
basis of many characteristics) have biperforate anterior coxal 
cavities, and are the only Stylifera that do (aside from Apotomus 
if this group really belongs to the Stylifera). The distribution of 
biperforate coxal cavities indicates that this character state is 
derived (apomorphic), but convergent in distantly related groups, 
perhaps in strengthening the prothorax under greater demands 
for digging in partial subterranean life (but see also Erwin, 1970: 
168).^'' 

The characteristics of members of the hobarti group show 
similarities to the trechines, but as in B. reicheelliim 1 think 
these characteristics indicate an old lineage surviving in an old 
but stable habitat, and maintaining certain characteristics of an 
early "trechine-bembidiine" stock. 

The relationships within the Stylifera and particularly within 
the trechine-bembidiine complex will be thoroughly discussed 
along with the supporting evidence in my current revision of the 
Tachyina (Erwin, MS) and need not be dealt with at length 
here. 



1972 SOUTH TEMPERATE CARABID-BEETLES 19 

LITERATURE CITED 

Ball. G. E. 1960. Carabidae (Latreille. 1810). //; R. H. Arnett. Jr., 
The Beetles of the United States. The Washington Catholic Uni- 
versity of America Press, pp. 55^181. 

Bell. R. T. 1967. Coxa! cavities and the classification of the Adephaga 
(Coleoptera). Ann. Ent. Soc. America. 60(1): 101-107. 

Blackburn, T. 1901. | Australian Bembidiini.] Trans. R. Soc. South 
Australia. 25: 120-124. 

CsiKi. E. 1929. Coleopterorum Catalogus, pars 104, Carabidae; 
Harpalinae. pp. 347-527 

Darlington, P. J.. Jr. 1962. Australian Carabid beetles XI. Some 
Tachys. Psyche. 69( 3 ) : 1 1 7- i 28. 

Erwin, T. L. 1970. A reclassification of Bombardier Beetles and a 
taxonomic revision of the North and Middle American species (Cara- 
bidae: Brachinida). Quaestiones Ent., 6: 4-215. 

. (MS). A revision of the New World species of Tachyina 

and a reclassification of the generic components of the world. 

Germain, P. 1906. Apuntes entomolojicos. Anal. Univ. Chile, 117: 
589-653. 

Jeannel, R. G. 1926. Monographic des Trechinae. L"Abeille, 32: 
224-550. 

1932. Revision des genre Lininastis (Coleoptera, Cara- 
bidae). Soc. ent. Fr., Livre der Centenaire, pp. 167-187. 

1936. Les Bembidiides Endogcs (Col. Carabidae). Rev. 



Fr. Ent.. 3: 241-399. 

1941. Faune de France 39, Coleopteres Carabiques. 



Paris, 1. 571 pp. 
1946. Coleopteres Carabiques de la Region Malgache, 



part I, Faune de lEmpire Fran^ais. Paris, 6. Pp. 1-372. 
1962. Biologic de TAmerique Australe. Vol. 1. Etudes sur 



la Faune du Sol. Paris, VII. Pp. 611-655. 
Lindroth, C. H. 1961-69. The Ground-beetles (excluding Cicindelidae) 
of Canada and Alaska, parts I-VI. Opusc. ent. Suppls. XX, XXIV, 
XXIX, XXXIII, XXXIV, XXXV. 

PuTZEYs. J. 1846. Prcmices entomologiques: Note monographique sur 
le genre Pasimachus: Nouv. Cicindel et Carabid. Mem. Soc. Sc. Liege, 
2: 353-417. 

Valentine, J. M. 1932. Horologion, a new genus of cave beetles 
(Fam. Carabidae). Ann. ent. Soc. America. 23(1): 1-11. 



MUS. CGMP. ZOOL 
LICRARY 



* ' » 



■• 198; 



B R E V I O R. A 

UM!VFRS1TY 

Meseiuin of Comparative Zoology 

Cambridge, Mass. 25 February. 1972 Number 384 



AN ATTEMPT TO DETERMINE THE SYSTEMATIC 

POSITION OF ELLOPOSTOMA MEGALOMYCTER, 

AN ENIGMATIC FRESHWATER FISH FROM BORNEO 

Tyson R. Roberts^ 



Abstract. The types and only known specimens of EUopostoma 
megalomycter (Vaillant) are figured and redescribed. This highly distinc- 
tive fish may have a Weberian apparatus with osseous swimbladder 
capsules similar to those in Cobitidae, but it differs from all Cobitidae in 
several important respects. Resemblance between EUopostoma and 
Kneriidae evidently is mainly superficial. EUopostoma cannot be classified 
with assurance because fundamental osteological information about it is 
lacking, owing to the present condition of the specimens. Its relationships, 
when worked out on the basis of fresh material, are likely to be of phyletic 
significance. The present account will permit the identification of new 
material of EUopostoma with greater facility than the older accounts, 
because some errors are now corrected and more descriptive details are 
given. 

INTRODUCTION 

The rich fish collections obtained by the Netherlands Borneo 
Expedition of 1893-97 included specimens of a pecuhar little fish 
that Vaillant (1902: 141-149, figs. 42-45) described as Aperiop- 
tus megalomycter and assigned to the Cobitidae. He also proposed 
a provisional new genus, EUopostoma, for it. Vaillant's figures 
depict a moderately elongate, small-scaled fish with very large 
nostrils and eyes, and a most peculiar inferior mouth with a single 
pair of barbels. No modern ichthyologist has found it possible to 
assign a firm systematic position to this strange fish. Despite 



1 Museum of Comparative Zoology, Cambridge, Mass. 02138. 



2 BREVIORA No. 384 

Vaillant's placement with the cobitids, later accounts of that 
family have neglected the species. Its appearance calls to mind 
the Kneriidae, a family of gonorynchiform fishes known only from 
fresh water in Africa. Weber and de Beaufort (1916: 237-239, 
fig. 97) examined the type specimens and doubted that they be- 
long to the Cobitidae. They reproduced Vaillant's figures and 
pointed out that the species is not an Aperioptus and should be 
known as Ellopostoma megalomycter (Vaillant) (see below). No 
additional specimens have been found and nothing further about 
it has appeared in print. 

Dr. M. Boeseman, Curator of Fishes of the Rijksmuseum van 
Natuurlijke Historic in Leiden, kindly lent his institution's type 
specimens of Ellopostoma so that I might try to classify it. The 
following account gives an emended description of Ellopostoma 
and points out the difficulties in trying to place it on the basis of 
the available material. There remains a major question as to 
whether the structures interpreted as a Weberian apparatus by 
Vaillant are really such. If they are, Ellopostoma' s closest rela- 
tionship probably is with Cobitidae. If its postcranial bony 
capsules have some other origin the relationships are more prob- 
lematic and might lie with Kneriidae. In the description given 
below its characteristics are compared with those of both Cobiti- 
dae and Kneriidae. 

I wish to thank Mr. Andrew Konnerth for preparing radio- 
graphs and Professor George S. Myers for reading the manu- 
script, and Dr. M. L. Bauchot for information about the Paris 
specimen of Ellopostoma. 

ELLOPOSTOMA Vaillant 1902 

Type species. Aperioptus megalomycter Vaillant 1902, by 
monotypy. 

Note. In describing Aperioptus megalomycter, Vaillant pro- 
posed for it the new generic name Ellopostoma in a footnote in 
case it should prove not congeneric with Aperioptus. Aperioptus 
Richardson 1848 (type species Aperioptus pictorius Richardson 
1848, by monotypy) was based on a rather unsatisfactory draw- 
ing and notes made before the only two specimens were inad- 
vertently thrown away, and its identity may never be resolved. 
The only locality information for the specimens is that they came 



1971 BORNEAN FRESHWATER FISH 3 

from Borneo. If they were freshwater they may have been 
cobitids, as implied by Giinther (1868: 371). In any event, 
Richardson's description and figure (reproduced on pp. 238-239 
in Weber and de Beaufort, 1916) indicate a fish differing in 
several major respects from Vaillant's fish, the proper designation 
for which is thus Ellopostoma megalomycter, as pointed out by 
Weber and de Beaufort. 

Ellopostoma megalomycter (Vaillant) 
Figure 1 

Material. The type series of Ellopostoma consists of four 
specimens. No holotype was designated and thus they are equiv- 
alent syntypes. Three specimens, 24.3, 39.5 and 41.4 mm in 
standard length, were retained by the Rijksmuseum in Leiden 
(RMNH 7777) and one specimen, 40.5 mm in standard length, 
was presented to the Museum National d'Histoire Naturelle in 
Paris (MNHN 03-202). I have examined all four and find them 
very soft and poorly preserved. The 24.3-mm specimen belongs 
to the genus Noemacheilus. It has the characteristic three pairs 
of barbels and 1 1 dorsal fin rays. Its eye is relatively smaller 
than that of Ellopostoma, but unusually large for a Noemacheilus. 
The Paris specimen has the mouth, snout, orbit on one side and 
possibly some gill arches badly damaged. The 41.4-mm speci- 
men, badly damaged to begin with, was dissected by Vaillant; 
the posterior gill arches are missing and structures in the anterior 
region of the vertebral column are missing or badly damaged. 
The 39.5-mm specimen (RMNH 7777) is intact and is hereby 
designated lectotype. The following description is based pri- 
marily on the lectotype and 41.4-mm paralectotype. Vaillant 
reported that the 41.4-mm specimen contained eggs which seemed 
ripe or nearly so. It is now completely eviscerated. 

The specimens arrived in the Netherlands along with other 
unlabelled material collected by Dr. J. Biittikofer in the vicinity 
of Sintang, near the middle portion of the Kapuas River (Vaillant, 
1902: 149). 

Coloration. The specimens are now discolored and their 
original color pattern has largely disappeared. Vaillant (p. 147, 
fig. 42) gave a rather good color description, to which I can add 
but little: "La coloration est identique sur les trois grands ex- 



BREVIORA 



No. 384 



emplaires [i. e., in all but the 24.3-mm specimen, which is a 
Noemacheilus] et doit se rapporter sans doute a une livree fonda- 
mentale, plus ou moins masquee peut-etre sur le vivant, mais qui 
se retabUt lorsque les animaux sont plonges dans la liqueur. 
Dans I'etat actuel la teinte generale est roux tres pale, devenant 
blanchatre argente en descendant vers le ventre, tout a fait ar- 
gentee a la region operculaire, ainsi que sur les cotes et le dessous 
de la tcte. Une serie de taches plus ou moins en quadrilateres, 
occupant de 3 a 5 rangees d'ecailles, ornent le dos et les flancs. 
Sur le premier on en compte 7, les deux anterieures, les plus 
petites, sont I'une a la region nuchale, I'autre a mi-distance de 
la dorsale, les trois suivantes, respectivement sous I'origine de 
cette dorsale, en son milieu, enfin juste en arriere d'elle; les deux 
dernieres Tune a distance egale de cette tache metepipterique 
et de la septieme tache, celle-ci placee un peu en avant de I'inser- 
tion de la caudale. Sur la ligne laterale se voient 7 ou 8 taches, 
les anterieures peu distinctes; la derniere est au milieu du pedon- 
cule caudal a I'insertion mcme de Furopterc, les deux precedentes 
repondcnt aux intervalles clairs des trois dernieres taches dorsales, 
disposecs par rapport a celles-ci en damier; la quatrieme en 
procedant toujours d'arriere en avant est sous la partie anterieure 
de la tache dorsale metepipterique, les 3 ou 4 taches anterieures 




Figure 1. Ellopostoma megalomycter, 39.5-mm lectotype, RMNH 
7777 (camera lucida). First elongate ray of anal and one or two rays of 
dorsal fin that are broken off restored on basis of 41.4-mm paralectotype. 
Arrow indicates position of vent. Base of pectoral fin should be slightly 



more anterior than figured. 



1971 BORNE AN FRESHWATER FISH 5 

moins developpees, moins distinctes, sont plutot placees au-des- 
sous de la ligne laterale, la plus avancce scrait juste en avant de 
I'origine de la dorsale; dans I'espace compris cntre ces taches 
lateroanterieures et les taches dorsalcs se voicnt 4 macules, nu- 
ageuses, repondant aux intervalles des cinq taches dorsales 
anterieures. Enfin il ne faut pas negliger d'attirer Tattention sur 
une petite tache de forme demi-circulaire, d'un noire bleuatre 
accentue, qui pourrait bien ctre cerclee de blanc, c'est-a-dire 
ocellee, laquelle orne la caudale dans sa partie infericure au point 
d'insertion avec la pedoncule. L'iris parait sombre; un cercle 
pupillaire argente." The silvery white coloration on the opercle 
and lower side of the head is still evident, as is the small black 
spot on the ventral half of the caudal peduncle. A similar spot 
is present on the caudal peduncle in many Cobitidae and in some 
Kneriidae. 

Proportional measurements. Proportional measurements ex- 
pressed as times in standard length, those of 39.5-mm lectotype 
followed in parentheses by those of 41.4-mm paralectotype: 
head 4.4 (4.8); eye 14.9 (13.8); snout 16.5 (17.6); bony inter- 
orbital 27.3 (26.7); depth 6.5 (5.9); depth of caudal peduncle 
11.5 (10.6); snout-tip to dorsal origin 2.5 (2.45); snout-tip to 
pelvic origin 2.08 (1.91); snout-tip to vent 1.80(1.72); snout- 
tip to anal origin 1.23 (1.20); length of dorsal base 3.9 (3.75); 
length of anal base 12.4(14.3); length of caudal peduncle 8.0 
(8.6); length of pectoral fin 5.1 (4.9); length of longest dorsal 
ray 4.5 (broken); length of anal fin 6.7 (6.2). 

Fins (Fig. 1). Dorsal iii 16, last ray divided to base. Anal 
iii 5 or iii 6, last branched ray divided to base. Pectoral i 1 1 or 
i 12. Pelvic i 7 (pelvic splint present). Caudal with nine prin- 
cipal rays in both lobes, about seven procurrent rays in upper 
lobe and four in lower lobe. 

Dorsal fin origin slightly posterior to a vertical midway between 
insertions of pectoral and pelvic fins. Pelvic insertion shghtly 
anterior to a vertical through base of seventh segmented ray of 
dorsal fin. Anal fin set far back, its origin somewhat anterior to 
a vertical midway between base of last dorsal fin ray and end of 
hypural fan. Origin of dorsal on a vertical through ninth vertebra. 
Pelvic insertion on a vertical through 13th or 14th vertebra. 
Anal origin on a vertical through 25th vertebra. 



6 BREVIORA No. 384 

Pectoral fin somewhat larger than pelvic fin, none of its rays 
hypertrophied or otherwise modified, its longest ray reaching al- 
most to insertion of pelvic fin. Height of dorsal fin almost twice 
that of anal fin. Dorsal fin base about three and one half times 
longer than anal fin base. Dorsal fin margin slightly falcate, the 
first three branched rays sHghtly longer than the rest. Anal fin 
margin more or less straight. Caudal deeply forked (damaged 
in both specimens). Membranes between rays in all fins uni- 
formly thin. 

Anal fin position variable in Kneriidae, several of them having 
the anal fin as far posterior as in EUopostoma. In Kneriidae 
dorsal fin short-based, never with more than 10 rays. Fin counts 
and placements highly variable in Cobitidae. Relatively few 
cobitids have the anal fin as far posterior as EUopostoma. Some 
(e.g., Noemacheilus pavonaceus) have fin placements and number 
of fin rays near those of EUopostoma. Number of principal caudal 
rays variable in both Cobitidae (from 17 to 19) and Kneriidae 
(16 in Grasseichthys, 18-20 in Parakneria). 

Position of vent (Fig. 1). Vent located between pelvic fins, 
somewhat closer to insertion of outermost pelvic ray than to tip 
of longest pelvic ray. In cobitids the vent position varies from 
midway between pelvic insertion and anal origin (in some Noe- 
macheilinae) to immediately anterior to anal fin origin (in some 
Cobitinae and Botiinae). In at least some (all?) kneriids vent 
slightly or immediately anterior to anal fin origin. 

Squamation (Fig. 1). Head scaleless. Body entirely scaled 
except for a small area in front of pectoral base. Isthmus scaled 
up to attachment of gill membranes. Fin bases scaleless. Ap- 
proximately 75 scales in a lateral series from upper angle of gill 
opening to base of caudal fin, 20 scales between supraoccipital 
spine and dorsal fin origin, 20 scales between dorsal fin origin 
and pelvic fin insertion (11 rows above and eight rows below 
lateral line), and 17 scales around caudal peduncle. 

Scales cycloid, height about 1.25 times width. A scale from 
side of body above pectoral fin has 12 radii and 22 circuU on its 
anterior field and 14 radii and 16 circuli on its posterior field. 
Focus well within anterior half of scale. 

Vaillant regarded the scales of EUopostoma as belonging to the 
type (multiradiate) characteristic of Cobitidae, and quite distinct 
from the type (pauciradiate) characteristic of Cyprinidae. 



1971 



BORNEAN FRESHWATER FISH 



Mouth (Figs. 1-4). Mouth inferior. No teeth on jaws or 
roof of mouth. Upper jaw apparently bordered exclusively by 
preniaxillaries, which have broad-based ascending processes and 
are well separated from each other at midline. Approximate 
shape of preniaxillaries and of portion of dentaries bordering 
mouth as in Figure 4. Preoral and oral barbels absent. A median 
fleshy thickening or mentum on lower lip behind symphysis of 
mandibles (Figs. 2, 3). The name Ellopostoina was given be- 
cause the mouth resembles that of a sturgeon (Vaillant, 1902: 
145). 




Figure 2. Close-up of head with mouth almost fully open, 39.5-mm 
lectotype (camera lucida). Arrows indicate extent of gill opening. 



Vaillant recorded a pair of barbels, one at each side of the 
mouth, at the junction of the upper and lower jaws, which he 
supposed were located at the ends of the maxillary bones (I am 
unable to ascertain the presence of maxillaries, although they 
may be present). He was able to view the barbels only with 
difficulty. There is a fold or flap of skin in the lip at or near the 
rictus of the jaws, but this can hardly be described as a barbel. 
Certainly there are no structures corresponding to the elongate 
pair of barbels depicted in Vaillant's figure 43 in either the lecto- 
type or the 40.5 and 41.4-mm paralectotypes. As noted above, 
the 24.3-mm paralectotype is actually a specimen of Noemachei- 
lus, and has three pairs of barbels. 



8 



BREVIORA 



No. 384 




Figure 3. Ventral view of head, 39.5-mm leciotype (camera lucida). 



Snout (Figs. 1-3, 5). Snout obliquely truncate and very rigid. 
Mesethmoid immovably articulated to frontals. Nasal organs oc- 
cupying a large cavity, its diameter about one-third that of eye. 
Nostrils separated by a membraneous strip with a small, pos- 
teriorly directed flap (vestigial nasal barbel?). Dorsal to nostril 
a slender nasal bone, ventral to nostril a small, broad-based tri- 
angular shaped bone (antorbital?) with its apex directly ventrally, 
and posterior to nostril a supraorbital bone and the lateral margin 
of the lateral ethmoid bone. Lateral ethmoid forming part of 
anterior rim of orbit, not extending ventrally to orbit, of general- 
ized morphology. The snout is unlike that in any other teleost 
I have seen. 




Figure 4. Ventral view of jawbones bordering mouth, 39.5-mm lecto- 
type (freehand). 



197 



BORNEAN FRESHWATER FISH 



Eye (Figs. 1, 2, 5). The eye is misshapen in all three speci- 
mens of EUopostoma. Figures 1, 2 and 5 depict its approximate 
shape. It presumably is oblong in live specimens, not perfectly 
round as in Vaillant's figures. Orbital rim free. Adipose eyelid 
absent. The combination of large eye and exceptionally large 
nostrils found in EUopostoma is unusual. The eye is relatively 
much smaller in Cobitidae and Kneriidae (in both of which 
orbital rim usually attached) with the exception of the minute 
kneriid Grasseichthys, which has relatively small olfactory organs. 

Laterosensory canals (Figs. 1-3, 5). Laterosensory system 
with supraorbital, infraorbital, preopercular, temporal, occipital 
and lateral canals. Supraorbital canal enclosed in a superficial 
bony tube on frontal bone until anterior margin of eye, then 
continued unenclosed for a short distance before it is again en- 
closed by the tubular nasal bone, without ethmoidal commissure 
or medially-directed branches. Infraorbital canal enclosed in a 
superficial, segmented bony tube (the infraorbital series?), con- 
tinued anteriorly onto snout, where it turns upwards abruptly 
and terminates immediately anterior to nostrils. The appearance 
of the infraorbital canal is similar to that in Cobitidae (and 
Kneriidae?). Preopercular canal with two or three short lateral 
branches terminating in a small pore (Fig. 3). Occipital canal 



supraorbital canal 
supraorbital 



frontal extrascapular 



nasal 




lateral ethmoid 



infraorbital canal 

temporal canal 



occipital canal 
temporal commissure 
supraoccipital 



posterior frontal fontanel 
parietal 



lateral canal 
supracleithrum 



Figure 5. Dorsal view cf head, based mainly on 39.5-mm lectotype 
(camera lucida). 



10 BREVIORA No. 384 

with a temporal commissure, the lateral portions enclosed in a 
superficial bony tube, the median portion unenclosed, traversing 
the posterior frontal fontanel distinctly anterior to supraoccipital 
bone (Fig. 5). The temporal commissure is usually (always?) 
present in Cobitidae and Kneriidae but usually it lies flush with 
the anterior margin of the supraoccipital bone, sometimes entirely 
enclosed in a bony tube well within the supraoccipital bone (cf. 
Ramaswami, 1953; Greenwood et ol., 1966). Lateral canal of 
Ellopostoma extending along middle of body to base of caudal fin, 
as in Cobitidae and Kneriidae. 

Rooj of cranium (Fig. 5). Cranial roof exposed, with only 
a thin cover of skin, unlike most Cobitidae in which dorsum of 
head is usually rather fleshy or at least covered with moderately 
thick skin. No anterior frontal fontanel. Posterior frontal fontanel 
slightly longer than eye, bordered by frontals, parietals, and 
supraoccipital, very similar to posterior frontal fontanel in Cobit- 
idae (cf. Ramaswami, 1953). In adults of the larger kneriids 
(cf. Greenwood et al., 1966; Giltay, 1934) the cranial roof is 
complete. Cromeiia (cf. Swinnerton, 1903) and Grasseichthys 
have a frontal fontanel extending to the snout, quite unlike that 
in either Cobitidae or Ellopostoma. 

Posterior region of basicranium (Fig. 6). Parasphenoid ex- 
tending posteriorly to anterior margin of supraoccipital, with 
which it is firmly sutured. Parasphenoid with a median sutural 
process. In Cobitidae and Kneriidae (cf. Ramaswami, 1953; 
Greenwood et al., 1966, fig. 6 on p. 376) posterior portion of 

parasphenoid 
basioccipital 




1 J 



lateral process 
basioccipital centrum 

transverse process? 
firEt vertebral centrum 



Figure 6. Ventral view of part of basicranium and first vertebral 
centrum, 41.4-mm paralectotype (freehand). 



1971 BORNEAN FRESHWATER FISH 11 

parasphenoid divided, the two divisions extending distinctly pos- 
terior to anterior margin of basioccipital; parasphenoid without 
median sutural process. Basioccipital centrum of Ellopostoma 
with anteriorly directed lateral processes, otherwise devoid of 
processes that might be identical with pharyngopophyses of 
Cobitidae. Pharyngopophyses of Cobitidae frequently poorly de- 
veloped or absent; if present, they are posteriorly directed. 
Pharyngopophyses absent in Kneriidae. 

Gill cover (Figs. 1, 2). Subopercle relatively large. Shape 
of opercle and subopercle generalized compared to cobitids in 
which they are often highly modified. Gill opening wide; gill 
membranes broadly united to isthmus at a point below middle of 
subopercle (Fig. 2). Branchiostegal rays apparently three, as 
Vaillant reported. Cobitidae invariably have three branchiostegal 
rays as do other cyprinoids. The number of branchiostegal rays 
is also reduced in Kneriidae, Kneria and Cwmeria with three 
(Giltay, 1934; Swinnerton, 1903), Grasseichthys with only two 
(Gery, 1965). The extent of the gill opening, variable in Cobit- 
idae, is always very narrow in Kneriidae. 

Gill arches (Fig. 7) The following notes on the gill arches 
are based on the damaged 41.4-mm paralectotype in which only 
the first three arches of the left side and the first arch of the 
right side are present. Suspensory pharyngeals large. First arch 
of right side with about ten gill rakers on leading edge (none 
on suspensory pharyngeal) and 15 gill rakers on trailing edge 

1 mm 
I » 

Figure 7. Dorsal view of left siispen^ory pharyngeal 1 with gill rakers, 
41.4-mm paralectotype (camera lucida). 



(including a few on suspensory pharyngeal). First arch of left 
side with about 12 rakers on leading edge (none on suspensory 
pharyngeal) and 18 on trailing edge (of which five are on sus- 
pensory pharyngeal). The rakers on the first arch are all on the 



12 BREVIORA No. 384 

uppermost third of the arch. Perhaps the rakers from the lower- 
most two-thirds have been stripped away. Second and third gill 
arches with a full complement of gill rakers, at least 25 on both 
leading and trailing edges. Teeth absent on first three gill arches. 

Vaillant did not mention the gill arches or rakers; he was 
unsuccessful in an attempt to remove the pharyngeals intact from 
the 41.4-mm specimen, and they are now missing. 

Pectoral skeleton. Extrascapular bone present, bearing junc- 
tion of temporal, occipital, and lateral branches of laterosensory 
canal system on its dorsal surface (Fig. 5). Pectoral arch of 
left side of 41.4-mm specimen detached from cranium and with 
supracleithrum exposed; posttemporal missing (normally absent?), 
postcleithrum evidently absent. 

In Cobitidae the posttemporal is reduced, sometimes absent, 
and there are no postcleithra (Rendahl, 1930, 1933a; Bacescu- 
Mester, 1970). The pectoral skeleton of the more typical 
kneriids, Kneria and Parakneria, is undescribed. Swinnerton 
(1903: 67, fig. N) described the pectoral skeleton of Cromeria, 
which has a posttemporal with a very elongate upper hmb articu- 
lated directly to the supraoccipital (quite unlike the posttemporal 
of Cobitidae) and a slender postcleithrum. 

Anterior region of vertebral column, postcranial bony capsules, 
swimbladder. Radiographs of the lectotype and 41.4-mm para- 
lectotypc and examination of the latter reveal a pair of large, 
rounded, bony capsules, one on either side of the vertebral column 
almost immediately posterior to the cranium. The capsules of 
the 41.4-mm specimen are broken (their ventral portions missing) 
and detached from the vertebral column. Their transverse diam- 
eter is about three-fourths of the eye diameter. Portions of a 
tough membrane, presumably that of the swimbladder, remain 
inside the capsules. Vaillant interpreted these structures as 
homologues of the swimbladder capsules in Cobitidae. As the 
anterior portion of the vertebral column itself is also damaged 
and detached, and some parts possibly missing, it is impossible 
to tell precisely what structures contribute to form the capsules. 
The appearance of the capsules in radiographs of various 
Cobitidae is at least superficially similar to that of the capsules 
of Ellopostoma. The first centrum of Ellopostoma is free and 
relatively short, and bears a small lateral projection on one side 



1971 BORNEAN FRESHWATER FISH 13 

(a transverse process?) which is absent (broken off?) on the 
other side (Fig. 6). 

Either Ellopostoma has a Weberian apparatus with swimbladder 
capsules like Cobitidae or else it has complicated bony structures 
superficially similar and perhaps comparable in function to the 
swimbladder capsules of Cobitidae. I am inclined toward the 
first possibility. I am unable to make a detailed comparison of 
the capsules of Cobitidae with Ellopostoma on the basis of the 
present material. Nothing comparable to the capsules of Cobitidae 
or Ellopostoma has been reported in Kneriidae. 

Vaillant (1902) was unable to find a swimbladder in 
Ellopostoma and stated (p. 142): "Pour la vessie natatoire je 
n'ai pu en trouver trace, pas plus que d'enveloppe osseuse 
pouvant la contenir, I'etat de I'exemplaire [41.4-mm] paraissait 
cependant devoir permettre d'en constater I'existence." The de- 
velopment of the swim bladder is variable in Cobitidae. Fre- 
quently it is reduced to only the left and right halves of the 
anterior portion enclosed in the bony capsules, but some forms 
also have a nonenclosed posterior portion, the genus Noema- 
cheiliis with several species in which it is particularly large 
(Rendahl, 1933b). The swimbladders of Kneriidae have not 
been described in detail; Rosen and Greenwood (1970: 11) 
indicated that Kneriidae possess a swimbladder divided by a 
distinct constriction into a small anterior and a much larger 
posterior chamber, as in Chanos and in ostariophysans generalized 
with respect to their swimbladders. 

Vertebral counts. Radiographs reveal about 32 vertebrae (ex- 
cluding hypural centrum) in the lectotype of Ellopostoma and 
about 31 in the 41.4-mm paralectotype. 

The following vertebral counts have been recorded for 
Kneriidae: Grasseichthys 36-37 (Gery, 1965); Parakneria 41-42 
(Poll, 1965); and Cromeria 42 and 45 (Swinnerton, 1903). 
Vertebral counts of Cobitidae range very widely. 

Caudal skeleton (Fig. 8). Figure 8 is based on a radiograph 
of the 41.4-mm specimen in which some elements are relatively 
well defined but others are indistinct, so that it is somewhat 
interpretive and should be used with caution. Hypurals appar- 
ently five. Proximal ends of hypurals 1-4 indistinct; impossible 
to determine from radiographs whether hypural 2 separate or 
fused with ural centrum. Parhypural evidently united to ural 



14 



BREVIORA 



No. 384 



epural 



uroneural ? 
urostyle 



Pj.Uj(.U2? 




parhypurapophysis ? 



parhypural 



Figure 8. Caudal skeleton, 41.4-mm paralectotype (camera lucida of 
radiograph). 



centrum by a saddlelike joint, and bearing a well-developed par- 
hypurapophysis. A single epural. Urostyle slender. What ap- 
pears to be a slender separate ossification (uroneural?) dorsal 
to urostyle may actually be part of it. Bases of neural and hemal 
spines of vertebrae 1 and 2 indistinct. 

The caudal skeleton of Ellopostoma, so far as can be deter- 
mined, agrees in almost all respects with the caudal skeletons of 
both Cobitidae (undetermined cobitid; Botia macracanthus) and 
Kneriidae (Cromeria, Grasseichthys) described and figured by 
Monod (1968). The cobitids and kneriids lack uroneurals; it is 
possible that one is present in Ellopostoma. The cobitids have 
hypural 2 fused to the ural centrum, whereas it is evidently free 
in the kneriids; the condition of hypural 2 in Ellopostoma is 
undetermined. All agree in having a single epural and five 
hypurals, except Botia macracanthus, which has six hypurals. 
Rosen and Greenwood (1970: 13, fig. 10) found six hypurals 
and two epurals in Kneria wittei and in Parakneria sp. In K. 
wittei the first epural is very reduced, but in Parakneria sp. both 
epurals are large. 



1971 BORNEAN FRESHWATER FISH 15 

DISCUSSION 

Ellopostoma differs from all Cobitidac and Kneriidae in the 
relatively large size of its nostrils and eyes and in the morphology 
of its snout, mouth, and (so far as known) parasphenoid bone. 
It also differs from all Cobitidae in lacking barbels, and from all 
Kneriidae in its possession of postcranial bony capsules, a long- 
based dorsal fin with relatively numerous rays, and an extensive 
gill opening, and in the morphology of its posterior frontal 
fontanel. 

Ellopostoma cannot be classified with reasonable assurance 
because important information about the structure of its snout, 
mouth, pharyngeal bones, and postcranial bony capsules is lack- 
ing. If the postcranial bony capsules are part of a Weberian 
apparatus (as seems likely), then it is probably closer to Cobitidae 
than to any other living group. It does not show the specialized 
traits (mesethmoid movably articulated to f rentals, lateral ethmoid 
highly modified as an erectile suborbital spine) of the subfamilies 
Botiinae and Cobitinae, and differs greatly from them in general 
appearance. It shares many characters with the Noemacheilinae 
and is somewhat like them in general appearance, but the resem- 
blances may be due to convergence and retention of generalized 
characters. Cobitidae, Kneriidae, and Ellopostoma are similar 
in many respects, and this is of particular interest in view of 
evidence that gonorynchiforms and ostariophysans are closely 
related (Greenwood et ai, 1966; Rosen and Greenwood, 1970). 
Whatever the relationships of Ellopostoma, it is likely to be of 
major phyletic significance. 

LITERATURE CITED 

Bacescu-M ESTER, L. 1970. The morphological comparative study of the 
shouldei girdle of the fishes belonging to the Cobitidae family from 
Rumania. Trav. Mus. Hist. Nat. Grigore Antipa (Bucharest), 10: 

251-272. 

Gery, J. 1965. Poissons du Bassin de I'lvindo. Biologia Gabonica, 

1(4): 375-393. 

GiLTAY, L. 1934. Contribution a I'etude du genre Xenopomaticbthys 
(Kneriidae). Bull. Mus. roy. Hist. nat. Belgique, 10(44): 1-22. 

Greenwood, P. H., D. E. Rosen, S. H. Weitzman, and G. S. Myers. 
1966. Phyletic studies of teleostean fishes, with a provisional classifi- 
cation of living forms. Bull. Amer. Mus. Nat. Hist., 131(4): 339- 
456, pis. 21-23. 



16 BREVIORA 384 

GiJNTHER, A. 1868. Catalogue of the Fishes in the British Museum. 

Vol. 7: XX + 512 pp. 
MoNOD, T. 1968. Le complexe urophore des poissons teleosteens. 

Mem. Inst. Fond. Afrique Noire (Dakar), No. 81. 705 pp. 
Poll, M. 1965. Contribution a 1 etude des Kneriidae et description d'un 

nouveau genre, le genre Parakneria (Pisces, Kneriidae). Mem. Acad. 

roy. Belgique, I. Sci, oct., 36(4): 1-28, 13 pis. 
Ramaswami, L. S. 1953. Skeleton of cyprinoid fishes in relation to 

phylogenetic studies. 5. The skull and gasbladder capsule of the 

Cobitidae. Proc. Nat'l. Inst. Sci. India, 19(3): 323-347. 
Rendahl, H. 1930. Einige Bemerkungen iiber den Schultergiirtel und 

die Brustflossenmuskulatur einiger Cobitiden. Ark. Zool., 21(16): 

1-31. 
1933a. Studien iiber innerasiatische Fische. Ark. Zool., 

25(11): 1-51. 
1933b. Weitere Untersuchungen iiber den Schultergiirtel 



und die Brustflossenmuskulatur der Cobitiden. Ark. Zool., 25(10): 

1-38. 
Rosen, D. E., and P. H. Greenwood. 1970. Origin of the Weberian 

apparatus and the relationships of the ostariophysan and gonorynchi- 

form fishes. Amer. Mus. Novitates, No. 2428: 1-25. 
SwiNNERTON, H. H. 1903. The osteology of Croineria nilotica and 

Galaxias attenuatns. Zool. Jahrb. (Jena), anat. abth., 18: 58-70. 
Vaillant, L. 1902. Resultats zoologiques de Texpedition scientifique 

neerlandaise au Borneo Central. Poissons. Notes Leyden Mus., 

24(1): 1-166, 2 pis. 
Weber, M., and L. F. de Beaufort. 1916. The fishes of the Indo- 

Australian Archipelago (Leiden, E. J. Brill), Vol. 3: XV + 455 pp. 



LIHRARY 

MAR 1 8 1985 

lARVARD 

B R E V I O R^A 



MMsemoi of Coioiparative Zoology 

Cambridge, Mass. 25 February, 1972 Number 385 

THE CHANARES (ARGENTINA) TRIASSIC REPTILE 

FAUNA. XII. THE POSTCRANIAL SKELETON 

OF THE THECODONT CHANARESUCHUS. 

Alfred Sherwood Romer 



Abstract. A description is given, with a reconstruction, of the post- 
cranial skeleton of the long-snouted thecodont Clutnaresiichus bonapcirtei 
from the early Middle Triassic Chaiiares Formation of Argentina. The 
general proportions are comparable to those of crocodilians, and life habits 
may have been similar. However, there is no positive indication of rela- 
tionship to either later Crocodilia or Phytosauria, and Chanaresiichiis and 
its relatives ( Proterochampsidae) appear to be a sterile offshoot of the 
primitive proterosuchian stock. Dermal armor is little developed; the 
pelvis, although somewhat advanced in build, shows no trend toward the 
crocodilian elimination of the pubis from the acetabulum; the pes is of 
a specialized nature, with emphasis on the inner digits. 

INTRODUCTION 

In a recent paper in this series (Romer, 1971), I described the 
cranial anatomy of two long-snouted thecodonts, Chanaresuchus 
bonapcirtei and Gualosuchus reigi, from the (? Anisian) Triassic 
Chaiiares Formation of Argentina. Few postcranial remains of 
Gualosuchus are available to me. For Chanaresuchus, however, 
a number of specimens from our 1964-65 expedition are at hand, 
and Sr. Bonaparte allowed me to study several further specimens 
in the Instituto Lillo of Tucuman collections. Below, I give a 
description, with restoration, of the postcranial skeleton of 
Chanaresuchus. 

Materials. The following specimens from the La Plata- 
Harvard expedition include postcranial remains: 

The holotype: (La Plata Museum 1964-X1-14-12) Skull, 
partial column, a few limb bones. 



2 BREVIORA No. 385 

MCZ 4035. Part of the contents of a large nodule, including 
a column articulated to caudal 3, and considerable girdle and limb 
material. 

MCZ 4036. The remainder of material in this nodule, in- 
cluding incomplete and in part poorly preserved remains of skulls 
and postcranial materials of two further individuals. 

MCZ 4037. A skull and presacral column. 

MCZ 4038. A slab containing a melange of materials; most, 
however, appear to belong to a nearly completely disarticulated 
Chanaresuchus skeleton. 

There is considerable variation in the size of the individuals 
concerned. Largest is MCZ 4037, in which skull length to the 
quadrate is 260 mm. In the type skull this measurement is 211 
mm. My figures of postcranial material (including restoration) 
arc based mainly on MCZ 4035, in which postcranial vertebrae 
and limb bone measurements indicate a size approximately that 
of the type. The two skulls somewhat imperfectly preserved in 
MCZ 4036 measure about 230 and 220 mm. The imperfect skull 
with which the array of postcranial material on MCZ 4038 appears 
to be associated had a length of approximately 165 mm. 

Several specimens in the Instituto Lillo collections were studied. 
Best was a specimen with articulated column extending to the 
proximal part of the tail, and much of the appendicular skeleton; 
the skull measured, to quadrate, 175 mm, and the individual was 
thus considerably smaller than the type. A second specimen in- 
cluded much of the postcranial skeleton of a moderately large 
individual; the skull is missing, but comparison of limb bone 
measurements indicate a size shghtly larger than MCZ 4035. A 
third Tucuman specimen is an immature individual, in which post- 
cranial measurements are about half those of the type; preserved 
is a hind foot exactly duplicating the peculiar structure, and a 
nearly complete tail. 

Vertebral column. The atlas-axis complex, best seen in MCZ 
4037 (Fig. 1, a, Z?), is of a type readily derivable from that of 
primitive reptiles. The atlas neural arches are paired. Dorsally a 
posterior extension was applied to the lateral surface of the axis 
arch; an anterior extension is flattened ventrally, presumably for 
articulation with the exoccipital (no pro-atlas has been found). 
A slight rugosity seen on one specimen indicates a rib attachment. 
Ventrally the stem of the arch extends stoutly downward to 



1972 



CHANARESUCHUS 




Figure 1. Vertebrae and rib of Chanaresuchus bonapartei. Composite, 
mainly from MCZ 4036. a, lateral view of vertebrae 1-3; b, anterior view 
of atlas-axis; c, d, lateral and anterior views of an anterior dorsal vertebra; 
e, lateral view of a posterior dorsal; /, an anterior dorsal rib, in postero- 
ventral view, x 1. 



articulate with the atlas intercentrum; internally it is excavated to 
form the dorsolateral segment of the socket for reception of the 
skull condyle. The lower part of this socket is formed by the 
atlas intercentrum, a stoutly developed crescent that articulates 
broadly on either side with the neural arches and is bevelled 



4 BREVIORA No, 385 

anteriorly for condyle articulation. Behind the atlas intercentrum 
lies the axis intercentrum, a thin crescent connecting ventrally the 
atlas intercentrum and the axis centrum. The atlas centrum is 
a stout element, which forms the deeper part of the condylar 
socket (slightly indented centrally by a notochordal pit); dorsally 
it forms the anterior end of the floor of the neural canal; posteriorly 
it is firmly apposed to the axis centrum. It is partially obscured 
in lateral view by the atlas arch and intercentrum and ventrally by 
the axis intercentrum. 

The axis neural arch is relatively low but long. As noted, the 
articulation of the atlas arch is simply a flattened area on the 
lateral surface of the arch, anteriorly; posteriorly, normal post- 
zygapophyses are present. A slight lateroventrally directed 
process, rugose at its tip, is an incipient transverse process for the 
tubercular articulation of the rib. The parapophysial articulation 
of the rib capitulum is indicated by a pronounced rugosity far 
down the side of the centrum and close to its anterior border. 
In contrast to the intercentrum anterior to it, the axis centrum 
(like the cervical vertebrae that follow it) is keeled ventrally. 

Presacral column (Fig. 1, c-e). MCZ 4035 includes a pre- 
sacral column and sacrum, showing definitely the presence of 23 
presacral vertebrae; MCZ 4037 includes, in two segments, the 23 
presacral vertebrae, the column having presumably broken off at 
the sacrum. Most of the description that foflows is drawn from 
these two specimens; unfortunately, in much of both columns the 
transverse processes are broken and incomplete. 

The column is, for the most part, of a typical thecodont 
character. In MCZ 4037 the length of typical dorsal segments, 
measured along the centra, averages 16 mm; in MCZ 4035, a 
somewhat smaller individual, this measurement averages 14 mm. 
The neural spines are centered somewhat to the back of the mid- 
length of the centrum; they are broadened anteroposteriorly, 
expanding in width above the region of the zygapophyses. The 
anterior spines are relatively tall and narrow; posteriorly, in the 
lumbar region, they are lower and broader distally. They are 
narrow in transverse diameter, without noticeable expansion at 
the top. When the surface is well preserved, faint longitudinal 
striations are frequently seen. In the larger specimens there 
appears to have been a late continuation of growth (or rather 
ossification) of the spines, the neural spines in larger forms being 



1972 CHANARESUCHUS 5 

taller in relation to other dimensions of the vertebrae. For 
example, in the large specimen, MCZ 4037, the total height of a 
posterior cervical (including both arch and centrum) is approxi- 
mately three times the length of the centrum; in the smaller MCZ 
4035, the height is barely above two and one-third times the 
central length. 

The posterior zygapophyses are situated directly below the 
posterior margins of the neural spines; the prezygapophyses, on 
the contrary, lie well in advance of the anterior margins of the 
neural spines, on arch processes running upward and forward 
from the anterior margins of the transverse processes. The 
articular surfaces of the posterior zygapophyses face somewhat 
outward as well as downward; the anterior zygapophysial surfaces, 
in contrast, face inward as well as upward. Beginning with the 
axis, transverse processes are developed, their bases centering at 
a point well down on the side of the arch and toward the front. 
They are directed slightly posteriorly and ventrally rather than 
directly laterally. The axis process projects but 2 mm or so 
from the surface of the arch; posteriorly, the lengths increase so 
that, for example, by vertebra 7 in 4037, the length is close to 
10 mm and there is obviously further increase more posteriorly, 
although, to my regret, the processes of the posterior dorsal and 
lumbar regions are mostly broken and imperfect. In MCZ 4035 
the transverse processes as far back as vertebra 7 terminate in a 
narrow tip; posteriorly the transverse processes are broadened 
anteroposteriorly for their entire length, with oval distal facets 
for tubercular attachment. 

The anterior margin of the arch curves almost directly down- 
ward from the process supporting the anterior zygapophysis to 
the anterior margin of the centrum; posteriorly the lateral margin 
of the arch is strongly concave in outline, allowing a large gap 
between arches here for passage of nerves and vessels. 

The centra are smoothly oval in end contours, with a height 
considerably greater than the width. The centra are amphicoelous, 
with thickened rims at either end. Between the two ends of the 
centrum, the element pinches in somewhat laterally, and in side 
view the lower margin is somewhat concave in outline. The 
anterior centra are distinctly keeled ventrally; at about the be- 
ginning of the dorsal region the keel fades out, and posteriorly 
the ventral surface is smoothly rounded. . 



6 BREVIORA No. 385 

In the axis the parapophysis is merely a rugose area far down 
the anterior margin of the centrum. Posteriorly the parapophysis 
gradually moves upward along the anterior margin of the centrum 
and slowly comes to project distinctly from the surface of the bone 
so that, for example, by presacral 10 it has risen nearly to the 
top of the front margin of the centrum and projects outward for 
several millimeters. Posteriorly beyond this point the para- 
pophysis continues movement upward and backward toward the 
transverse process and a thin ridge develops connecting the two 
(as in some primitive reptiles). Regrettably the processes are but 
imperfectly preserved in the "lumbar" region of materials available 
to me, but it appears that in the last four or so presacrals the two 
processes are united. 

Sacral vertebrae. The two sacral vertebrae arc present in MCZ 
4035, and are present also in a Tucuman specimen. They differ 
mainly from the vertebrae anterior to them in the broad base from 
both centrum and arch from which the apophyses supporting the 
sacral ribs arise. 

Caudal vertebrae. In the material available to me at Harvard, 
there is little identifiable caudal material; even in specimen MCZ 
4035 articulated vertebrae cease at the third caudal. In the 
Instituto Lillo material, however, one specimen shows twelve 
caudal vertebrae in articulation with the sacrals and eleven are 
present in a second specimen. There is little decrease in vertebral 
length in these series of proximal caudals. Immediately behind 
the sacrum, however, the neural spines begin to show a backward 
slant and show a decrease in breadth distally; in the end members 
of these series, the neural spine is reduced to a small elevation 
lying above the postzygapophyses. Transverse processes are well 
developed; broad at their bases, they extend laterally and sHghtly 
ventrally to become pointed at their tips. In a Tucuman specimen 
of rather small size (skull length 175 mm), the lengths of the 
transverse processes of the anterior caudals, measured from the 
mid-line, are, in sequence, 29, 30, 28+, 33, 29, 25 and 20 mm. 
In this specimen the first chevron preserved lies between the fifth 
and sixth centra. That this is truly the first of the series is 
suggested by the fact that there is little space between the more 
anterior centra for insertion of a haemal arch, whereas more 
posteriorly a sufficient ventral gap is present. That the chevrons 
extended far down the tail is indicated by the fact that in the small 



1972 CHANARESUCHUS 7 

specimen mentioned below, they are seen in the region of the 
20-23rd caudals. 

In an Instituto Lillo specimen considerably smaller than "adult" 
specimens of Clumaresuclnis but definitely representing a young 
individual of this form, 35 articulated caudals arc present, plus a 
few disarticulated elements beyond. In its present condition, 
little detail can be made out, but the last elements appear to be 
subterminal in nature, and a length of 40 or so caudals seems 
to be indicated. 

Ribs. Regrettably, there is no available specimen with a good 
set of articulated ribs. In two instances a single rod-shaped 
structure that appears to be a single-headed atlantal rib, is present 
at the anterior end of a column. Back of the atlas, the ribs are 
distinctly double headed in typical archosaur fashion. In the 
anterior part of the presacral column, the two heads are widely 
separated; in correlation with the rapid development of trans- 
verse processes as we progress backward, the capitular branch is 
much longer than the tubercular, curving downward and inward 
from the line of the shaft. A short series of articulated vertebrae 
in MCZ 4038, representing an individual somewhat smaller than 
the "adults," bears a sequence of cervical ribs, presumably be- 
ginning with the atlas; the lengths are: 6, 12, 26, 32, 52, 60, ?, 
75 ± and 85 mm. The more anterior, at least, of this series have 
pointed tips, indicating that they lay anterior to the point of rib 
connection with the (presumably cartilaginous) sternum. There 
is no indication in these cervical elements of an anterior "spur" 
from a point near the head of the rib that is found in crocodilians, 
and is highly developed in certain other thecodonts. 

Of ribs in the dorsal region, remains are sparse. A number of 
rib heads show some variation in the amount of separation of 
tubercle and capitulum, those in which the two are less widely 
separated presumably coming from the posterior part of the region, 
in which transverse processes and parapophyses are approaching 
one another. In MCZ 4037 a complete rib (Fig. 1,/) is present 
close behind the eighth vertebra and may belong to that segment. 
Regrettably, no ribs are preserved in association with vertebrae 
of the "lumbar" region, in which shortened ribs were presumably 
present. For most of its extent the rib is slender; proximally, 
however, there is a slight expansion in the form of a thin flange 
posterior to the main proximal end of the shaft, which runs to the 
capitulum. 



8 



BREVIORA 



No. 385 




1972 CHANARESUCHUS 



Figure 2. Chanarcsuchus bonapartei, girdle and limb bones. Composite, 
but mainly from MCZ 4035. a. ventral view of pelvic girdle; b, dorsal 
view of pelvic girdle and sacral ribs; c, pelvic girdle from right side; d, left 
humerus, dorsal surface; e, right scapulocoracoid (dorsal end of scapula 
at left); /, right hind foot, calcaneum incompletely preserved; g, dorsal 
and lateral views of left femur; /;, extensor surface of left tibia; /, extensor 
surface of left fibula. X Vz. 



10 BREVIORA No. 385 

Two sacral ribs are present in MCZ 4035, although some details 
are obscured. They extend outward from a broad base, including 
much of the lateral surface of the centrum as well as the lower 
part of the neural arch. They expand somewhat distally and are 
blunt-ended; as indicated by several specimens of ilia, they were 
not too firmly attached to the articular areas for them on the 
internal iliac surface. 

Girdles. Several examples of scapulae and coracoids are 
present in available material; that illustrated (Fig. 2, e) pertains 
to MCZ 4035. The two elements of the endochondral girdle are 
not tightly sutured to one another, and were found separated in two 
instances. The scapula is tall and slender, somewhat broadened 
at its upper margin, which is a "finished" one in the specimens 
studied. At its lower end the scapula is somewhat expanded 
anteriorly and its margin curved outward, presumably for clavicu- 
lar attachment. The posteroventral portion of the bone is 
thickened and bears externally the well-defined upper margin of 
the glenoid cavity, which faces posterolaterally. Scapula and 
coracoid meet in a long horizontal suture. The latter bone is an 
oval plate, thickened posterodorsally to bear the lower part of 
the glenoid, and having the usual coracoid foramen. The re- 
mainder of the bone is relatively thin and convex in contour 
externally, the lower portion obviously turning medially across the 
chest. There is no trend toward posteroventral expansion, such as 
would be expected in a possible crocodile relative. Presumably 
clavicles and interclavicles were present, but I have not been able 
to identify such elements in the material available. 

A number of specimens include pelvic girdle material, so that 
the nature of the ilium, ischium, and the anterodorsal portion of 
the pubis can be readily recognized. In no case, however, is there 
present a complete, well-preserved pelvis with the elements in 
proper relations with one another. Closest to this condition is 
that associated with MCZ 4035, and my figured restoration 
(Fig. 2, a-c) is based on this specimen. Here the proper position 
of the ilia can be determined by their relation to the preserved 
sacral vertebrae and ribs; the ischia are for the most part well 
preserved; the pubes, however, are imperfect. 

The ilium is of the primitive type seen in many thecodonts — 
essentially a simple vertical blade, extending posteriorly in tapering 
fashion some distance beyond the region of the acetabulum, the 



1972 CHANARESUCHUS 11 

front margin curving downward not far in advance of the aceta- 
bulum. On the inner surface are well-marked depressions for 
attachment of the two sacral ribs. On its outer surface the ilium 
forms the upper half of the acetabular cavity. This depression is 
most deeply incised posteriorly, where it reaches its most dorsal 
position, to be bordered dorsally by a sharp outwardly extending 
ridge on the ilium; thence the upper border, less sharply defined, 
slants forward and downward a considerable distance before 
reaching the pubis. 

The pubis forms the anteroventral margin of the acetabulum; 
this cavity is here not as sharply incised into the bone as is true 
of its iliac and ischiadic margins, and the pubis is not greatly 
thickened here. The anterior surface of the ilium adjacent to the 
acetabular border is somewhat thickened; this surface continues 
downward to form a stout rounded ridge along the anterior margin 
of the pubis, the ridge turning ventrally in its distal portion. 
Medial to this ridge the pubis extends as a thin sheet to the mid- 
line of the body, where it meets its fellow in a long symphysis. 
Although the condition of the specimens makes the matter some- 
what uncertain, it would seem that the anterior portion of this 
broad sheet of bone turns somewhat ventrally, while the posterior 
part of the symphysial region lies in a horizontal plane. 

At the posterior end of this ventral expanse of bone, the pubic 
symphysis terminates and the posterior margins of the two pubes 
retreat upward and outward toward the acetabular region. The 
exact nature of the posterior margin of the bone and the distance 
it retreats dorsally before making contact with the ischium is 
uncertain. The usual obturator foramen penetrates the pubis not 
far below the acetabular border. 

The ischium forms the posteroventral sector of the acetabular 
border. The dorsal portion of the bone is here greatly thickened, 
forming a horizontal shelf with a crescentic outer margin, to 
constitute a prominent segment of the acetabular cavity. Below 
this area the ischium rapidly decreases in thickness, but increases 
in anteroposterior breadth ventrally. The posterior margin of the 
bone, slanting medially, descends nearly straight downward for 
some distance, then curves posteriorly and, finally, ventrally to 
meet its fellow in symphysis. This symphysial union extends 
forward in a nearly horizontal plane to a point somewhat posterior 



12 BREVIORA No. 385 

to the level of the puboischiadic area of contact below the aceta- 
bulum. Here, the symphysis terminating, the anterior margin of 
the ischium ascends nearly straight upward and laterally toward 
the acetabular region. We thus have, in Chanaresuchus, a definite 
advance in pubic structure over the proterosuchian condition in 
the presence of a broad ventral "incision" in the primitive pubo- 
ischiadic plate and a marked break in the primitively continuous 
ventral symphysis. 

Pectoral limb. Little material of the pectoral limb is available. 
A humerus, 85 mm in length, was found with MCZ 4035 (Fig. 
2, d), but is not too well preserved. It is expanded at both ends 
in primitive reptilian fashion, but shows little of the primitive 
"twist" of one end on the other. The posterior margin of the 
"head" segment curves medially to the proximomedial corner of 
the bone, whence the proximal articular surface, facing dorso- 
medially, curves laterally around the head of the bone. At the 
proximal end of the lateral margin a typical deltopectoral crest is 
present (in crushed condition). The distal end of the bone is 
distinctly divided, both dorsally and ventrally, into medial and 
lateral condyles, the medial the stouter, with distal articular 
surfaces for radius and ulna. Neither of the foramina often 
present distally in the humerus of early reptiles is present. 

I cannot with confidence identify either radius or ulna in any of 
the adult Chanaresuchus specimens studied. They are present, 
however, in association with a humerus in the small specimen in 
Tucuman, mentioned above. As preserved in this obviously im- 
mature individual, they are merely elongate bony cylinders, with 
some expansion proximally of the element which I take to be the 
ulna, but without indication of an olecranon. These elements, 
measuring 32 and 33 mm in length, are associated with a humerus 
measuring 36 mm in length. If the same proportions were to hold 
in MCZ 4035, the radius and ulna would be expected to measure 
74 and 76 mm in length respectively. 

Pelvic limb. The series of three major elements of the hind 
legs — femur, tibia, and fibula — are preserved in articulation in 
three instances, and a number of isolated hind leg elements are 
present on MCZ 4038. In the three articulated specimens the 
lengths of these elements as preserved are 128 mm, 94 mm, and 
95 mm in MCZ 4035; 15 1± mm, 129 mm, 91+ mm in the leg 



1972 



CHANARESUCHUS 



13 




^^nr^^^^'^^^ 




LQ-j 



yi^\ 



m 



"41?: 



Plate 1. Skeletal restoration of Chanaresiichiis bonapartei. Composite, but skull from the holotype, post-cranial skeleton mainly from MCZ 4035. Manus and 
dermal shoulder elements unknown; ribs imperfectly preserved and those of "lumbar" region unknown. About V3 size of type and MCZ 4035. 



14 BREVIORA No. 385 

of MCZ 4036 with articulated foot, 150± mm, 135 mm and 117+ 
mm for another specimen in the same block. As noted elsewhere, 
the skeleton of MCZ 4035 is apparently somewhat immature; the 
limbs in MCZ 4036 appear to be those of essentially mature 
individuals. In all three articulated examples the femur is 
definitely longer than the tibia; there is, however, a very consider- 
able diflerence between the examples, in MCZ 4036 where the 
femoral length is 1 1 1 percent and 117 percent of the tibia, and 
MCZ 4035, where the figure is 135 percent. 

The femur (Fig. 2, g) is of a pattern seen in many thecodonts 
and preserved little changed in crocodilians. The bone is relatively 
long and slender, the ends little expanded. In side view, the shape 
is sigmoid, with the proximal end curved upward and somewhat 
medially, the distal end curved somewhat downward; the curvature, 
however, is less extreme than in more advanced thecodonts. The 
proximal end of the bone is much modified, so that the morpho- 
logically medial (or anterior) margin is turned upward, the lateral 
border downward. The curved head of the bone is thickened, the 
thickening increasing toward the medial (upper) border; preserva- 
tion in available material, however, is not good enough to clearly 
outline the articular area. About two-fifths the distance down, 
the bone is extended ventrally, with, medially, a large oval area 
that probably (as in crocodilians) lodged the insertion of the long 
caudifemoral muscle; the projecting lower margin is a trochanter 
that presumably afforded insertion to the caudifemoralis brevis. 
The bone broadens somewhat distally, with a partial division into 
condyles, a shallow intercondylar fossa above, and a ventral 
popliteal depression. The distal end of the bone is unossified in 
MCZ 4035, so that details of areas of articulation cannot be made 
out. 

Tibia and fibula, again, are of typical primitive archosaur 
construction (Fig. 2, h, i). The broad head of the tibia is essen- 
tially triangular in section; there is little development of a cnemial 
crest, which is represented merely by the top of a ridge that 
descends much of the length of the bone, separating a lateral- 
facing surface of the shaft from one facing anteromediaUy. 
Terminally, there is a broad oval surface, somewhat convex, for 
articulation with the astragalus. The fibula is, as expected, a 
relatively slender element, somewhat expanded proximally for 
appos tion to the femur; distally (in contrast with the situation in 



1972 CHANARESUCHUS 15 

many reptile groups) there is only a minor expansion in diameter 
of the bone to accommodate the oval terminal articulation with 
the calcaneum (and to a much lesser degree with the astragalus). 
Tarsal elements are present in MCZ 4035 and 4036. In 4035 
the astragalus is well preserved (Fig. 2,/). Dorsally it bears a 
large articular area for the tibia, gently concave, elongate medio- 
laterally and facing somewhat anteriorly as well as dorsally, and 
dorsolateral to this a smaller facet for the fibula. No calcaneum 
was preserved with this specimen. MCZ 4036, on the other hand, 
lacks the astragalus, but has an incompletely preserved calcaneum. 
This shows a posterolateral tuber of "crocodilian" type, but the 
main body of the bone is imperfect. In both specimens mentioned 
distal tarsals 3 and 4 are present as flattened ovals lying over the 
heads of the lateral metapodials. There is no trace of more medial 
distal elements, although they may have been present in cartilage. 

The structure of the pes is the most distinctive feature of 
Chanaresiichiis (Fig. 2,/). Much of the right foot is present in 
MCZ 4035, but digit IV is represented only by a metatarsal splint; 
a right foot complete except for the ungual of toe IV is present in 
MCZ 4036, but the proximal ends of the inner toes are obscured 
by a refractory matrix. Two closely comparable specimens of the 
Chanaresuchus foot are present in the Tucuman material. 

In archosaurs generally the trend in foot construction has been 
for a retention of the primitive phalangeal formula (except for 
frequent reduction of the fifth toe), but the development of a 
symmetrical pattern, with toe III the longest, toes II and IV some- 
what shorter but subequal in length, and toe I short. This is true 
of the Chanaresuchus foot; but whereas in most archosaur toes 
II-IV are subequal in development, here there is very strong 
emphasis on the inner toes, and digit IV, on the contrary, is very 
slender. Digit I is somewhat shortened, but very stoutly built. 
Digit II is massive. Digit III is relatively slim. Digit IV is 
slender, almost sphntlike in structure. Digit V is represented only 
by a metatarsal spur. 

In some other archosaurs, such as certain crocodilians, 
Ticinosuchus, Euparkeria and Stagonolepis, there is a modest trend 
toward strength in the more medial digits, but never to an extent 
approaching the condition seen here. A similar trend (but usually 
a much less extreme one) towards a strengthening of the inner 
toes is found in mosasaurs and a number of chelonians. These 



16 BREVIORA No. 385 

are aquatic forms, and this specialization is apparently an "im- 
provement" in paddle-action — a fact that gives strength to the 
assumption that Chanaresuchiis was in great measure a water 
dweller. It has been assumed that Chanaresuchiis and its relatives 
in the Proterochampsidae might be antecedent to the Crocodilia, 
but such strong emphasis on the inner toes is not to be expected 
in an ancestor of the group. As regards possible phytosaur 
relationship, foot material of that group is rare, and the pes, when 
restored, is generally based on poor material. I am indebted to 
Dr. Chatterjee, of the Geological Study Group of the Indian 
Statistical Institute, for a figure of a well-preserved foot of a 
phytosaur from the Maleri Formation. This shows almost no 
trace of emphasis on the inner toes, and relationship of the 
Proterochampsidae to the phytosaurs seems highly improbable. 

Dermal armor. No gastralia are present in articulated fashion 
in any specimen. In contrast to a large fraction of the thecodont 
assemblage, other body armor appears to have been feebly 
developed, and consisted merely of a single row of thin scales 
lying over the neural spines. As preserved in several specimens, 
scales are definitely present from the axis to the last presacral; 
there is no evidence as to whether they extended onto the tail 
region, although this was probably the case. The scales are 
wedge-shaped as seen from above, narrow anteriorly, broader 
posteriorly, with a low longitudinal median keel. The anterior 
end of each scale underlies that anterior to it. There are approxi- 
mately three scales the length of each vertebra, although there is 
no apparent relationship between vertebral segmentation and scale 
arrangement; in one specimen 57 scutes were present above the 
first 20 vertebrae. Their appearance and arrangement is very 
similar to that pictured in Ticinosuchiis by Krebs (1965, fig. 8). 
This author restores the dorsal scales in the presacral region in a 
double row, in analogy, I suspect, with the pattern seen in various 
other thecodonts. I doubt if this was the case. The scales pre- 
served in the presacral region number only about enough to make 
up a single median row. The presence of scales above the tail 
vertebrae in Ticinosiichus suggests that, were appropriate materials 
to be discovered, they would be found to continue along the tail 
in Chanaresuchiis as well. 

Restoration. In the accompanying plate I have attempted a 
skeletal restoration of Chanaresuchus at one third the natural size 



1972 CHANARESUCHUS 17 

of a typical adult. Much of the skeleton is taken from MCZ 4035. 
The dermal shoulder elements, the manus. and the posterior ribs 
are restored, and the tail is imperfectly known. Despite these 
lacunae and the fact that in ribs and certain other elements the 
restoration is composite, 1 believe that the restoration gives a 
fairly accurate picture of the animal's skeleton. 

In restoring an early archosaur, an important question is whether 
the pose was that of a biped or a quadruped. In all primitive 
reptiles the front legs are somewhat shorter and less massive than 
the hind. In most archosaurs this disproportion in limb lengths 
is usually much greater and led to a general (but not universal) 
belief (which I shared) that archosaurs ab initio were more or 
less bipedal in locomotor tendencies, and that most quadrupedal 
archosaurs, such as the sauropods and various ornithischians, had 
relapsed from a bipedal mode of progression back to progressing 
on all fours. Charig, Attridge and Crompton (1965) have, as a 
result of a study of probable sauropod history, come to the con- 
clusion that quadrupedal pose was primitive for archosaurs, and 
that the development of powerful hind legs and a highly developed 
tail were, to begin with, adaptations for an amphibious life, and 
that bipedalism arose later. 

It is not unreasonable to believe that, as these authors argue, 
the sauropods developed without passing through a bipedal stage. 
I think, however, that on present evidence, the primitive orni- 
thischians were at least partially bipedal, although a majority of 
the ornithischian subgroups later became quadrupedal. 

The possibility that the ancestral archosaurs were amphibious 
leads to speculation as to the general nature of early reptilian 
history. I have argued (although not without strenuou; oppo- 
sition ) that the ancestral reptiles, although having acquired an 
amniote style of development, were still amphibious, and perhaps 
even mainly aquatic in habits. Of the great group of synapsids, 
which were dominant in the Permian and earlier Triassic, most 
became fully terrestrial, but the most primitive (and oldest) 
synapsids, the ophiacodont pelycosaurs of the Pennsylvanian and 
early Permian, were still essentially aquatic fish-eaters. Ancestral 
Permian archosaurs are almost unknown, but they may have 
followed a similar life pattern, but for a longer period of time. 
Further, we have no undisputed knowledge of the early ancestors 



18 BREVIORA No. 385 

of the euryapsid sauropterygians and placodonts or of the ichthyo- 
saurs; at their earHest appearance these reptiles were aquatic — 
and perhaps primitively so. Mesosaurus, of the Permo-Carbo- 
niferous boundary, shows that if the primitive reptiles had be:ome 
terrestrial, reversion to the water took place at a very early stage 
of reptilian history. And while I do not want to even suggest any 
close relationship between archosaurs and Mesosaurus (whose 
phylogenetic relationships are quite problematical), the similar 
adaptations in this genus and in the archosaurs in the powerful tail 
and highly developed hind limbs are strikingly suggestive. 

But to return from this discussion to the question of limb dis- 
parity and body pose in archosaurs. Correlation between limb 
proportions and presumed posture is none too simple a matter. 
Such obviously amphibious and quadrupedal forms as crocodilians 
and phytosaurs have front legs nearly as well developed as hind; 
in a "sample" crocodilian, for example, the humerus plus radius 
are about 84 percent the length of femur plus tibia, and in a 
phytosaur described by McGregor (1906), the figure is 87 percent. 
But forms that seem quite surely quadrupedal may show a con- 
siderably greater disparity in length between front and hind legs. 
The heavily armored aetosaurs, for example, are universally con- 
sidered to be quadrupeds, but in Aetosaurus and Stagonolepis, 
according to figures given by Walker (1961), the front legs 
are but 61 percent and 64 percent the length of the hind. 
Protewsuchus [Chasmatoscmnis] is an early and surely primitive 
thecodont that is reasonably regarded as amphibious in habits 
although the available materials suggest that the front legs were 
but about 66 percent the length of the hind. 

On the other hand, Euparkeria, Hesperosuchus, and Ornitho- 
suchus are regarded by those who have studied them (Ewer 1965, 
Colbert 1952, and Walker 1964) as bipedal; and yet their hmb 
ratios are as high or higher than some of the presumed quadrupeds, 
being 70 percent, 67 percent, and 70 to 75 percent, respectively. 
Obviously limb proportions in themselves are not sufficient to 
enable one to give a firm conclusion as to pose; body structure 
in general must be taken into consideration. 

As regards Chanaresuchus, the Umbs are very markedly dis- 
proportionate. If the radial length in MCZ 4037 was that esti- 
mated in relation to the humerus, the front leg would be but 53 
percent that of the hind. However, the humerus of this specimen 



1972 CHANARESUCHUS 19 

may be incomplete distally. If, instead, we take the small specimen 
from Tucuman in which all four bones conceived are present (with 
lengths of 39 mm for humerus, and 32, 59, and 59 for radius, 
femur, and tibia), we get a somewhat higher figure of 61 percent. 
These figures taken by themselves would seem to strongly suggest 
bipedality. But in other regards there is considerable reason to 
believe that Chanaresuchus was amphibious, spending much of its 
time in the water, where the body would be in a horizontal 
position; and while the animal could quite probably assume a 
semi-erect bipedal posture, this would only occur at such times 
when, ashore, speed seemed imperative. In consequence, I have 
restored the animal as a quadruped. 

Relationships. In the preceding paper of this series, I have 
commented on possible relationships of Chanaresuchus and its 
relatives in the Proterochampsidae. Reig (1959) and Sill (1967) 
have argued for Proterochampsa as a pre-crocodile; Walker 
(1968), on the other hand, maintains that this genus is a pre- 
phytosaur rather than a pre-crocodilian, but maintains that its 
"cousin" Cerritosaurus is pre-crocodilian. Study of skull structure 
led me to believe that there was little in the cranial anatomy of 
Chanaresuchus to suggest relationships to either Crocodiiia or 
Phytosauria, and that this group, the Proterochampsidae, was 
essentially a sterile one, representing a modest advance over the 
Proterosuchia, which in middle Triassic days occupied a position 
in the ecology similar to that of the phytosaurs of the late Triassic 
and the crocodilians of the later Mesozoic. Study of the postcranial 
anatomy does not lead me to change my previous conclusions. 

The postcranial skeleton is for the most part of a primitive and 
generalized archosaurian type; the only advances over the protero- 
suchian condition lie in the fact that the pelvis has progressed to 
a more typical archosaurian condition in the "cleavage" between 
pubes and ischia for much of their height, with interruption of the 
primitively continuous ventral symphysis, and in the initiation of 
body armor not found in known proterosuchians. The armor is 
very lightly developed — much less than we would expect in an 
Anisian(?) predecessor of either crocodilians or phytosaurs. The 
pubis shows no indication of a trend toward exclusion from the 
acetabulum, as one might expect in a crocodile ancestor. The 
cervical ribs lack the specialization seen in crocodilians (and 
certain other thecodonts) of an anterior "spur" near the head. 



20 BREVIORA No. 385 

Still further, the pes is of a highly specialized nature, not of a 
type antecedent to either crocodilians or phytosaurs. 

Giialosiichus postcranial material. I may note that a small 
amount of postcranial material was found with the type skull of 
Gualosiichus; this included a scapulocoracoid, femur, parts of two 
tibiae, and a few bones of the pes. The elements preserved are in 
general comparable, except for larger size, to those of Chanaresu- 
chus. The anterior border of the scapula is less everted than in 
Chanaresuchus. The height of the scapulocoracoid is 148 mm, 
the femoral length 158 mm. The femoral length is close to 50 
percent the overall length of the skull, indicating that the Gualosii- 
chus skull was rather shorter in proportion to body measurements 
than that of the relatively long-snouted Chanaresuchus. Further 
postcranial remains of Gualosuchus, which I trust will be described 
presently by Sr. Bonaparte, are present in the Instituto Lillo col- 
lections. In one specimen in that collection the humerus is some- 
what less than two-thirds the length of the femur, thus indicating 
the same disproportion in limb length as in Chanaresuchus. 

For aid in the collection, preparation, and publication of the 
materials described above (as in other numbers in this series), 
I am deeply indebted to grants GB8171 and 22658 from the 
National Science Foundation. 

LITERATURE CITED 

Charig, a. J., J. Attridge, and a. W. Crompton. 1965. On the origin 

of sauropods and the classification of the Saurischia. Proc. Linn. Soc. 

London, 176: 197-221. 
Colbert, E. H. 1952. A pseudosuchian reptile from Arizona. Bull. 

Amer. Mus. Nat. Hist., 99: 565-592. 
Ewer, R. F. 1965. The anatomy of the thecodont reptile Euparkeria 

capensis Broom. Phil. Trans. Roy. Soc. London, ser. B, 248: 379-435. 
Krebs, B. 1965. Ticinosuchus ferox nov. gen. nov. sp. Ein neuer 

Pseudosuchier aus der Trias des Monte San Giorgio. Basel, Birk- 

hauser Verlag, 140 pp. 
McGregor, J. H. 1906. The Phytosauria, with especial reference to 

Mystriosuchus and Rhyddodon. Mem. Amer. Mus. Nat. Hist., 9: 

29-101. 
Reig, O. a. 1959. Primeros datos descriptivos sobre nuevos reptiles 

arcosaurios del Triasico de Ischigualasto (San Juan, Argentina). Rev. 

Assoc. Geol. Argentina, 13: 257-270. 



1972 CHANARESUCHUS 21 

RoMER, A. S. 1971. The Chanares (Argentina) Triassic reptile fauna. 

XI. Two new long-snouted thecodonts, Chanarcsiichus and Gualo- 

suchits. Breviora. Mus. Comp. Zool., No. 379: 1-22. 
Sill, W. D. 1967. Proterochampsa harrionuevoi and the early evolution 

of the Crocodilia. Bull. Mus. Comp. Zool., 135: 415-446. 
Walker, A. D. 1961. Triassic reptiles from the Elgin area: Stagonolepis, 

Dasygnathiis and their allies. Phil. Trans. Roy. Soc. London, ser. B, 

244: 103-204. 
1964. Triassic reptiles from the Elgin area: Ornitho- 

suchus and the origin of carnosaurs. Phil. Trans. Roy. Soc. London, 

ser. B, 248: 53-134. 

1968. Protositchus, Proterochampsa, and the origin of 



phytosaurs and crocodiles. Geol. Mag., 105: 1-14. 



^^ f^^L'S. CGMP. ZOOL 

^^'^ LinnARY 

MAR 1 8 1985 

B R E V I O ft-^ 

MmseuiM of Comparative Zoology 

Cambridge, Mass. 25 February. 1972 Number 386 

A REVIEW OF THE AUSTRALIAN SPECIES OF 
ELAPHROSYRON AND TELOSTEGUS, WITH NOTES 
ON OTHER GENERA (HYMENOPTERA, POMPIUDAE) 

Howard E. Evans 



Abstract. The genera EUiphrosyron and Tcloslci^'iis are reported fiom 
Australia for the first time, and a brief review of the species is presented. 
The following new species are described: E. sociiis. T. thomisivoriis. and 
T. tiinicri; T. nii^rocinerascens (Turner) is also redescribed. Notes are 
presented on the genera Ponipiliis. Ctcnoitciinis. and Fahriogeuia, and the 
following new species described: P. hclanloo, C. warragai. F. Canberra, 
and F. dilga. Notes on the nesting behavior of these and other species 
will be presented separately. 

During the summer of 1969-1970, Robert Matthews and I 
made a number of observations on the nesting behavior of 
Australian spider wasps (Pompilidae). Since the pompilid fauna 
of that continent has been little studied, we experienced much 
difficulty in identifying the species we worked on. Study of the 
types of Smith's and Turner's species in the British Museum has 
clarified many of our problems, but there remain several previously 
unnamed species for which we have field data that we consider 
worth publishing. Some of our most interesting data pertain to 
the related genera Elaphrosyron and Telo.stei^iis (neither previously 
reported from AustraHa), and 1 present here a review of the 
Australian members of those genera. Notes and descriptions 
relating to three other genera are also appended. 

The material considered here will be deposited in the following 
museums, which are abbreviated in the text as follows: 
AMS: The Australian Museum, Sydney 
ANIC: Australian National Insect Collection, Canberra 
BMNH: British Museum (Natural History), London 
MCZ: Museum of Comparative Zoology, Cambridge 
WAMP: Western Australian Museum, Perth 



2 BREVIORA No. 386 

Genus Elaphrosyron Haupt 

Eluphrosyron Haupt, 1929, Mitt. Zool. Miis. Berlin, 15: 120 (type-species: 

E. heinrichi Haupt). 

— Arnold, 1937, Ann. Transvaal Mus., 19: 40-43 (African species). 
Protelostegiis Priesner, 1955, Bull. Soc. Ent. Egypt, 39: 167 (type-species: 

P. anioldi Priesner). New synonymy. 

Both Haupt and Arnold regarded this genus as closely related 
to Episyron, with which it shares several important features, most 
notably the bifid claws and the longitudinal folding of the wings 
along a vena spuria. However, the postnotum is shaped differently 
and there is no evidence of squamose pubescence on the body. 
The genus is poorly known but is apparently confined to warmer 
parts of the eastern hemisphere. The type species was described 
from the vicinity of the Caspian Sea, and Arnold recognized one 
species from central Africa and another from southern Africa. 
The genus has not previously been reported from Australia. There 
appears to be only one species on that continent, but it is widely 
distributed and locally common. 

Arnold (1937) provided an accurate generic diagnosis, and 
there seems no need to present one here. The Australian species 
resembles the African insidiosus Smith closely. 

Elaphrosyron socius new species 

Holotype. 9 , Boundary Bend, Victoria, Australia, 25 Feb. 
1970 (H. E. Evans and R. W. Matthews) [ANIC]. 

Description of type female. Length 10 mm; fore wing 9.7 mm. 
Body and appendages entirely black. Fore wings lightly infuscated 
except very narrowly darker at basal vein and broadly darker in 
marginal and outer two submarginal cells, also with a still darker 
band broadly margining the wing beyond the cells, but not quite 
reaching the extreme margin, which is whitish; hind wing very 
lightly infuscated, darker apically. Body clothed with conspicuous 
silvery pubescence over much of head, thorax, and leg-bases (but 
pubescence dark on much of mesoscutum and scutellum, meta- 
pleura, and anterior part of propodeum); abdomen with dark 
pubescence except most of first segment silvery, also posterior 
margins of sternite 2 and tergites 2-4. Head and thorax covered 
with pale erect hair which is especially long and dense on the 
temples, prothorax, mesopleura, and propodeum; coxae with short, 



1972 AUSTRALIAN POMPILIDAE 3 

pale hairs, femora weakly hairy; first abdominal tergitc with pale 
hairs, apex of the abdomen with dark setae above and below. 

Clypeus 2.4 X as wide as high, somewhat convex, the apical 
margin weakly concave. Front broad, middle interocular distance 
.65 X head width; inner eye margins subparallel below, strongly 
convergent above, upper interocular distance .77 X lower inter- 
ocular distance; ocelli in a broad, flat triangle; postoccllar line: 
occllo-ocular hne = 5:4; third antennal segment equal to .70 X 
upper interocular distance. Pronotum short, broadly subangulate 
behind. Postnotum smooth, produced backward medially as an 
obtuse angle which is rounded at its apex. Front basitarsus with 
three long, weakly spatulate pecten spines, the apical basitarsal 
spine 1.6 X length of second segment; basitarsus also with two 
slender accessory spines nearly as long as the pecten spines, the 
accessory spines located on the inner margin and alternating with 
the pecten spines. Fore wing with the second submarginal cell 
1.4 X as wide as the third, measured below, but of approximately 
the same width when measured on the radial vein; hind wing with 
the anal vein meeting media slightly basad of the cubital fork. 

Allotype. S , same data as type [ANIC]. 

Description of allotype male. Length 7 mm; fore wing 5.8 mm. 
Coloration as in female; pubescence and erect hairs also as de- 
scribed for that sex. Fore wings subhyaline, with a brown band 
beyond the cells which docs not quite reach the apex of wing; hind 
wings subhyaline, very slightly darkened apically. 

Clypeus 2.2 X as wide as high, truncate apically. Middle inter- 
ocular distance .64 X head width, 1.15 X lower interocular 
distance; upper interocular distance .95 X lower; oceUi rather 
large, in a flat triangle; postocellar fine: ocello-ocular line = 4:3. 
First four antennal segments in a ratio of 10:4:8:9, segment three 
twice as long as thick, segments three and four together equal to 
.7 X upper interocular distance. Pronotum broadly angulate 
behind. Postnotum arcuately produced backward medially. Wing 
venation as described for female except second submarginal cell 
much larger than third, 1.6 X as wide measured below, 2.5 X 
as wide measured along the radial vein. Subgenital plate slender, 
tapering, its midline strongly elevated, surface with numerous 
strong setae and margin fringed with short, stiff setae. Genitalia 
as shown in Figure 1. 



4 BREVIORA No. 386 

Paratypes. 13 9 9,2^5, same data as type; 1 S , Yaapeet, 
Victoria, 18-22 Feb. 1970 (Evans & Matthews); 1 c5 , 5-15 miles 
south of Rainbow, Victoria, 21-22 Feb. 1970 (Evans & 
Matthews) ; I S , Wyperfeld Nat. Park, 25 miles north of Rainbow, 
Victoria, 18-23 Feb. 1970 (Evans & Matthews); 1 9, Near 
Adelaide, South Australia, 7 Jan. 1966 (O. W. Richards); 1 9, 
1 S , Dedari, 40 miles west of Coolgardie, Western Australia, 
11-21 Jan. 1936 (R. E. Turner); 1 9,55c?, Merredin, West- 
ern Australia, 13 Dec. 1935 (Turner); I 9 ,2 $ S , Perth, West- 
ern Austraha, Jan., Feb. 1914, 1936 (Turner); 1 9,1 S, 
Yanchep, Western Australia, Nov., Dec. 1935 (Turner); 1 $, 
Mundaring Weir, Western Austraha, 19-23 Feb. 1936 (Turner); 
1 9, "New Holland" [ANIC, AMS, BMNH, MCZ, WAMP]. 

Variation. The females vary in length from 7.5 to 10.5 mm. 
In some specimens the apical margin of the clypeus is suffused 
with reddish brown, and the mandibles have a variable amount of 
this color. Otherwise there is httle variation in color except that 
the female from Yanchep has slightly darker wings than usual and 
the silvery pubescent bands on the abdomen are reduced. This 
specimen also has the pecten spines unusually strongly spatulate. 

The males vary in length from 5 to 9 mm. In several the second 
submarginal cell is only slightly larger than the third, and there is 
some variation in the patterning of silvery pubescence, but other- 
wise there is no noteworthy variation in this series. 

Remarks. The females differ from those of the African species 
insidiosiis Smith in the following particulars: clypeus slightly con- 
cave apically; third antennal segment considerably shorter; two 
strong accessory spines present on the front basitarsus in addition 
to the three pecten spines. I have seen no males of insidiosus, 
but I judge from Arnold's figures that the genitalia bear a close 
resemblance to those of socius but have somewhat differently 
shaped volsellae and parapenial lobes. 

Genus Telostegus Costa 

Telostegiis Costa, 1887, Prospetto Imenotteri Italiani, II: 88 (type-species: 
T. major Costa). — Haiipt, 1930, Mitt. Zool. Mus. Berlin, 16: 703-718 
(Paiaearctic spp.). — Arnold, 1937, Ann. Transvaal Mus., 19: 35 
(African spp.). — Priesner, 1955, Bull. Soc. Ent. Egypt., 39: 168-183 
(Egyptian spp.). 



1972 AUSTRALIAN POMPILIDAE 5 

This genus is very closely related to Elaphrosyron, the major 
difference being that the second transverse cubital vein is absent. 
In all other respects, including the male genitalia, the two genera 
are scarcely separable. Our field notes suggest that the two genera 
are much alike ethologically. 

Telostegus is widely distributed in warmer parts of the Eastern 
Hemisphere, but it has not previously been recorded from Aus- 
tralia. Although the specimens from that continent display little 
morphological diversity, I believe that they represent three species 
(one known only from males). The species may be separated by 
the following key: 

Females 

Vertex weakly arched above the eye tops; antennae elongate, third segment 
at least .75 X upper interocular distance; known specimens with a wing 
length of 7-8 mm iiigrocinerascens (Turner) 

Vertex strongly arched above tops of eyes; antennae shorter, third segment 
.50 to .58 X upper interocular distance; known specimens with a wing 
length of 4.5 to 6 mm thoniisivorus new species 

Males 

1. Third antennal segment about 3 X as long as thick, third and fourth 

together subequal to upper interocular distance; postocellar line 
barely exceeding ocello-ocular line; vertex little arched above eye 

tops; genitalia as in Fig. 3 iiigrocinerascois (Turner) 

Third antennal segment 2.0-2.7 x as long as thick, third and fourth 
together equal to from .65 to .80 x upper interocular disiance; 
postocellar line: ocello-ocular line = 14:11; vertex strongly arched 
above eye tops 2 

2. Clypeus 2.5 x as wide as high; tibial spurs nearly as dark as the bgs; 

volsellae slender, nearly parallel-sided (Fig. 4). .tiirneri new species 

Clypeus 2.8 X as wide as high; tibial spurs stramineous, much lighter 

than legs; volsellae much expanded, somewhat truncate apically 

(Fig. 2) thoniisivorus new species 

Telostegus nigrocinerascens (Turner) new combination 

Aporus nigrocinerascens Turner, 1910, Proc. Zool. Soc. London, 1910: 334 
(type: $ , MacKay, Queensland, Australia, summer 1899, Turner Coll. 
[BMNH]). 



6 BREVIORA No. 386 

Description of type female. Length 8.6 mm; fore wing 7.5 mm. 
Black; antennae dark brown; legs black to dark brown, tibial spurs 
dusky testaceous. Fore wings subhyaline, with a brown cloud in 
the marginal and second submarginal cells that is partially sepa- 
rated from a brown band along the outer wing margin (not quite 
reaching the wingtip, which is hyaline); hind wings subhyaline, 
slightly darker apically. Body clothed with conspicuous silvery 
pubescence over much of head, thorax, and leg bases (but 
pubescence dark on much of mesoscutum, scutellum, and base of 
propodeum); abdomen silvery-pubescent over first segment and 
second sternite, also in a narrow posterior band on tergite 2, 
otherwise dark. Head and thorax covered with pale erect hair 
that is especially long and dense on the temples, prothorax, meso- 
pleura, and propodeum; abdomen with sparse, dark setae ventrally 
and on the apical two segments dorsally. 

Clypeus 2.7 X as wide as high, its apical margin truncate, 
polished and slightly elevated. Front broad, middle interocular 
distance .64 X head width; inner eye margins strongly convergent 
on the upper half, upper interocular distance .8 X lower; vertex 
passing nearly straight across between eye tops; ocelli in a flat 
triangle, postocellar line: ocello-ocular line = 7:6; third antennal 
segment equal to .8 X upper interocular distance. Pronotum 
short, broadly subangulate behind. Postnotum arcuately produced 
backward medially. Front basitarsus with three long pecten 
spines, the apical one 1.2 X the length of segment two; basitarsus 
also with two accessory spines ventrally, one quite short and the 
other nearly as long as the pecten spines. Fore wing with the 
second submarginal cell 2.5 X as wide as high; hind wing with 
the anal vein meeting media slightly basad of the cubital fork. 

Plesiallotype. $ , same data as type [BMNH]. 

Description of plesiallotype male. Length 6 mm; fore wing 
5 mm. Black; antennae and legs, including tibial spurs, dark 
brown. Fore wings subhyaline, with a broad brown band along 
the outer margin; hind wings subhyaline. Body pubescence as 
described for the female. Head, thorax, and abdominal venter 
witli sparse erect hair, including pale hairs on each side of the 
propodeum. 

Clypeus 2.4 X as wide as high, its apical margin truncate. 
Front of moderate breadth, middle interocular distance .66 X head 



1972 



AUSTRALIAN POMPILIDAE 




Figures 1-6. Male genitalia of Pompilidae, ventral aspect (drawn to 
same scale). Fig. 1, Elaphrosyron sociiis n. sp. Fig. 2, Telostegus 
thomisivorus n. sp. Fig. 3, T. nigrocinerascens (Turner). Fig. 4, T. 
turneri n. sp. Fig. 5, Pompilus belardoo n. sp. Fig. 6, Fabriogenia Canberra 
n. sp. 



8 BREVIORA No. 386 

width; upper interocular distance .95 X lower; vertex weakly 
arched above eye tops; postocellar line: ocello-ocular line = 11:10. 
First four antennal segments in a ratio of 15:8:20:19, segment 
three 3 X as long as thick, segments three and four together equal 
to the upper interocular distance. Pronotum angulate behind. 
Postnotum arcuately produced backward medially. Venational 
features as in female. Subgenital plate elevated along the midline, 
pointed apically, the margin beset with strong spines. Genitalia 
as shown in Figure 3. 

Other specimens examined. 4 9 9,5 S S , same data as type 
[BMNH]; 1 9, Yeppoon, Queensland, 3-6 Feb. 1970 (H. E. 
Evans) [ANIC]; 1 9, Ku-ring-gai Chase, 20 miles north of 
Sydney, New South Wales, 4-8 Jan. 1970 (H. E. Evans) [ANIC]. 

Variation. The females from Yeppoon and from Ku-ring-gai 
Chase resemble one another and differ from the MacKay series 
in the following respects: middle and hind tibial spurs white; 
abdominal tergites 3-6 with silvery pubescence apically; posterior 
margin of pronotum rather sharply angular; front basitarsus with 
both accessory spines rather long. These females resemble those 
from MacKay in all other particulars, and I feel it probable that 
all are conspecific. 

Telostegus thomisivorus nev\^ species 

Holotype. 9 , Nilemah Station, 50 miles south of Denham, 
Western Australia, 8-9 October 1969 (R. W. Matthews, note no. 
AM22) [ANIC]. 

Description of type female. Length 7 mm; fore wing 6 mm. 
Black, except anterior margin of clypeus and much of mandibles 
dull ferruginous, tibial spurs dusky testaceous. Fore wings lightly 
infuscated, more heavily clouded at the basal vein and in the 
marginal and second submarginal cells, also with a brown sub- 
apical band, the extreme outer wing margin hyaline; hind wings 
subhyaline, shghtly darker apically. Body clothed with silvery 
pubescence over much of head, thorax, and leg-bases (but 
pubescence dark on mesoscutum, scutellum, and base of pro- 
podeum); abdomen with dark pubescence except silvery on first 
segment, second sternite, and apical margin of second tcrgite. 
Head and thorax covered with pale hair that is especially dense 
and long on the temples and propodeum; abdomen with sparse. 



1972 AUSTRALIAN POMPILIDAE 9 

dark setae ventrally and on the apical two segments dorsally. 

Clypeus 2.8 X as wide as high, its apical margin truncate, 
polished and slightly elevated. Front broad, middle interocular 
distance .68 X head width; upper interocular distance .87 X 
lower; vertex forming a strong, even arc above tops of eyes; post- 
occllar line: ocello-ocular line = 7:6; antennae rather short, third 
segment equal to only .55 X upper interocular distance. Pro- 
notuni broadly angulate behind. Postnotum arcuately produced 
backward medially. Front basitarsus with three slender pecten 
spines, the apical one 1.3 X the length of segment two; basitarsus 
also with two slender accessory spines, as described for nigro- 
cinerascens. Fore wing with the second submarginal cell 2.5 X 
as wide as high; hind wing with the anal and cubital veins inter- 
stitial. 

Allotype. $ , same data as type except taken in a Malaise 
trap (H. E. Evans & R. W. Matthews) [ANIC]. 

Description of allotype male. Length 5.5 mm; fore wing 5 mm. 
Color of body and of tibial spurs as in female; pubescence as in 
that sex except abdominal dorsum with brownish pubescence 
beyond segment one. Fore wing lightly infuscated, slightly darker 
in marginal cell and in a subapical band; hind wing subhyaline, 
slightly darker apically. Sparse, pale hairs present on head, pro- 
thorax, pleura, and propodeum; abdominal venter with a few dark 
setae, especially toward the apex. 

Clypeus 2.8 X as wide as high, its apical margin truncate. 
Front broad, middle interocular distance .7 X head width; upper 
interocular distance .9 X lower; vertex forming a strong, even arc 
above eye tops; postocellar line: ocello-ocular line = 14:11. 
First four antennal segments in a ratio of 8:3:7:7, segment three 
about twice as long as thick, segments three and four together 
equal to .66 X upper interocular distance. Pronotum angulate 
behind. Postnotum arcuately produced backward medially. Wing 
venation as in female except anal vein reaching media slightly 
before the cubital fork. Subgenital plate slender and tapering, its 
midline strongly elevated, surface with strong setae. GenitaHa 
differing from those of nigrocinerascens chiefly in having the 
vohellae more abruptly truncate apically (Fig. 2), 

Paratypes. 1 9,1 $ , same data as allotype; 1 9 , 6 $ $ , 2 
miles west of Coorow, Western Australia, 12 Oct. 1969 (Evans 
& Matthews); 1 9,1 5 , 13 miles SW of Carnemah, Western 



10 BREVIORA No. 386 

Australia, 12 Oct. 1969 (Evans & Matthews); 1 9,1 $ ,21 miles 
north of Northampton, Western Australia, 10 Oct. 1969 (Evans & 
Matthews); 1 9 , Miaboolya Beach, 9 miles north of Carnarvon, 
Western Australia, 4 Oct. 1969 (Evans & Matthews) [AMS, 
ANIC, BMNH, MCZ, WAMP]. 

Variation. The females vary slightly in size but are consistently 
smaller than the known females of nigrocinerascens (length 4.5- 
7.0 mm). The fore wings vary from hyaline to moderately infus- 
cated basally, and the abdomen often lacks silvery bands beyond 
the first segment. The middle interocular distance varies from 
.67 to .70 X the head width, the third antennal segment from .50 
to .58 X the upper interocular distance. 

The males vary in length from 4 to 6 mm. Some of the smaller 
specimens have more silvery pubescence on the abdomen than 
described for the allotype, and in some of these specimens the 
upper interocular distance exceeds the lower interocular distance 
slightly. 

Telostegus turneri new species 

Holotype. $ , YaUingup, Western Austraha, 1-12 Dec. 1913 
(R. E. Turner) [BMNH]. 

Description of type male. Length 7 mm; fore wing 5.3 mm. 
Body black; antennae and legs dark brown except tibial spurs 
stramineous. Wings very lightly infuscated except fore wing with 
a darker band along the outer margin, hind wing slightly darker 
apically. Pubescence brownish except conspicuously silvery on 
front, temples, prothorax, coxae, lower pleura, posterior part of 
propodeum, first abdominal tergite, and first two stemites. Head, 
thorax, and abdominal venter sparsely setose, the temples, pro- 
pleura, and propodeum with an abundance of pale hair. 

Clypeus 2.5 X as wide as high, truncate apically. Front broad, 
middle interocular distance .69 X head width; upper interocular 
distance .95 X lower; vertex forming a strong, even arc above 
eye tops; postocellar line: ocello-ocular line = 14:11. First four 
antennal segments in a ratio of 16:7:19:17, segment three 2.7 X 
as long as thick, three and four together equal to .80 X the upper 
interocular distance. Pronotum angulate behind. Postnotum 
subangularly produced backward medially. Fore wing with the 
second submarginal cell 2.5 X as wide as high; hind wing with 



1972 AUSTRALIAN POMPILIDAE 1 1 

the anal and cubital veins interstitial. Subgenital plate strongly 
elevated medially, pointed apically, surface with several strong 
setae and margin with a row of stout spines. Genitalia very 
similar to those of nigrocinerascens, but the volsellae more slender 
and parallel-sided (Fig. 4). 

Paratypes. 2 $ $ , same data as type except one dated Nov. 
1913 [ANIC, BMNH]. 

Variation. Both paratypes are smaller than the type (fore 
wing 3.8-4.3 mm). In the smaller specimen, the upper and lower 
interocular distances are subequal and antennal segments three 
and four only .65 X the upper interocular distance. In the larger 
male these measurements approximate those of the type. 

Genus Pompilus Fabricius 

Pompilus Fabricius, 1798, Suppl. Ent. Syst., p. 212 (type-species: P. pulcher 
Fabricius). 

The precise limits of this large, cosmopolitan genus have never 
been defined. Many generic and subgeneric names are available, 
but for the most part these are difficult to apply when the genus 
is considered from a world point-of-view. Pompilus-Vikc wasps 
are abundantly represented in Australia, and they are somewhat 
diverse morphologically, but there seems little hope of fitting them 
neatly into accepted subgenera or closely related genera at this 
time. Hence, I shall use the generic name in its broad sense and 
when discussing species simply attempt to point out the closest 
known relatives within the genus. 

Pompilus cinereus (Fabricius) 

Sphex cinerea Fabricius, 1775, System. Ent., p. 350. 
Sphex pliimbea Fabricius, 1787, Mant. Insect., I: 278. New synonymy. 
Pompilus pulcher Fabricius, 1798, Suppl. Ent. Syst., p. 249. New 
synonymy. 

This is a very common wasp in eastern Austraha, its typical 
habitat being sea beaches and sand banks along streams. I have 
studied the Fabricius specimen in the Banks collection at the 
British Museum, probably the type, and my interpretation of the 
species is based on that specimen. Australian specimens compare 
very favorably with specimens of pliimbeiis and pulcher from 
India, Africa, and Europe, even to minor details of the male 



12 BREVIORA No. 386 

genitalia. Evidently this one species ranges in suitable habitats 
throughout the warmer parts of the eastern hemisphere. Arnold 
(1937, Ann. Transvaal Mus., 19: 47) recorded it from China, 
and one assumes that it also occurs in the East Indies. 

There is an extensive literature on this species, mainly under 
the name Pompilus plumbeus. Our field observations, so far as 
they go, agree well with the accounts of various European authors. 
Arnold (1937) placed three additional species names in synonymy 
with plumbeus. 

Pompilus belardoo new species 

Holotype. 9 , Rottncst Island, Western AustraUa, 21-22 Oct. 
1969 (H. E. Evans & R. W. Matthews) [ANIC]. 

Description of type female. Length 14 mm; fore wing 11 mm. 
Body and appendages entirely black; wings moderately infuscated, 
broadly darker along outer margins. Pubescence in large part 
dark, with a blue-green sheen in certain lights, but conspicuously 
silvery as follows: much of scape, clypeus, front, and temples; 
anterior third and posterior margin of pronotum; sides of scutellum 
and sides and central part of metanotum; much of upper surface 
of tibiae; narrow posterior margins of abdominal segments one to 
four, these bands narrowly interrupted mid-dorsally and broadly 
interrupted ventrally. Body clothed with rather long, dark hairs 
over much of the head and thorax, including the scape, pro- 
podcum, coxae, and to some extent the femora; abdomen with 
dark hairs ventrally and on the apical two tergites. 

Clypeus 2.3 X as wide as high, truncate apically; malar space, 
at its minimum, about half the width of the anterior ocellus. 
Front rather narrow, middle interocular distance .56 X head 
width; upper interocular distance .82 X lower; vertex passing 
nearly straight across between eye tops; postocellar line: ocello- 
ocular hne = 6:5; third antennal segment subequal in length to 
upper interocular distance. Pronotum broadly subangulate be- 
hind. Propodeum with a median sulcus anteriorly, posterior third 
with a flat declivity. Front basitarsus with a strong pecten con- 
sisting of five slender spines, the apical one 1.7 X as long as the 
second segment, the basitarsus also with two accessory spines 
nearly as long as the pecten spines. Fore wing with the marginal 
cell removed from the wing tip by approximately its own length, 



1972 AUSTRALIAN POMPILIDAE 13 

the radial vein somewhat angled at the third transverse cubital 
vein; third submarginal cell about as wide below as the second, 
but more strongly nai rowed above. 

Allotype. S, same data as type |ANIC]. 

Description of male allotype. Length 1 1 mm; fore wing 9.7 
mm. Color of body and wings as in female, but silvery pubescence 
less extensive than in that sex, restricted to the base of the 
mandibles, sides of the clypeus and front, temples, and interrupted 
apical bands on abdominal tergites 1 and 2 and sternite 2. Head 
and thorax with an abundance of long, dark hairs, including some 
on the scape and on the coxae and to some extent the femora; 
abdomen sparsely setose ventrally. 

Clypeus convex, twice as wide as high, its apical margin trun- 
cate; malar space rather long, nearly equal to width of anterior 
ocellus. Middle interocular distance .58 X head width; upper 
interocular distance .97 X lower; vertex passing nearly straight 
across between eye tops; postocellar line very slightly exceeding 
ocello-ocular line. Third antennal segment 2.4 X as long as thick, 
third and fourth together subequal to upper interocular distance. 
Pronotum subangulate behind. Propodeum rounded, abruptly 
dechvous on posterior fourth. Apical segment of front tarsus 
symmetrical, unmodified, but the inner claw strongly curved, bifid. 
Venation as in female. Subgenital plate elevated along the mid- 
line, broadly rounded apically. GenitaHa with the basal hooklets 
absent, the aedeagus with an abrupt apical expansion that is turned 
sharply downward (Fig. 5). 

Paratypes. 6 9 9 , 5 <5 $ , same data as type [AMC, ANIC, 
BMNH, MCZ, WAMP]. 

Variation. The females vary in length from 11 to 14 mm. 
In the majority the third antennal segment is very slightly shorter 
than the upper interocular distance (.93-1.00 X this distance) 
and in three the silvery pubescence on the posterior margin of the 
pronotum and on abdominal tergite 4 is weakly developed. The 
males vary in length from 8.5 to 11 mm. Three of them have 
slightly more silvery pubescence than the allotype, including some 
on the pronotum and sides of the scutellum; however, the banding 
of the abdomen is relatively constant. 

Remarks. This species is closely related to several other 
Australian species, such as semiluctuosus Smith, but the patterning 



14 BREVIORA No. 386 

of silvery pubescence, the male genitalia, and the nature of the 
pecten spines of the female are distinctive. Belardoo is an abo- 
riginal word from Western Australia referring to coastal sand 
dunes. 

Genus Ctenostegus Haupt 

Ctenostegus Haupt, 1930, Mitt. Zool. Mus. Berlin, 16: 685 (type-species: 
Sphex cingiilata Fabricius). 

This is a dominant genus of Pompilidae in Australia and on 
adjacent islands; in fact it may well be the largest genus of the 
family in that zoogeographic region. It is closely related to 
Pompiliis and probably a derivative of that genus. Only a few of 
the species have been described. One of the commoner species, 
for which we have nesting data, is described below and compared 
with the type species. 

Ctenostegus warragai new species 

Holotype. 9 , 3 miles west of Wentworth, New South Wales, 
27 Nov. 1969 (R. W. Matthews, note no. AM92) | ANIC]. 

Description of type female. Length 12 mm; fore wing 9.5 mm. 
Body and appendages entirely black. Wings rather heavily infus- 
cated, broadly darker along outer margin. Pubescence wholly 
dark, on the abdomen with dark blue-green reflections in certain 
lights. Body with short, sparse, rather dark hairs as follows: 
front, vertex, temples, propleura, front coxae, propodeum, and 
tip of abdomen. 

Clypeus 2.4 X as wide as high, apical margin narrowly polished, 
weakly concave; malar space well developed, at its minimum 
nearly as long as width of anterior ocellus. Front narrow, middle 
interocular distance .55 X head width; upper interocular distance 
.92 X lower; postocellar line: occllo-ocular hne = 6:5; third 
antennal segment equal to .77 X upper interocular distance. 
Posterior margin of pronotum angulate. Propodeum rounded, 
with a flat, oblique declivity on the posterior third. Front basi- 
tarsus with four rather broad pecten spines, the apical one 1.7 X 
as long as the second segment. Second submarginal cell of ap- 
proximately the same width as the marginal cell, measuring 2.2 X 
as wide as high, narrowed by two-thirds above. 



1972 AUSTRALIAN POMPILIDAE 15 

Paratypes. 10 9 9, same data as type except collected by 
H. E. Evans & R. W. Matthews; 6 9 9,4 miles east of Wilcannia, 
N.S.W., 1-2, 20-21 Nov. 1969 (Evans & Matthews); 1 9, 5 
miles west of Wilcannia, N.S.W., 1 Nov. 1969 (Evans & 
Matthews); 4 9 9, Packsaddle, 111 miles north of Broken Hill, 
N.S.W., 31 Oct., 21-26 Nov. 1969 (Evans & Matthews); 1 9, 
Port Germein, South Australia, 28 Oct. 1969 (Evans & Matthews) 
[AMS, ANIC, BMNH, MCZ]. 

Variation. The paratypes vary in length from 8 to 14 mm; 
the middle interocular distance varies from .52 to .57 X the head 
width, the third antennal segment from .75 to .85 X the upper 
interocular distance. Several of the specimens have a small 
amount of silvery pubescence on the sides of the clypeus and/or 
lower front. 

Remarks. This species is closely related to the type species, 
cingulatus Fabricius, but the pubescence is wholly or almost wholly 
dark, the propodeum more hairy, and the pecten spines longer. 
I have not been able to associate any males with these females 
with any certainty. 

All specimens were collected in areas of extensive sand dunes 
or ridges. The species name warragai is an aboriginal word from 
New South Wales meaning "plenty of sand." 

Genus Fabriogenia Banks 

Fabriogenia Banks, 1941, Occ. Papers B.P. Bishop Mus., Honolulu, 16: 240 
(type-species: F. incompta Banks). 

This is the dominant genus of the tribe Auplopodini in Aus- 
tralia. Turner (1910, Proc. Zool. Soc. London, 1910: 310) 
presented a key to several of the species (under the name Pseiida- 
genia), but there are many additional species. Both Banks 
(1941) and Townes (1957, Bull. U.S. Nat. Mus., 209: 141) 
pointed out the similarity of Fabriogenia to Phanagenia, but I 
regard them as generically distinct. Not all of the species have 
a well-developed malar space and lateral spines beneath the apical 
tarsal segments, as described by Banks for the type species, but 
they agree in having a central polished area on the apical tergite 
of the female (approaching the condition in Auplopiis) , as well as 
a group of stout bristles on the mentum. The generic classifica- 
tion of the Auplopodini i:, in a deplorable state, but for the present 



16 BREVIORA No. 386 

it seems sufficient to assign the majority of the Australian species 
to Fabriogenia. 

Fabriogenia Canberra new species 

Holotype. 9 , Canberra, A.C.T., Feb. 1970 (R. W. Matthews, 
note no. AM191) [ANIC]. 

Description of type female. Length 9 mm; fore wing 8 mm. 
Body and legs black; antennae bright orange except scape and 
apical segment very weakly infuscated. Wings clear hyaline except 
fore wings with a narrow brown band at the basal and transverse 
median veins and a somewhat wider band through the base of the 
marginal cell, across the second submarginal and base of the third 
submarginal cell, and into the outer discoidal cell; tip of wing also 
darkened. Body covered with fine, silvery pubescence which is 
darker on the vertex and mesoscutum. Pale, erect hairs are 
present over much of the head, prothorax, front coxae, mesopleura, 
and propodeum; abdomen sparsely setose ventrally and densely so 
on the apical tergite. 

Clypeus convex, weakly produced medioapically, measuring 
1 .9 X as wide as high. Middle interocular distance .64 X head 
width; upper interocular distance .95 X lower; postocellar line: 
ocello-ocular line = 7:9; vertex forming a low, even arc above 
eye tops. Antennae moderately thick, third segment 3.5 X as 
long as thick, exceeding the fourth segment as 10:9, equal to .70 
X the upper interocular distance. Pronotum broadly subangulate 
behind. Postnotum, along the midline, about two-thirds the 
length of the metanotum. Propodeum evenly rounded in lateral 
view. Hind tibiae with numerous short spines laterally, smooth 
above. Second submarginal cell twice as wide as high, receiving 
the first recurrent vein slightly before the middle; third submarginal 
cell 1 .7 X as wide as second, receiving the second recurrent vein 
.3 X the distance from the base. 

Allotype. S , same data as type [ANIC]. 
Description of allotype male. Length 7.5 mm; fore wing 6.8 
mm. Black, except as follows: mandibles with a cream band 
about halfway from the base, the tips rufous; clypeus cream 
laterally and apically; sides of lower front narrowly cream; front 
legs suffused with light brown on their inner sides; hind tibial spurs 
white, other spurs more or less infuscated; antennae orange except 



1972 AUSTRALIAN POMPILIDAE 17 

basal two segments weakly infuscated, apical four segments 
strongly infuscated. Wings clear hyaline, unhanded, fore wings 
slightly darkened at apex. Body clothed with fine, silvery 
pubescence. Head and thorax extensively covered with pale, erect 
hairs, including the propodeum and front coxae. 

Clypeus 2.2 X as wide as high, truncate apically. Middle 
interocular distance .64 X head width; upper intcrocular distance 
1.1 X lower; postocellar line: ocello-ocular line = 3:4; vertex 
forming a smooth arc above tops of eyes. Antennae elongate, 
third segment subequal to fourth, 3.4 X as long as thick, .67 X 
upper interocular distance. Pronotum broadly subangulate behind. 
Postnotum medially only slightly shorter than metanotum. Vena- 
tion as in female. Abdominal sternite 6 with a pair of toothlike 
elevations bordering a median, flat area. Subgenital plate broad, 
outer apical margins rounded, extreme apex truncate; surface 
somewhat concave, margins fringed with setae. Genitalia as 
shown in Figure 6. 

Paratypes. 4 9 $ , same data as type [ANIC, BMNH, MCZ]. 

Variation. The four paratypes range in size from 7 to 9.5 mm 
and resemble the type closely in all details; the upper interocular 
distance varies from .93 to 1.0 X the lower, antennal segment 
three from .65 to .72 X the upper interocular distance. 

Remarks. This species runs to the final couplet in the key to 
Pseudagenia provided by Turner (1910, Proc. Zool. Soc. London, 
1910: 310). It is a smaller species than australis Cameron, and 
the male lacks banding on the fore wings and has the sixth sternite 
somewhat differently modified. The size is comparable to that of 
jusijormis Saussure, but the latter species is much less hairy and 
has a longer pronotum. 

Fabriogenia dilga new species 

Holotype. 9, Canberra, A.C.T., 31 January 1970 (H. E. 
Evans, note no. A118) [ANIC]. 

Description of type female. Length 9.5 mm; fore wing 8.8 
mm. Coloration of body appendages, wings, and pubescence 
exactly as described for Canberra. Head and pronotum, including 
the front coxae, with numerous pale, erect hairs; propodeum with 
a few pale hairs on each side, but the pleura weakly hairy; abdo- 
men sparsely setose ventrally and densely so on the apical tergite. 



18 BREVIORA No. 386 

Clypeus convex, strongly and subangularly produced apically, 
measuring 1.6 X as wide as high. Middle interocular distance 
.61 X head width; upper and lower interocular distances subequal; 
postocellar line: ocello-ocular line = 2:3; vertex forming a low, 
even arc above eye tops. Antennae elongate, third segment 4 X 
as long as thick, exceeding the fourth segment as 10:8, equal to 
.82 X the upper interocular distance. Pronotum very broadly 
angulate behind. Postnotum along the midline about half the 
length of the metanotum. Propodeum evenly rounded, with a 
weak median sulcus. Hind tibiae weakly spinose laterally, almost 
without spines above. Second submarginal cell 1.7 X as wide as 
high, receiving the first recurrent vein slightly beyond the middle; 
third submarginal cell nearly twice as wide as the second, receiving 
the second recurrent vein .38 X the distance from the base. 

Paratype. 9, same data as type [ANIC]. 

Variation. The paratype is smaller (7.5 mm, fore wing 7 mm) 
and the third antennal segment equal to only .73 X the upper 
interocular distance. Otherwise it is very similar to the type. 

Remarks. This species is exceedingly similar to the preceding, 
but differs in the shape of the clypeus, the more elongate antennae, 
the shorter postnotum, and minor details of wing venation. Only 
the female is known. 

Dilga is an aboriginal word from New South Wales, meaning 
"a stick of wood." Two females were reared from a trap-nest, 
so I assume the species typically nests in hollow twigs or other 
cavities in wood. 



ML'S. CO MP. ZOOL 
LIHRARY 



•1 <^ ,._ 





B R E V I O 

Miaseitaini of Comparative Zoology ' 



HARVARD 



Cambridge. Mass. 15 March. 1972 Number 387 

GROWTH CHANGES IN AMPHILIMNA OLIVACEA 

(LYMAN) AND THE SYSTEMATIC STATUS OF 

AMPHITARSUS SPINIFER SCHOENERi 

Lowell P. Thomas- and Amy Schoener^ 

Abstract. Analysis of the developmental sequence in the brittlestar 
Aiuphiliniiui olivacea (Lyman) has revealed the similarity between its 
young stages and those of adult Amphitarsiis spiiiifer Schoener. Evidence 
is provided to justify synon\mizing the latter species with the former. 

INTRODUCTION 

Two recent papers published almost simultaneously have dealt 
with the enigmatic brittlestar genus Ainphitarsus. Schoener 
(1967a) described two new species of this previously monotypic 
genus and discussed its possible family affinities, while Thomas 
(1967: 126) pointed out the similarities between Ainphitarsus 
mirahilis H. L. Clark 1941 and Amphilimua olivacea (Lyman, 
1869). The information contained in these two papers has 
prompted a re-examination of Ainphitarsus spinijer and a study 
of the growth stages of Amphilimua olivacea. As a result of our 
studies we synonymize A. spinijer as a junior synonym of 
Amphilimua olivacea and discuss changes that take place during 
the growth of this species. 



' Contribution No. 1454 from the University of Miami. Rosenstie! School 
of Marine and Atmospheric Science 

- Rosenstiel School of Marine and Atmospheric Science, University of 
Miami. Id Rickenbacker Causeway, Miami, Fla. 33149 

•^ Museum of Comparative Zoology, Harvard University, 
Cambridge, Mass. 02138 



2 BREVIORA No. 387 

Amphilimna olivacea (Lyman, 1869) 

Ophiocnida olivacea Lyman. 1869, BulL Mus. Comp. ZooL, 1(10): 340 

(off Alligator Reef, Caribbean, 79 fms. ). 
Amphitarsus spinifer Schoener, 1967a, Breviora No. 267: 6, fig. 2 (NW 

Atlantic, 200 m). See Thomas (1967) for complete synonymy of 

A. olivacea. 

Diagnosis. Adult specimens (see below for discussion of young 
animals). Jaws with two, occasionally three, infradental papillae 
at apex, two or three slender oral papillae bordering each adoral 
plate; two oral tentacle scales, resembling papillae, in each oral slit. 
Arms slender, six times disc diameter; tentacle pores large; eight 
to ten proximal ventral arm plates with two tentacle scales on each 
side, innermost attenuated; beyond tenth ventral arm plate a single 
tentacle scale on each side; seven to nine slender arm spines, 
ventralmost largest; arm spines of arm segments under disc greatly 
flattened, fused together to form peculiar flanges occupying genital 
slits; dorsal arm plates slightly wider than long. Disc scales 
studded with slender spines; primary plates present, often in- 
conspicuous. Radial shields narrow, joined proximally, slightly 
separated distally; disc deeply notched at each pair of radial shields. 
Color variable, disc gray, tan, or brown, arms pink or orange. 

SYSTEMATIC DISCUSSION 

Although this species has been known for over a century, it has 
been only in the last nine years that the peculiar fused arm spines 
under the disc have been described. Early references to A. 
olivacea, in addition to omitting mention of this important char- 
acter, are generally brief and often unillustrated. Lyman's only 
figures (1871, pi. 1, figs. 7, 8) show primitive plates, although 
only indistinctly, and a view of the ventral arm and disc surfaces, 
omitting the fused arm spines. The only other illustrations prior 
to 1962 are VerrilFs (1899; pi. 42, figs. 1, la) stylized figures of 
the ventral disc surface and one row of arm spines. Again, the 
fused arm spines are omitted. Finally, Cherbonnier (1962) 
described and illustrated the fused arm spines, erroneously referring 
to them as "ecailles genitales." They were described and properly 
identified by Thomas (1967), who also considered the similar 
"winglike flanges" of Amphitarsus niirabilis to be fused arm spines. 
If the latter observation is correct, it is almost certain, in view of 



1972 AMPHILIMNA OLIVACEA AND AMPHITARSUS SPINIFER 3 

the other similarities discussed by Schocner (1967a) and Thomas 
(ibid.), that Amphitarsus mirabilis and Amphilimna olivacea are 
congeneric. External features and dissection of oral and dental 
plates indicate that Amp/iilimna olivacea belongs in the family 
Ophiacanthidae, but material of Amphitarsus mirabilis is not 
available for dissection. 

GROWTH CHANGES 

In the following section small (presumably young) and large 
(presumably adult) specimens of Amphilimna olivacea are figured 
in dorsal and ventral aspect (Figs. lA, B & 2 A, B). Synopses of 
the growth changes are given, purposely written so as to trace the 
growth sequence in reverse, going from the larger to the smaller 
individual. This was done so that specimens of the even smaller 
species, Amphitarsus spinijer (Figs. IC, D & 2C, D), could be 
viewed as initial stages of a growth series terminating with large 
specimens of A. olivacea. 

Growth changes on the dorsal surface 
Amphilimna olivacea. In the larger specimen (9.7 mm disc 
diameter) important systematic characters of the dorsal surface 
(Figure lA) appear as follows: (1) Only the central plate of the 
six primary plates is conspicuous. (2) The radial shields are 
contiguous for much of their length and are greatly attenuated. 

(3) Most of the disc is covered by fine overlapping scales, each of 
which frequently bears a spine. (4) The dorsal arm plates are 
rectangular or with a convex outer edge. 

In a smaller specimen (4.5 mm disc diameter): (1) All six of 
the primary plates are conspicuous and occupy a greater proportion 
of the disk than in the larger specimen. (2) The radial shields, 
which are less attenuated, are not always contiguous. (3) The 
fine scale covering of the disc is comprised of fewer scales than 
in the larger specimen, and the scales less frequently bear spines. 

(4) The dorsal arm plates have convex distal edges and are con- 
cave laterally. 

Amphitarsus spinijer. The holotype of this species (Fig. IC) 
has a disc 3.8 mm in diameter, and is therefore comparable to the 
specimen of A. olivacea illustrated in Figure IB. Here: (1) The 
six primary plates are conspicuous, although in this case they are 



BREVIORA 



No. 387 




Figure 1. Dorsal views of specimens arranged in order of decreasing size. 
A. Amphilimna olivacea (9.7 mm); B. A. olivacea (4,5 mm); 
C Amphitarsus spinifer (3.8 mm); D. A. spinifer (ca. 3 mm). 



1972 AMPHILIMNA OLIVACEA AND AMPHITARSUS SPINIFER 5 

less perfectly arranged and other large irregular scales are present 
on the disc. (2) The radial shields are separated for their entire 
length and, although longer than wide, do not appear greatly 
attenuated. (3) The fine disc scales bear spines only occasionally. 
(4) The dorsal arm plates are nearly identical to those figured for 
A. olivacea of this size (Fig. IB). 

Figure ID illustrates a paratypc of A. spinifer whose disc 
diameter is smaller (ca. 3 mm) than that of the holotype. Here: 
( 1 ) The primary plates, of which one central and three radial are 
figured, are large relative to the size of the disc, occupying a sizable 
fraction of the dorsal surface. (2) The radial shields are much 
shorter, relative to their width, and are not contiguous. (3) The 
scalation of the disc consists of even fewer fine overlapping scales, 
which bear scale spines only occasionally. Here again, several 
larger scales, particularly one in each interradial area, are quite 
noticeable, although other large scales are present around the 
central disc area. (4) The dorsal arm plates, which are basically 
rectangular, show some convexity at the distal edge and slight 
indentations laterally. 

Growth changes on the ventral surface 
Amphilimna olivacea. In the specimen with a disc 9.7 mm in 
diameter: (1) Seven sets of flanges of fused arm spines are borne 
by those arm segments overlain by the disc. (2) The oral shield 
is basically triangular with a distal edge that forms a slight outward 
bulge toward the middle. (3) On the arm plates overlain by the 
disk two tentacle scales are usually present. (4) Eight or nine 
arm spines are present on each side arm plate at the point where 
the arm becomes free of the disc. (5) The ventral surface of the 
disc is covered by many small overlapping scales bearing spines. 
Figure 2B shows the smaller specimen of this species: ( 1 ) There 
are only three sets of flanges per arm. (2) The oral shield, with 
an even more convex distal side, almost forms a rhombus. (3) 
Two tentacle scales are often present on the arm plates overlain 
by the disc. (4) The arm bases have fewer arm spines (five or 
six) than in the above specimen. (5) The ventral surface of the 
disc is covered by fewer overlapping scales than in the above 
specimen. Some of these bear single scale spines. 

Am phi tars us spinifer. The specimen illustrated in Figure 2C 
(the holotype) is almost the same size as the A. olivacea shown 



BREVIORA 



No. 387 





Figure 2. Ventral views of specimens arranged in order of decreasing size. 
A. Amphilimna olivacca (9.7 mm); B. A. olivacea (4.5 mm); 
C. Amphitarsiis spinijer (3.8 mm); D. A. spinifcr (ca. 3 mm). 



1972 AMPHILIMNA OLIVACEA AND AMPHITARSUS SPINIFER 7 

in Figure 2B. In the characters considered, except perhaps for 
the length of the disc scale spines, it scarcely differs from A. 
oUvacea. 

Figure 2D shows the ventral aspect of the smaller specimen of 
A . spinijer: ( 1 ) The number of flanges in the genital area is 
reduced still further to two sets per arm. (2) The distal side of 
the oral shield is more rounded. (3) Two tentacle scales usually 
are present on the arm plates overlain by the disc. (4) The 
number of arm spines near the arm base (four or five) is less than 
that in the larger A. spinijer (above). (5) The ventral surface 
of the disc is covered by still fewer scales, which overlap less, and 
only a few of which bear spines. 

CONCLUSION 

In order to conclude on the basis of evidence from growth 
sequences that two supposed species, the larger Amphilimna 
olivacca and the smaller Amphitarsiis spinijer, are in reality only 
one species, two criteria must be satisfied. First, for specimens of 
each species that overlap in size, one must show that the variation 
in important characteristics is negligible. Second, one must be 
able to offer a logical progression of growth stages from one to the 
other. 

The first point is readily satisfied upon examination of specimens 
of the same size range. This has been done (see Figs. IB, C & 
2B, C) and the specimens are found to be very similar. 

The second point, that growth sequences should generally agree 
within reasonable bounds with those of other species investigated, 
is also satisfied. Superficially the following sequences are con- 
sistent with knowledge of developmental series for other ophiuroid 
species (Schoener, 1967b, 1969). These points include the 
following: (1) The six primary plates of the dorsal disc surface 
may become less conspicuous as the specimens in a series increase 
in size. (2) The radial shields are initially small and may elongate 
with an increase in the size of the specimen. (3) More arm spines 
are added as the adult condition is approached. (4) There may 
be an increase in the number of specialized elements (e.g., flanges 
in the genital area) as the adult condition is reached. (5) There 
will be an increase in the number of scales covering the disc if 



8 BREVIORA No. 387 

their absolute size remains constant while the size of the specimen 
increases. 

Several other characters remained fairly constant in this series. 
These were the deep notching of the disc at the distal ends of the 
radial shields, the shape of the dorsal and ventral arm plates, and 
the number of infradental papillae at the jaw apex. 

Points on which no judgment is presently made concern the 
variation in the shape of the oral shield, which in any case seems 
slight, and the fact that there is no documented sequence of 
development in which radial shields in the smallest specimens are 
initially separated but later become contiguous. However, since 
our knowledge in this area is just being expanded, this may indeed 
occur in other species. 

Based on the above evidence, it is concluded that the smaller 
species A mphitarsus spinifer Schoener is part of the developmental 
series of Amphilimna olivacea (Lyman), the latter name having 
priority. 

This research has been supported by National Science Founda- 
tion Grant GB- 16556 and a grant from the Radcliffe Institute, for 
which grateful acknowledgment is made. We also thank Claire 
Ulanoff, who illustrated our specimens. 

LITERATURE CITED 

Cherbonnier, G. 1962. Ophiurides. Exped. Oceanogr. Beige Eaux Cot. 
Afr. Atlant. Sud., 3(8): 1-24, pis. 1-7. 

Clark, H. L. 1941. Reports of the scientific results of the Atlantic 
expeditions to the West Indies, under the joint auspices of the 
University of Havana and Harvard University. The echinoderms 
(other than holothurians). Mem. Soc. Cubana Hist. Nat., 15(1): 
1-54, pis 1-10. 

Lyman, T. 1869. //; Pourtales, L. F. de. Preliminary report on the 
Ophiuridae and Astrophytidae dredged in deep water between Cuba 
and the Florida reef . . . Bull. Mus. Comp. Zool., 1(10): 309-354. 

. 1871. Illustrated catalogue of the Museum of Comparative 

Zoology at Harvard College. No. 6. Supplement to the Ophiuridae 
and Astrophytidae. Mem. Mus. Comp. Zool., 2(6): 1-18, pis. 1-2. 

Schoener, A. 1967a. Two new species of Ainphitarsiis (Ophiuroidea) 
from the western North Atlantic. Breviora, No. 269: 1-9, figs. 1-3. 

1967b. Post-larval development of five deep-sea ophiuroids. 

Deep-Sea Res., 14: 645-660, figs. 1-9. 



1972 AMPHILIMNA OLIVACEA AND AMPHITARSUS SPINIFER 



1969. Atlantic ophiuroids: some post-larval forms. Deep- 



Sea Res., 16: 127-140. 

Thomas, L. 1967. The systematic position of Amphilimna (Echinoder- 
mata; Ophiuroidea). Proc. Biol. Soc. Washington, 80: 123-130. 
figs. 1-9. 

Verrill, a. E. 1899. North American Ophiuroidea. I. Revision of 
certain families and genera of the West Indian ophiurans. II. A 
faunal catalogue of the known species of West Indian ophiurans. 
Trans. Connecticut Acad. Arts. Sci., 10(7): 301-385, pis. 42-43. 



* 



f^L'S. CrOMP. 200U' 

B R E V I O'WM 

U.\JJVGRSiTY 

Rluseuni of Comparataye Zoology 

Cambridge, Mass. x^pril, 1972 Number 388 



INTENSE LOW-FREQUENCY SOUNDS FROM 

AN ANTARCTIC MINKE WHALE, 

BALAENOPTERA ACUTOROSTRATA 

William E. Schevill and William A. Watkins 

Abstract. Intense low-freqnencv underwater sounds, somewhat similar 
to those heard from other species of Btihwitoptoa, have been recorded from 
ininke whales, Balaenoptera acuturostrata Lacepede 1804, in tlie Ross Sea, 
Antarctica. 

The small Balaenoptera f ininke whales) of the Antarctic had 
for many vears been identified as B. acutorostrata Lacepede 
I8U4, until \Villiamson 1959 and 1961 ) indicated that some 
of them might be referable to B. bonaerensis Burmeister 1867, 
which van Utrecht and \an dcr Spoel 1962) considered no 
more than a \ariety. Since our Ross Sea whales showed us only 
the top of the back and the part of the head from the blowholes 
forward . Fig. 1 ) , we could not judge whether they were this 
form or the t\pical acutorostrata, so we refrained from reporting 
the sounds till the relationship of these Antarctic minke whales 
to those in other parts of the world was more clearly defined. 
Ohsunii, Masaki, and Kawamura (1970) have now compared 
the southern and northern forms and concluded p. 116) that 
an\ differences were minor and that the Antarctic minke whale 
was not taxonomically separable from the typical northern 
Balaenoptera acutorostrata, and this conclusion we happily 
accept. 

Our recordings were made from the edge of the Ross ice 
shelf 2 km east of Cape Crozier, Ross Island, on 22 Noxember 
1964. A whale had been sighted earlier in the 4-km stretch of 
open water between the ice shelf and the loo.se pack-ice further 
out, but it was too far away for identification. A light northerly 
wind e\entually clo.sed this open water and dro\e the pack-ice 
against the ice shelf. Large (hunks of ice were forced on edge, 



2 BREVIORA No. 388 

and the loose ice was pushed together against the shelf, foiining 
a solid co\er as far as the eye could see. 

^\'e had been listening underwater for ice sounds as the pack 
came in and had forgotten about the whale sighting, when we 
were startled by the characteristic sound of a whale blow in air. 
A minke whale had thrust its head out through a hole in the 
ice far enough to breathe (Fig. 1 ). In an 8-m whale, this means 
nearly 1.5 m. There were three holes nearby, apparently kept 
open by emperor penguins, Aptenodytes forsteri G. R. Gray 
1844, in relatively thin ice that had formed behind a projecting 
tongue of the ice shelf. We were using the nearest for our hy- 
drophone, and the whales appeared in one about 5 m further 
away. 

The whale sounds had been noted, but not identified, in the 
underwater ambient before the whales' appearance. There were 
two minke whales, though at first we saw only one at a time. 
Later both were \isible at once as they blew in adjacent holes. 
The whales came from the direction of the ice pack and would 
return again in that direction after a series of blows. Five to 
ten minutes elapsed between series of four to seven blows. 

Examination of the recordinos reveals that initially there were 
two whales producing sounds, one relatively close by and a sec- 
ond at a distance. Sounds produced when the whale was close 
by often took us by surprise and massi\ely overloaded the sound 
equipment. The loud sounds were heard when a whale was near 
the breathing holes, and therefore near the h)drophone, either 
just before or just after a series of blows. The blows were barely 
audible underwater. 

The l)ackground ambient on the day of these recordings was 
filled with a wide variety and range of sounds. Though most 
of these remain imidentified, occasionally some could be matched 
to ice movement and also to a single \isit from a leopard seal, 
Hydruroa leplonyx (Blain\'ille 1820). Others were recognized 
as sounds from Weddell seals, Leptonychotes weddelli (Lesson 
1826). We had been working for some weeks in McMurdo 
Sound in an acoustically pure culture of Leptonychotes and we 
were confident that we could recognize most of their vocaliza- 
tions. Throughout this entire day's listening, these seals were 
heard quite often as they moxed from crack to crack under the 
more solid ice co\tr. They were audible except during the 
period of the approach of the minke whales. During this time, 
a period of about 30 minutes, the Weddell seals were silent. 



1972 SOUXI1S 1 Rrnr ax antarctic minkk whale 3 

EQUIPMENT AND METHODS 

The sounds were recorclcd with an LC-34 (Athmtic Research) 
hydrophone ancl a ^\'H()I-huilt ampHfier and sprin^"-dri\en re- 
corder i^\\'aikins, 19()jj. AnaKsis pla\ljack was on Clrown 800 
series) recorders. The combined frequency response was flat 
wiihiii Vl> dR from M) to 30,000 Hz. Spectograms were made 
on a Kay l^lectric 7029A Sound Spectograph. 

The h\ch'ophone was thro\vn fiom the ice shelf across thin 
ice through holes used by emperor penguins, Aptcnodytes jor- 
sleri. The hydrophone was suspended 3 meters or more in the 
water from the ice edge, but since the cable was operating at 
low impedance, no noise was generated by motion of the cable 
arainst the ice. 



\->" 



CHARACTERISTICS OF THE WHALE SOUNDS 

The minke whale '>;ounds were intense. The loudest of these 
in undistorted recording are 60 to 65 dB above the local rela- 
ti\ely high background aml:)ient, \vhich a\-eraged about dB 
re 1 d\ ne cm". Of course, we do not know how near the whales 
were, I^ut we assume they were quite near, since the loudest 
sounds occurred either a few seconds before or after a whale 
was seen in a breathing hole. Thus, 65 dB re 1 dyne/cm- ma\ 
be nearly the maximum 1 m) signal strength of these whale 
sounds. 

The sounds were composed of a single downward sweep in 
frequency, starting at 130 to 115 Hz and sw^eeping to about 60 
Hz see Fig. 2). Since the frequency sweep continued through- 
out the sound, the rate of drop in frequency varied with both 
the span of frequencies and the duration of the call. The sweep 
rate was fairly regular throughout each sound. 

Sounds that were recorded at a low enough le\-el to be free 
from distortion had no harmonic structure; they appear to ha\-e 
l^een composed of nearly sinusoidal wa\'es. The second and third 
traces \isible with each sound in Figure 2 are from reflections 
off ice walls or the bottom. 

The minke whale sounds began with gradualK increasing in- 
tensitv for the first few cycles and ended with a gradual lethic- 
tion in intensity for the last few cycles. Thus the sounds appeared 
to rise out of background and disappear back into it. The endings 
of the sounds also were further obscured b)' reflections and re\er- 
berations. Measurement of the duration of a sound depended 



4 BREVIORA No. 388 

on its relati\'e intensity oxer ambient. Indi\idual minke whale 
sounds lasted from 0.2 to 0.3 seconds. 

No repetitive pattern was evident in sequential sounds either 
from one individual or from both. Of nine sounds presumed to 
be from one whale and recorded on one continuous tape, the 
intervals ]:)etween sounds (in seconds) were approximately 19, 
8, 13, 97, 35, 10, 89, and 12. 

The sounds were not all identical, but had a basic similarity 
in duration, frequency sweep, and intensity. Differences could 
not be attributed easily to characteristics of indi\idual whales, 
though certainly that possibility exists. 

DISCUSSION 

Intense low-frequency sounds have been recorded in the pres- 
ence of other species of Balaenoptera. Schevill and Watkins 
(1962) reported a 75- to 40-Hz sound from B. physalus (Lin- 
naeus 1758), and Schevill, Watkins, and Backus (1964) iden- 
tified a 20-Hz (23-18 Hz) sound also with B. physalus. More 
recently, other low-frequency sounds have been reported from 
B. physalus, 20 Hz to 100 Hz, and B. edeni Anderson 1879, with 
average frequency of 124 Hz, by Thompson and Cummings 
(1969), and from B. musculus (Linnaeus 1758), with most 
energy below 50 Hz, by Cummings and Thompson ( 1971 ) . The 
intensity of most of these low frequencv whale sounds has been 
estimated to be 60 dB or more re 1 dyne/cm" at 1 m. 

Other sorts of sounds also have been reported from Balaeno- 
ptera. These include pulses at about 25 kHz from B. musculus 
by Beamish and Mitchell (1971), and chirps and whistles at 
1500 to 2000 Hz from B. physalus by Perkins ( 1966) . Our gear 
is capable of receiving such sounds, and with it we have listened 
to a few hundred P>alaenoptera over many years, and yet we 
have recorded only lower-frequency sounds from them. 

The minke whale sound is similar in most respects to both the 
75- to 40-Hz and the 23- to 18-Hz sounds in our recordings of 
the larger fin whale, B. physalus. The one from B. acutorostrata 
and these two from B. physalus (1 ) are relatively intense, (2) are 
composed of low frequencies, (3^ have a downward sweep in 
frequency, (4) are nonharmonic. nearly sine-wave, (5) are made 
UD of about the same number of cycles duration at the same 
relative intensity ( about 20 cycles at 40 dB above background ) , 
and (6) begin with gradually increasing intensity and end with 



1972 SOUNDS FROM AN AXTARCTU: MINKK WHALE 5 

dropping intcnsit\ . The differences in these sounds are mainly 
those related to freqiienc\-. This \\ould seem to indicate a com- 
mon method of sound production and similar acoustic structures. 
Neither the 75- to 40-H7 finback whale sound nor the 120- 
to 60-Hz sound of the minl<e whale has shown the regular, re- 
peated patterns often found in the 23- to 18-Hz finback sounds. 
(These hitler are often called simply 20-Hz pulses because they 
usually ha\e lieen examined through band-pass filters centered 
at 20 Hz. ) 

The conspicuous silence of Lept on yr holes while the minke 
whales were about is puzzling. Did thev confuse them with 
killer whales? One might expect the seals to be good enough 
cetologists to differentiate between killers and minkes. Killer 
whales, Orcinus orca (Linnaeus 1758), are frequent visitors to 
the ice edge, and the seals, one would suppose, might have de- 
\eloped some respect for them. There was no ob\ious panic, 
howe\er, in the demeanor of a band of emperor penguins con- 
gregated near the holes. Ne\ertheless, thev staved awav from 
the water during the whales' \isit. 

We presume that the minke whales sought out the holes in 
the thin ice because the shift in the hea\y pack had closed other 
breathing spaces in the \icinity. Our failure to hear the whale 
sounds again after the minkes' disappearance suggests a swim 
under the pack to some other distant open water. 

ACKNOWLEDGEMENTS 

The recordings were made during a trip sponsored bv the 
United States Antarctic Research Program and NSF s.rant GA- 
141, and the analysis and laboratory study ha\e been supported 
by contract Nonr 241.48 with the Office of Naval Research, 
Oceanic Biology Program. This is contribution number 2801 
from the ^Voods Hole Oceanographic Institution. 



BREVIORA 



No. 388 






Figure 1. A minke whale pushes its head through broken ice for a breath 
of air. The liole was about one-lialf this size l^cfore the whale began using 
it. Initially, thev had to push their heads nearly vertically through the 
ice hole, but as more and more ice was broken by their efforts, the hole 
became large enough for nearly normal (horizontal) attitudes during 
breathing. Ross Island in background. Watkins phot. 



200 — 



1972 SOUNDS FROM AN ANTARC 1 IC MIXKL WHALE 



150- 



loo- 



se- 







mimtitmmi iifmmmxi mmmmmmmitifiitiimi aiimtms 



'immniii>! m i m (mi^ias m^Hntitnmfimmniimm )iesti&^ 



Seconds 



T* 

6 



Figure 2. Fom miiikr wluilc sounds are analyzed witliout intervening 
inter\als — up to iwo niihules elapsed between sounds. J he high level ot 
the sounds relali\e lo background peiniiis .uiaKsis withniu oiivions iiuer- 
fcrence from anihitnt sound. Ilie seiond and iliird iiaces ac(oinpan\ nig 
Ihe niinke whale .sounds aie from lellections oil the bottom or otf ice walls. 
The slightly beaded ajjpearanee of liie sound traces probably is a result 
of multiple-path sound tiansnusMon with constructive and destructive rein- 
forcement of the sound as the wa\e length (from about 11.5 to 25 m) varies 
with the sweep in frequency. The elfective analyzing idter bandwidth was 
II Hz. 



8 BREVIORA No. 388 

LITERATURE CITED 

BE.\^rISH, P.. AND E. Mitchell. 1971. lUlrasonic sounds recorded in the 
presence of a blue whale Balaenoptera inusculus. Deep-Sea Research, 
18: 803-809. 

CuMMiNGS, VV. C, AND P. O. THOMPSON. 1971. Underwater sounds from 
the blue whale. Jour. Acoust. Soc. Anier.. 50, 4: 1193-1198. 

Ohslimi, S.. V. Masaki. and A. Kawamura. 1970. Stock of the Antarctic 
minke whale. Sci. Rep. Whales Res. Inst., 22: 75-125, Appendix 1, 3 pis. 

Perkins, P. J. 1966. Communication sounds of linback whales. Norsk 
Hvalfangst-Tidende, 55: 199-200. 

Schevill. W. E., and W. A. Watkins. 1962. Whale and porpoise voices, a 
phonograph record. Woods Hole, ^Voods Hole Oceanographic Institu- 
tion. 24 pp. anil phonograph disc. 

, , and R. H. Backis. 1964. The 20-cycle 

signals and Balatnoptera (fin wiiales) . Pp. 147-152, in Marine Bio- 
acoustics, W. N. Tavolga (ed.) , Oxford, Pergamon Press. 413 pp. 

Thompson, P. O., and W. C. C(iMMiNf;s. 1969. Sound production of the 
finback whale, Balaenoptera pliysalus, and Eden's whale, B. edeni, m 
the Gulf of California (.\bstract) . P. 109. Proceedings of the 6th an- 
nual conference on biological sonar and di\ing mannnals. Stanford, 
Calif., Stanford Research Institute. 

Utrechp, W. L. a AN. AND S. VAN DER Spoel. 1962. Observations on a minke 
whale (Mammalia, Cetacea) from the Antarctic. Ztschr. fiir Saugctier- 
kundc, 27: 217-221. 

AVatkins, W. A. 1963. Portable underwater recording system. Undersea 
Technology, 4 (9) : 23-24. 

■\V^iLLiAMsoN, G. R. 1959. Three unusual rorqual whales from the .Antarctic. 
Proc. Zool. Soc. (London) , 133: 135-144. 

. 1961. Two kinds of minke whale in the Antarctic. 



Norsk Hvalfangst-Tidende, 50: 133-141. 



LIHRARY 




B R E V I O R 



IMiuiseiuijii of Co:miparsitive Zoology 

Cambridge, Mass. August 11, 1972 Number 389 

THE CHANARES (ARGENTINA) TRIASSIG 

REPTILE FAUNA. XIII. 

AN EARLY ORNITHOSUGHID PSEUDOSUGHIAN, 

GRACILISUCHUS STIPANICICORUM, 

GEN. ET SP. NOV. 

Alfred Sherwood Romer 

Abstract. A description is given of the skull and skeleton of a small 
ornithosuchid thccoilont. Gidcilisuchns stipcniicicorum gen. et sp. nov., from 
the Triassic ( ? Anisian) Chanares Formation of Argentina. The skull is of 
an advanced pseiidosiichian character, with a large antorbital opening, a 
very large orbit and a lateral temporal opening of an advanced type. The 
front limbs are short, being about 3/5ths the length of the hind; the 
tibia is nearly as long as the femur. Armor consisted of a double row of 
dorsal scutes. 

INTRODUCTION 

Work in the Triassic of Argentina and southern Brazil in re- 
cent decades has resulted in the discovery of a considerable 
number of new thecodonts; some 21 genera have now been 
described from the Triassic of this area. Most are known from 
incomplete remains; howexer, in the last paper in this series 
(Romer, 1972), I have been able to give a fairly complete 
skeletal restoration of the long-snouted Chanaresuchus, and be- 
low I give a description of the nearly complete skull and skeleton 
of a small ornithosuchid. 

Gracilisuchus stipanicicoruvt, gen. et sp. nov. 

Cotnbined generic and specific description. A small ornitho- 
suchid, with a skull length on the order of 95 mm and a pre- 
sacral column of 23 segments, with a length of about 21 cm. 
Skull of advanced pseudosuchian type. Premaxillae extending 
upward back of nares, excluding the maxillae from that open- 
ing. Antorbital fenestra large, included in a recessed area of 



2 BREVIORA No. 389 

maxilla and lacrimal; antorbital bar moderately narrow. No 
pineal opening- ; a tiny postparietal bone present. V-shape of 
lateral temporal opening so pronounced that the upper part of 
the opening is closed by apposition of squamosal to postorbital 
and jugal. Basicranial kineticism lost; the pterygoids meet me- 
dially beneath the basisphenoid region. Lateral flanges of ptery- 
goids highly developed, extending direcdy outward the entire 
width between lower jaws. Normal stance possibly bipedal; 
femur and tibia combined about 1 1/2 times skull length and 
about 1 2/3 the length of humerus plus radius; tibia and fibula 
somewhat shorter than femur. Dorsal scutes are about 1/2 ver- 
tebral length and paired, each element having a \ertical lateral 
portion and a horizontal median flange that overlaps its mate. 

The generic name refers to the obviously graceful build of 
the little reptile. The specific name is in honor of Drs. Pedro 
and Maria Stipanicic, able students of Triassic stratigraphy and 
palaeobotany. 

I am indebted to National Science Foundation grant GB-2454 
for aid in the collecting of the specimen and grant GB-22658 for 
preparation and publication. 

Holotype. La Plata Museum No. 64-XI-14-11 (Field no. 
146). A slab (Fig. 1) exhibiting a skull in dorsal view, much 
of the presacral column and scattered limb and girdle material. 
Also present on the slab is the type material of Lagosuchus talam- 
payensis (Romer, 1971), and originally a gomphodont skull was 
likewise present. Collected from the Chaiiares Formation in La 
Rioja Province, Argentina, about 2 km north of the Rio Cha- 
fiares. 

Other yyiaterial. A number of further Chaiiares specimens 
include remains of the prescHt animal. These are: 

MCZ' 4117 (Field no. 153, partim). A nearly complete 
skull and jaws from the same locality as the holotype. 

MCZ 4118 (Field no. 153, partim). In the same nodule as 
the last, but separated from it by a short distance, was a speci- 
men including \entral elements of the skull, the jaws and a par- 
tial postcranial skeleton, including a well-preserved cer\-ical re- 
gion. 

MCZ 4116 (Field no. 174). A slab including a crushed 
skull and jaws and considerable postcranial material, part of it 
pertaining to a smaller reptile. From the holotype locality. 

^Museum of Comparative Zoology. 



1972 GRACILISUCHUS STIPANICICORUM 3 

Further Gracilisuchus material, collected by Sr. Jose Bona- 
parte, is in the Instituto Lillo, Tucuman. Notable is a specimen 
with an incomplete skull and the greater part of a skeleton that 
as^rees with Gracilisuchus in all identifiable features; the individ- 
ual is about 20 percent larger than the holotype. 

CRANIUM 

(Figures 1-4) 

Cranial materials are present in all four specimens listed above. 
The skull in MCZ 41 17 is nearly completely preserved, except for 
the palate, and is uncrushed. In the holotype the skull roof and 
right side of the face are seen on the upper side of the slab; Neu- 
trally there is present part of the palate and the disarticulated 
left maxilla and jugal. In MCZ 4118 the ventral margins of the 
skull and complete lower jaws are preserved. In MCZ 4116 the 
right aspect of the skull, crushed and elements disarticulated, is 
seen on one surface; on the opposite surface are disarticulated 
elements of the left side. 

Skull length to the posterior end of the table is 85 mm in MCZ 
4117, and appears to have been similar in the holotype and in 
4116 and 4118. The general proportions are those to be ex- 
pected in a moderately ad\anced pseudosuchian. As viewed 
dorsally, the shape is essentially triangular, expanding gradually 
from a slender snout to a greatest width across the temporal re- 
gion. Anteriorly the skull outline curves up sharply abo\e the 
external nares, but from this point backward there is little fur- 
ther increase in height, the roof being essentially flat and the 
height nowhere great. The external nares are moderately large. 
The facial length is not excessixe, the distance from snout to 
orbital margin being somewhat less than half the total skull 
length. The antorbital xacuities are large, and set in an oval 
recess, which is bounded anteriorh b\ a pronounced curved line 
running upward along the maxilla; it is deeply overhung dor- 
sally by the prominent lateral edge of the skull roof. The orbit 
is verv large, its diameter being about 1/3 the skull length. It 
not only occupies nearly the entire height of the face, but also 
extends medially across much of the dorsal surface of the skull. 
The temporal region, in contrast, is short. The superior temporal 
fenestra is triangular in shape, narrow anteroposteriorly but ex- 
tending broadly outward behind the postorbital bar. The lateral 
temporal opening is of unusual structure. .\s in advanced theco- 
donts generally, the posterior border is V-shaped, the apex of 



BREVIORA 



No. 389 



popr. 




Figure 1. The skull in dorsal view. This and the following skull figures 
are based mainly on ISICZ 4117 XI. Abbreviations for Figs. 1 to 5: 
a, articular; an, angular; c, coronoid; d, dcntary; ec, ectopterygoid; /, frontal; 
/, jugal; I, lacrimal; m, maxilla; n, nasal; p, parietal; part, prearticular; 
pf, postfrontal; pw, premaxilla; po, postorbital; popr, paroccipital process; 
pp, postparietal; prj, prefrontal; pt, pterygoid; q, quadrate; qj, quadratojugal; 
san, surangular; soc, supraoccipital; sp, splenial; sq, squamosal. 



the V pointing anteriorly. Here, howe\er, in contrast to normal 
advanced forms, the forward push of the V is so pronounced 
that the upper limb of the V is in contact with the postorbital 
and jugal; as a consequence the upper half of the fenestra is 
completely closed, the opening remaining being a ventral triangu- 
lar area. A similar situation is present in the aetosaurs. 

The premaxillae are thickened ventrally, with accommodation 
for tooth roots. Anteriorly each element sends a slender process 
upward to meet the nasal medial to the narial openings. Pos- 
teriorly the premaxilla sends a stout process upward to form 
part of the posterior border of the naris. 

A pronounced ridge extends backward on either side of the 
skull roof along the upper margin of the antorbital region toward 
the upper anterodorsal margin of the orbit; this pair of ridges, 
formed anteriorly by the nasals, sharply separates the flattened 
dorsal surface of the skull from the essentially \ertical lateral 
surfaces. Below this rids^e the nasal extends downward to form 
the upper boundary of the naris. This flange is in contact with 
the premaxilla ventrally, both in front of and behind the naris. 
On the dorsal surface the nasals extend well backward, broad- 



1972 



GRACILISUCHUS STIPANICICORUM 




Figure 2. The skull in lateral view. X 1. 



ening posteriorly. The lateral ridges form their lateral bound- 
aries for most of this distance; part way along the upper margin 
of the facial region, however, the width of the nasals is some- 
what constricted by the presence of the lacrimals. Nasals and 
frontals meet in a broad trans\erse suture, not readily discern- 
ible, a short distance anterior to the anterior margin of the orbits. 

The frontals are elongate, their length being roughly that of 
the large orbits. Broad anteriorly, they are constricted in width 
by the strong medial curx-ature of the dorsal orbital margins, 
the prefrontals being interposed for some distance between them 
and the frontals. Posteriori)- the frontals are bounded laterally 
by the postfrontals, and terminate at an irregular transverse su- 
ture with the parietals. 

The last elements are short in longitudinal extent, occupying, 
roughly, only the length of skull roof opposite the short temporal 
region. There is no pineal opening; posteriorly the median su- 
ture is absent. Laterally the parietals send out a flange on either 
side, anterior to the superior temporal fenestra and behind the 
postfrontal, to gain a postorbital contact. Posterior to the upper 
fenestra the parietals send out on either side strong flanges that 
form the upper margins of the occipital surface and, along the 
posterior margins of the fenestrae, overlap posteriorly ascending 
flanges of the squamosals. 

The postparietal, seen on MCZ 4117, persists as a small trian- 
gular wedge of bone that projects strongly backward from the 
middle point of the dorsal occipital surface; it is essentiallv the 
most anterior portion of the dorsal armor that continues down 
the back. 

The maxilla is prominently de\eloped. Its major ramus is a 
long strip of bone that carries the tooth bases and extends back- 
ward from a premaxillary contact to a point not far forward of 



6 BREVIORA No. 389 

the posterior border of the orbit. Anteriorly it forms the lower 
margin of the antorbital fenestra. Near the posterior hmit of 
the fenestra the anterior end of the jugal is found above the 
maxilla, and from this point back the maxilla is reduced, to be- 
come a slender splint. Anteriorly the maxilla fails to reach the 
external naris. It sends upward a broad process, bounded pos- 
teriorly by the antorbital vacuity and anteriorly by the premax- 
illa, from which it is separated by a very prominent incised 
suture. Curving upward along the surface of this process is a 
well-marked ridge separating the proper outer surface from the 
depressed area in which the antorbital vacuity develops. Dor- 
sally this maxillary process extends backward below the lateral 
skull ridge to form part of the upper margin of the antorbital 
vacuity. 

The lacrimal forms most of the vertical bar of bone between 
orbit and antorbital \acuity. This bar is of limited width; it 
has a conspicuous ridge along its anterior border, and a less de- 
veloped ridge posteriorly. Below, the lacrimal meets the jugal. 
Abo\e, the lacrimal forms the projecting process between the 
orbital rim and the lateral ridge abo\'e the facial region, and 
extends forward above the antorbital opening. It has (unusual- 
ly) a modest exposure on the dorsal surface, lateral to frontal 
and nasal. The prefrontal is a small wedge-shaped element, lat- 
eral to the frontal and posterior to the dorsal exposure of the 
lacrimal; it forms the anterior half of the curved dorsal margin 
of the orbit. 

The postfrontal is a triangular element of modest size, lying 
aboxe the posterodorsal margin of the orbit. Medially it is in 
contact with the frontal, posteriorly with a lateral flange of the 
parietal; laterally it is barely in contact with the postorbital. 

The postorbital is centered at the point of meeting of the 
postorbital bar with the bar of bone separating upper and lower 
temporal openings. The bone is here thickened, with a promi- 
nent external knob, posterior and dorsal to which it is gently 
concave externally. A \entral flange forms much of the poste- 
rior margin of the orbit, oxerlapping the jugal anteriorly. Above, 
the bone is in contact with postfrontal and parietal along the 
upper part of the postorbital bar. A posterior flange joins with 
the squamosal in the formation of the bar separating the tem- 
poral openings. 

The jugal is of typical construction, its main ramus lying be- 
low the orbit and extending from a point anterior to the orbit 
backward to a contact with the quadratojugal below the lateral 



1972 GRACILISUCHUS STIPAXICICORUM 7 

temporal fenestra. Behind the orbit a ramus ascends, shmming 
as it goes, behind the lower branch of the postorl)ital. 

The squamosal is a complex element. Its structural center 
lies at the posterior end of the bar separating the two temporal 
fenestrae. Anterodorsally a thin but deep flange extends forward 
to overlap laterally the posterolateral extension of the parietal. 
Anterolateralh a ramus extends for\\ard to form, with the post- 
orbital, the bar between the temporal openings. A backward 
extension of the bone from its center affords a broad area of 
articulation for the paroccipital process. Beneath this region the 
concave \entral surface of the bone supports the head of the 
quadrate. Forward and somewhat ventrally from this region a 
broad flange of bone forms a firm union with the dorsal end of 
the quadratojugal. In many pseudosuchians this flange forms 
the upper half of a \^-shaped posterior border of the lateral tem- 
poral fenestra. Here, however, (as previously noted) an unusual 
condition exists. This flange turns so sharply forward that the 
upper half of the "normal" lateral opening is obliterated, and 
the flange is apposed to the postorbital, which forms the anterior 
margin of the upper part of the lateral opening under "normal" 
conditions. 

The quadratojugal is a well-developed element. It forms the 
posterior part of the skull rim below the lateral temporal fenestra, 
between jugal and quadrate. A broad ramus extends upward 
and forward posterior to the lateral fenestra to terminate be- 
neath the anteroventral ramus of the squamosal. The quadrate 
is highly de\eloped. It presents a convex, trans\ersely broad- 
ened, articular surface for the lower jaw on the under side of 
its posterior termination. From the medial edge of the articular 
area a sharply defined ridge runs upward and somewhat forward 
to terminate beneath the squamosal. The quadrate is broadly 
developed lateral to this ridge for nearly the entire height of the 
bone; this area is somewhat concave in transverse section. The 
lateral margin of the quadrate is in contact with the quadrato- 
jugal for most of the extent of the two bones. Not far abo\-e 
their \entral margins, howe\er, there appears to be a small fora- 
men, as in many early tetrapods, between the two bones, and 
quadrate and quadratojugal separate dorsally to reach their dif- 
fering termini. Medial to the major vertical ridge on the quad- 
rate there is seen a thin sheet of bone runnino- mediallv for some 
distance in the position appropriate for the quadrate ranuis of 
the pterygoid; at the bottom of this sheet is a cur\ed ridge, run- 
ning upward and anteromedially from the articular region of the 



8 



BREVIORA 



No. 389 



pop 




Figure 3. Palatal surface of the skull; the anterior portion of the palate 
is not preserved in available material. X 1. 



quadrate. I am unable to determine how much of this bone per- 
tains to the quadrate and how much to the pterygoid. 

The upper rim of the occipital surface is formed by the two 
posterior flanges of the parietals, which curve backward, outward 
and somewhat downward to their lateral points of termination 
near the posterior angles of the squamosals. Between the two 
flanges, as noted earlier, is the small projecting postparietal ; be- 
neath, on either side, are the posttemporal fenestrae. Below and 
beneath the central part of these flanges is a broad plate of bone, 
the supraoccipital. It is essentially flat, but with a slight dorso- 
ventral swelling in the mid-line. Fused with the supraoccipital 
on either side are the paroccipital processes, which extend out- 
ward to form a broad (but not tightly sutured) union with the 
squamosals. The paroccipital processes are relatively narrow 
proximally, expanding somewhat in vertical breadth distally and, 
except at their bases, thin anteroposteriorly; they are roughly 
oar-shaped. Exoccipitals and basioccipitals have been lost on 
MCZ 4118. They appear to be present in the crushed remains 
of the occiput in the holotype, but little detail can be made out. 

The palatal surface is poorly preserved in available specimens. 
Posteriorly is a pair of well-developed basisphenoidal tubera. 
I ha\'e obtained no data regarding the lateral walls of the brain- 
case in the otic region. No stapes has been discovered. The 
two quadrate rami of the conjoined quadrates and pterygoids 
slant strongly inward anteriorly, so that the two pterygoids are 



1972 GRACILISUCHUS STIPANICICORUM 9 




Figure 4. Occipital view of the skull. X 1. 

apposed in the midline below the basisphenoid; it seems reason- 
ably certain that moxement of pterygoids on the braincase had 
been lost. A short backward median projection is present on 
each pterygoid at its point of apposition. Transverse pterygoid 
flanges are highly developed, extending straight out laterally and 
somewhat \entrally to occupy the total space available between 
the lower jaws when in occlusion. A pair of ridges extends out- 
ward along the under surface of each flange. Anterior to the 
flange, a curxed sheet of bone, on which there appears to be a 
suture between pterygoid and ectopterygoid, slants upward an- 
teriorly for a short distance. No data are available regarding the 
anterior portion of the palate. 

The jaws (Figs. 2, 5) are slender. The symphysis is moderate- 
ly elongate but shallow and is formed entirely by the dentaries. 
The dentary occupies practically the entire outer surface of the 
jaw for more than half its length. Posteriorly the dentary has a 
V-shaped suture with the surangular dorsally and a diagonal 
suture with the angular \entrallv. Between dentarv and surangu- 
lar abo\e and angular below is a typical archosaur lateral man- 
dibular fenestra. There is a well-de\eloped retroarticular process 
behind the broadly concave articular cavity. A splenial is well 
developed, occupying a considerable area on the inner surface 
of the jaw, but is barely visible externally. The prearticular 
forms a buttress at the anterior marsfin of the articular surface 
and thence extends forward and downward below the adductor 
fossa. The jaw is strongly compressed mediolaterally, and hence 
the large fossa looks medially rather than dorsally. The dorsal 
rim of the fossa is marked by a well-developed longitudinal ridge 
along the upper margin of the surangular. I am not sure whether 
or not a coronoid was present at the anterior margin of the 
fossa; the material is imperfect but suggests that a thin sheet of 
coronoid may have been present anterior to the adductor fossa. 



10 BREVIORA No. 389 



son 




Figure 5. Internal surface of the lower jaw. Sutures in the presinned 
coronoid region are uncertain. X 1. 



The teeth of Gracilisuchus are of the typical thecodont type 
common among carnixorous archosaurs — somewhat flattened 
mediolaterally, sharp-pointed and curxed somewhat posteriorly 
toward their tips. Only two small teeth are definitely preserved 
in the only specimen in which the premaxillae are present, but 
the space available suggests the presence of a third. A disarticu- 
lated maxilla of the txpe skull has fortunately preserved a nearly 
complete series of maxillary teeth, not fully thecodont. About 14 
appear to have been present. From a small first tooth, there is 
a steady increase in size to the fourth, following which there is 
a steady reduction to small elements for the last half dozen of 
the series. The lower teeth are not fully preserved, but the evi- 
dence suggests that none were of large size, and that there was a 
rather e\en row of small teeth, spaced about 2.5 mm apart, to 
a total of 16 or so. 

AXIAL SKELETON 

A considerable amount of xertebral material of Gracilisuchus 
is present in the collection. The holotype when entombed pos- 
sessed a complete articulated presacral series ( Fig. 6 ) . However, 
the nodule in which the specimen was preserved had undergone 
considerable damage before collection. A spht had occurred in 
the nodule that slanted back down most of the length of the 
series of dorsal vertebrae and, with the loss of a large chip adja- 
cent to it, caused the complete loss of the posterior cervical and 
anterior dorsal vertebrae and damage to the remainder of the 
dorsals. The second and third vertebrae from the end of the 
series are obviously sacrals. Most of the ribs of the right side 
are preserved in their original position. This aids greatly in de- 
termining the spacing of the missing vertebrae, as does, further, 
a calculation, from known lengths of cervical and posterior dor- 
sals, of the number of vertebrae contained in the missing seg- 
ment of the column. As a result, it seems rather certain that 23 
presacrals were present in life — a reasonable thecodont count. 



1972 



CKACII.lSLCll us S'lIPANICICOKL'.M 



11 




X 



u 
^ 



h 



tc 



be 
'iZ 



12 BREVIORA No. 389 

In MCZ 4118 a number of short sequences of articulated 
vertebrae are present: 

( 1 ) A series of cervicals articulated with the fragmentary 
skull. 

(2) Six articulated centra that appear to lie in the cervical- 
dorsal transition. 

(3) Six vertebrae, three well preserved, lying close to the 
skull-cervical series; they are probably anterior dorsals. 

(4) Five poorly preserved vertebrae, presumably dorsals. 

(5) Nine vertebrae, probably posterior dorsals, sacrals and 
proximal caudals. 

(6) T.WO poorly preserved dorsals, with ribs. 

(7) Ten caudals, probably from the middle of the tail. 
These series will be referred to below by number. Excluding' 

no. 6, 35 vertebrae are present, most of which are surely presa- 
crals. It is obvious that not all belong to one individual; some, 
apart from Jios. 1 and 3, may have been associated with MCZ 
4117, a skull found in the same nodule. 

The Tucuman specimen mentioned aboxe has present much 
of the cohunn; in all observed respects it agrees well with the 
material from the holotype and no. 4118. MCZ 4116 contains 
a melange of vertebral material, some of which appears to belong 
to Gracilisiiclms (as does the skull included in this slab). 

The cerx'icals are best seen in the series (no. 1 ) connected with 
the skull remains in 4118 (Fig. 7); a similar but less well-pre- 
served series is present in the holot\pe. The atlas is partially 
concealed by other materials, but as far as can be seen, has a 
typical archosaur structure — a well-developed intercentrum, 
above which arc paired neural arches, and back of this articular 
ring a small atlas centrum and a small axis intercentrum. The 
axis has a well-developed centrum and a neural arch and spine 
of stout construction. The vertebrae in the cervical region pos- 
terior to the axis are similar in pattern, but with a less expanded 
neural spine. In these vertebrae the spines slant forward dor- 
sally; in typical members of this series they are, as preserved, 
covered by dermal armor at their tips; isolated posterior cervi- 
cals and anterior dorsals show that their tips were flattened, 
obviously for close apposition of armor scutes. Two well-de- 
veloped apophyses for rib articulation are seen on each centrum 
from the axis backward. The capitular articulation is a short 
parapophysis developed low down anteriorly on the centrum and 
terminating in a round articular area. 'Fhe area for tubercular 
attachment is a short transverse process slanting sharply down- 



1972 



(iRACILI.SUCHUS STIPANIGICORUM 



13 




Plate 1. Skeleton of Gracilisiirhiis stipanicicoiiini, restored. Dermal slioiililer elements, manus. pubis, and distal part of tail unknown. X 1/2 



14 



BREVIORA 



No. 389 







Figure 7. a. Dorsal scales of the cervical region, seen lioni above, b, The 
cervical region in side view (the atlas is not induded) . r. Incomplete rihs 
ol ihe riglil side, hansilional l)el\veen cervical and th)rsal series, d, A doisal 
vertel)ra in hiteial and anterior views. All from MCZ 4118. X 1. 



ward fioni a point near the anterior edge of the neural arch. 
By the tinie the seventh \'ertebra is reached, the diapophysis 
has moved to a point somewhat higher on the arch and is di- 
rected more laterally. The parapophysis, lunvever, appears to 
have been persistently ventral in position in the cervical region. 
No intercentra are preserxed ])ack of the atlas. In MCZ 4118 
cervicals two to six are completely preserved, the seventh is in- 
complete; in the holotype six vertebrae are similarly preserved. 
In MCZ 4118 the typical cervicals average 10.5 mm in length 
and in the holotype the length is approximately the same. From 
the axis back all centra are keeled \entrally. 

Of the posterior cervicals and anterior dorsals, nothing can be 
made out on the holotype. Associated, however, with 4118 are 
two series of vertebrae that appear to represent this area. One 
such series (no. 2) includes six articulated centra and a fraction 
of the next posterior centrum; central lengths here average 8.5 
mm. Neural arches and spines are not preserved, but ribs are 
present on four, the first two well preserved. Since the first rib 
of the series resembles that which (as noted later) appears to be 
associated with vertebra 7 in the holotype, it is reasonable to 
assu!ue that this series includes \crtebrae 7—12 as well as part of 
centruiu 13. In presumed xertebrae 7-9 the capitulum still ar- 
ticulates well down the side of the centrum, and the build of 



1972 GRACILISUCHUS STIPAXICICORUM 15 

tubercular process indicates that in these segments the transverse 
process still slanted strongly downward. 

Further vertebrae from the anterior part of the column are 
present in series no. 3, thiee in articulation and well preser\-ed. 
^Ve are dealing here with typical dorsals (Fig. 7). \>rtebral 
lengths a\erage close to 8 mm. Rib attachments are markedly 
different from those of the series described abo\e. The diapophy- 
ses now extend directh out from the arches. In the first of the 
three complete \ertebrae the parapophysis lies about at the 
boundary between centrum and arch at the anterior edge of 
the %-ertebra ; in the third member of the series, the parapophysis 
is well up on the side of the arch, not far hdow and anterior to 
the base of the diapophysis. The neural spines are broad antero- 
posteriorly, but low, with a height of but 7 mm above the zyga- 
pophysis. Their upper ends are broadly o\al, for apposition to 
the dorsal armor. In the arch region the surface is poorly pre- 
ser\-ed, so that little can be seen regarding rib articulations. 

Series 4, 5 and 6 are poorly preserved and crushed, so that 
it is difficult to determine the nature of the apophyses, and no 
ribs are attached, except in no. 6 where the vertebrae show 
little). An ilium lies close to the posterior end of series 5, sug- 
gesting that this sequence may include proximal caudals and 
sacrals as well as posterior dorsals. No chevrons are present, and 
the condition of preser^•ation of the lateral surfaces of the verte- 
brae is such that it is impossible to determine whether sacrals are 
included. It seems clear that, as in archosaurs generally, the 
capitular articulation has joined the tubercular one in originat- 
ing from the transverse process; details, however, are not clear. 
In no. 5 central lengths are about 8 mm. In series 6 the mean 
central length is a little o\er 7 mm; the neural spines are broad 
anteroposteriorly but low, rising only about 5 mm abo\"e the 
zygapophyses, and the total height of a vertebra is about 13 
mm. As preserved, the centra in series 5 and 6 are prominently 
keeled \-entrally and thin from side to side, but this may be due 
to post-mortem compression. 

One further series of \-ertebrae on the 4118 slab, no. 7, in- 
cludes 10 articulated caudal vertebrae, evidently from the mid- 
portion of the tail. Central length of an anterior member of this 
series is 6 mm. that of posterior members 5.5 mm. Chevrons 
are present, the most anterior one being 10.5 mm in length; in 
contrast, that between the sixth and se\enth of the series is but 
7 mm long. The vertebrae are still fairly tall, an anterior mem- 
ber measuring 8 mm from bottom of centrum to zygapoph\"sis, 



16 BREVIORA No. 389 

while at the posterior end of the series this height is somewhat 
reduced. The first two members of the series have well-developed 
spines, slanting backward and rising to a \ertical height above 
the zygapophyses of 6 mm. Posteriorly the spines are reduced, 
and at the end of the series are merely low triangles above the 
zygapophyses. 

As noted earlier, sacral vertebrae are present, although poorly 
preserved, in the type. The Tucuman specimen (Fig. 8) in- 
cludes well-preserved sacrals in articulation with ilia. Beyond 
them are 16 articulated caudal vertebrae. Anterior members of 
this series have a central length of about 9 mm, posterior seg- 
ments about 7.5 mm. In the anterior members of the series the 
neural spines are narrow anteroposteriorly, but moderately high, 
with heights of 10 mm or so above the zygapophyses; they are 
tilted posteriorly. In the posterior members of this series the 
spines are much reduced in height. The first chevron is present 
between vertebrae 3 and 4. Most chexrons are imperfect dis- 
tally, but that between vertebrae 8 and 9 is 1 7 mm long. If we 
compare this series with that of no. 7 in MCZ 4118 — and take 
into account the larger size of the Tucuman specimen — it 
would seem that the 4118 series includes segments comparable 
to the posterior part of the Tucuman series. 

Vertebrae present on the holotype slab, although disconnected 
from the main specimen, appear to represent much of the length 
of the tail. One series of seven \ertebrae, and part of an eighth, 
is articulated and for the most part well preserved. Each meas- 
ures 8 mm in length. The neural spines are low, and capped by 
dermal plates to give a total height of about 8 mm above the 
le\el of the zygapophyses. The transverse processes are incom- 
pletely preser\ed. There then follows a series of about 14 par- 
tially disarticulated and poorly preserved vertebrae which, as 
indicated by long transverse processes, are definitely caudals. Of 
several that are moderately well preserved, the length of the cen- 
trum is, again, about 8 mm; the width over the transverse proc- 
esses is 15 mm. Beginning near the end of this series is a further 
series of about 15 poorly preser\ed caudals of smaller size, with 
lengths of 5.5 to 6 mm in the distal members. There is thus defi- 
nite evidence of 39 caudals making up a considerable part of the 
presumed length of the tail. 

Cervical ribs (Fig. 7) are well seen in both 4118 and the 
holotype. There is no evidence as to the presence or absence of 
an atlantal rib. Typical members have the highly specialized 
pattern developed among certain early archosaurs and retained 



1972 GRACILISUCHUS STIPAXICICORUM 17 

todav in crocodilians. Each rib is essentially plow-shaped, the 
two "handles"" represented b\ conxerging rami running outward 
from parapophysis and diapophysis. The "blade" of the plow, 
formed distally by the union of the two "handles," includes a 
short point anteriorly and a long posterior extension. In typical 
cer\icals each "blade"' overlaps its more posterior neighbor to 
make a continuous rib series from the axis as far back, at least, 
as vertebra 7. As in the case of the vertebrae, the rib transition 
to the dorsal series is somewhat uncertain. In series 2 of no. 
4118 (Fig. 7), four incomplete ribs are present on the left, and 
one on the right. The first three ribs extend strongly forward 
from the point of union of capitulum and tuberculum to form 
a triangular sheet of bone, the front end of which underlies the 
next anterior rib. The main shaft of the rib extends outward 
and downward in line with the tuberculum, rather than turning 
backward as in typical cervicals; shaft lengths are uncertain. The 
width of the proximal rib expansion presumably associated with 
serratus muscles) decreases from the first to the third of the 
series, and rib 4 appears to lack any expansion. 

In the holotype most of the dorsal ribs are preserved. Mid- 
dorsals have a\"erage lengths of 62 mm. There is, as expected, a 
diminution in length toward the sacrum. The fourth presacral 
rib measures but about 46 mm, the three following, as preser\ed, 
39, 18 and 10 mm. The main shaft of typical dorsal ribs aver- 
ages but about 1 mm in diameter for most of their length; they 
thicken somewhat toward their heads. The direct proximal 
course of each rib is toward the tubercular attachment; the capit- 
ular head slants downward medially from the course of the shaft 
to extend se\eral millimeters further than the tuberculum. The 
ribs show considerable curxature proximally, little distally, thus 
suggesting (reasonably) that the trunk was relati\ely high and 
narrow in its proportions. 

DERMAL ARMOR 

As in manv other pseudosuchians, Gracilisuchus was armored 
dorsallv. Best preserved is the armor of the cervical region (Fig. 
7). The plates, thin but highly sculptured, are paired, and ap- 
proximately two pairs are present for each vertebral segment. 
When articulated, the series of plates form a dorsal shield with 
a flat area, about 5 mm wide, running longitudinally down the 
column over the neural spines and with, on either side, a ver- 
tical sheathing about 5 mm in height. A sharp ridge separates 



18 BREVIORA No. 389 

dorsal and lateral portions of each plate laterally; each plate 
oxerlaps its posterior neighbor. The plates of either side join to 
form the dorsal area, each member of a pair overlapping dor- 
sally (as laterally) its posterior neighbor, and with members of 
each pair overlapping its partner; in 4118 the left plates overlap 
the right. Anteriorly, over the atlas region, the plates appear to 
narrow dorsally as a pointed terminus, but details are uncertain. 

The plating of the dorsal region is imperfectly preserved in 
available material. Only small plate fragments are present in 
the isolated dorsal series found with 4118. In the badly pre- 
served dorsal series of the holotype, a lateral plate covering is 
seen in the area of the third to fifth presacral neural spines, in- 
dicating that the type of plating seen in the cervical region was 
continued down the back. In the holotype the series of proxi- 
mal caudals described definitely carry dorsal scutes, although 
details are somewhat obscure. 

No articulated series of abdominal ribs is preserved, but in the 
holotype a scattered series of typical archosaur gastraHa are 
present in the area near the shoulder girdle and front legs de- 
scribed above. Those gastralia that are completely preserved 
measure about 35 mm in length. They are essentially straight 
for most of their length, but gently curved toward their presumed 
medial ends. 

APPENDICULAR SKELETON 

Much of the girdles and appendages are preserved, although 
partially disarticulated, in the holotype; a number of elements 
are preserved in MCZ 4118; disarticulated elements are to be 
found on the MCZ 4116 slab. 

In Gracilisuchus the front limbs are much shorter than the 
hind, and hence, as expected, the shoulder girdle is of small 
size. A right scapulocoracoid (Fig. 8) is present in the type (as 
well as an incomplete left scapula ) . The height of the scapula 
is 24 mm. The structure is typically thecodont; the scapular 
blade is slender, but ventrally the anteroposterior width increases 
to 1 1 mm before the anterior margin retreats to the clearly 
marked scapulocoracoid suture. The back margin ventrally is 
strongly bevelled off for a prominent area of articulation for the 
humerus, facing diagonally outward and backward. A some- 
what less marked articular area is present below on the coracoid. 
This latter element, with rounded borders, is much broader an- 
teroposteriorly than dorsoventrally, the measurements concerned 



1972 



GRACILISUCHUS STIPANICICORUM 



19 





Figure 8. a, Right shoulder girdle, humerus, radius, and ulna. From the 
holotype. b. Left ilium and ischium, c. Sacral ribs and ilia, seen from above. 
From a specimen in the Instituto Lillo, Tucuman. All X 1. 



beino- 1 7 mm and 1 1 mm ; there is thus no suggestion of croco- 
diloid build. Just below the suture with the scapula the coracoid 
is slightly notched anteriorly; below this point the bone expands 
anteriorly to a considerable degree. A coracoid foramen is pres- 
ent anteroventral to the articular area. A similar scapulocora- 
coid is present on MCZ 4116. Of dermal girdle elements, none 
are identifiable with certainty in the material studied. Of pehic 
girdle elements ,Fig. 8), iha are present in the holotype, in 
MCZ 4116, MCZ 4118, and in the Tucuman specimen men- 
tioned. The element is of generalized thecodont proportions, 
with an iliac blade extending only a short distance anterior to 
the acetabulum, but much better developed posteriorly. The 
upper edge of the blade is thin; below, howe\er, it swells con- 
\exly on the inner surface to allow for exca\ation of the aceta- 
bulum externally. The acetabular margin, semicircular in out- 
line, is well defined ; it is deeply incised into the bone, particularly 
anteriorh , where the margin de\elops as an oxerhanging shelf. 
The lower margin of the bone is convex in outline, with some 
differentiation of pubic and ischiadic contacts. It is ob\ious that 
the pelvis was imperforate. 



20 BREVxORA No. 389 

Little evidence of the ventral elements of the girdle can be 
identified in the material. I have found nothing in the available 
specimens that I can identify with confidence as pertaining to 
the pubis. Imperfectly seen ischia are present in the Tucuman 
specimen and a pair of conjoined ischia are present in MCZ 
4116. Unfortunately, the front margins of the ischia are imper- 
fectly preserved, so that only a fraction of the acetabular margin 
is present and nothing can be said concerning the relations of 
ischium and pubis. The bones are bladelike, tapering posteriorly 
and having a long median contact between the two elements. 

Of the short front legs, the long bones of both sides are present 
in semi-articulated fashion close to the right scapulocoracoid of 
the holotype (Fig. 8). A humerus is present in 4118, and two 
incompletely preserved specimens are present in 4116. Except 
for the last, the bones are exposed from the \'entral surface. The 
structure is that typical of primiti\'e archosaurs generally — hour- 
glass shaped, moderately expanded at either end, and constricted 
at mid-length of the shaft. There is a well-developed deltopec- 
toral crest. Distally, there is a circular convex area for articula- 
tion with the head of the radius; lateral to this the bone is 
somewhat notched for reception of the olecranon. 

The right radius and ulna ( Fig. 8 ) are present and articulated 
with the humerus in the holotype; the left radius and ulna are 
incomplete distally. I have not been able to identify these ele- 
ments in other specimens. As preserved, they show little char- 
acter; both are slender elongate cylinders. The ulna is some- 
what expanded proximally, but there is no olecranon ossification. 

Regrettabh there are no identifiable remains of the manus 
preserved. 

Femora (Fig. 9a) are present in the holotype, nos. 4116, 
4118, and the Tucuman specimen mentioned earlier; in the 4116 
slab there are four femora of appropriate size and shape, indicat- 
ing the presence of two indi\iduals of Gracilisuchus. The bone 
has the typical sigmoid curvature of a proper archosaur. The 
head is turned somewhat medially from the shaft, but this curva- 
ture is less pronounced and the distinction between head and 
shaft less marked than in Lagosuchus, for example. There is no 
evidence of the presence of a "fourth trochanter." Tibia and 
fibula are present in articulation with the femur on the right 
side of the holotype, and these elements are present also in nos. 
4116, 4118, and the Tucuman specimen (Fig. 9b and c). These 
elements are long and slender Ijut where associated are never- 
theless somewhat shorter than the femur. As always, the tibia is 



1972 



GRACILISUCHUS STIPANICICORUM 



21 






d 



Figure 9. a, b, c, Left femur, tibia, and fibula, d. Left pes, composite. 
All X I. 



much .stronger than the fibula, with an expanded triangular 
head, a well-developed cnemial crest, and a broadh o\al distal 
area for astragalar articulation. The fibula is moderately broad 
but flattened throughout, with a gently sigmoid cur\ature. 

The pes (Fig. 9d) is incompletely known. In the foot per- 
taining to the right leg of the holot\pe the usual two small distal 
tarsals are present. But little is preser\ed of the two proximal 
tarsals, and nothing worthy of description can be made out from 
tarsal remains associated with tibia and fibula in MCZ 4118. 
Astragalus and calcaneum are, howe\'er, moderately well pre- 
served in the Tucuman skeleton mentioned earlier. The astraga- 
lus is a stout element, broad mediolaterally and moderately deep 
proximodistally; the anterior surface is somewhat conca\-e. At 
its lateral margin the astragalus, as articulated, is in contact with 
the fibula, and below this area a diagonal surface of contact 
with the calcaneum is present. The calcaneum is less developed 
proximodistally. Details of the astragalocalcanear articulation 
cannot be determined; the calcaneum, howex'er, had a well-de- 
\eloped "'heel"" in the fashion of crocodilians and manv pseudo- 
suchians. 



22 BREVIORA No. 389 

Data on the digits of the pes are available only in the holo- 
type and the Tucuman skeleton; the foot of the former is of the 
right side, of the latter, the left. Neither is complete distally. In 
the holotype metatarsals I-IV are present, but metatarsal IV is 
incomplete. Metatarsal IV is notably more slender than I-III. 
Metatarsal lengths as preserved are: 12 +, 23, 28 and 24 + 
mm. In the Tucuman specimen all five metatarsals are present, 
with lengths of 24, 28, 33, 32 and 18 mm; metatarsal V is of the 
"hooked" type, pointed distally. On the holotype no phalanges 
are present on toe I, but on digit II all three phalanges are pres- 
ent, with lengths of 7, 6, and 8 mm. On toe III a single phalanx 
is imperfectly preserved ; no digits are present with metatarsal IV. 
In the Tucuman specimen, the two phalanges of digit I are 
present with lengths of 6 and 8 mm. On digit II all three 
phalanges are present, the third incomplete, the first two with 
lengths of 7 and 4 mm. On digit III three phalanges are present, 
but only the first, with a length of 10 mm, is well preserved. 
With digit IV there is found only a single phalanx, 7 mm long. 

Allowing for a disparity in size of about 20 percent, the data 
from the two specimens agree well and allow a complete con- 
struction of the foot except for the more distal phalanges of 
toes III and IV. Assuming that as regards these phalanges, the 
pattern is that found in other advanced thecodonts and primitive 
saurischians, the restoration shown in Figure 9d cannot be far 
from the actual condition. It may be noted that in the specimens 
with articulated foot material, the toes are closely appressed to 
one another, as in my figure, with no trace of the fan-shaped 
spreading seen in many reptiles. 

RESTORATION 

Between the various available specimens, nearly the entire 
skeletal structure of Gracilisuchus is identifiable, and hence a 
skeletal restoration is justified ( Plate 1 ) . As usual in archosaurs 
in which the front limbs are notably shorter than the hind, the 
question arises as to whether a bipedal or quadrupedal pose is 
suitable. In the case of Chanaresuchus (Romer, 1972) I re- 
stored the animal as a quadruped, despite considerable disparity 
in limb lengths, because of the probably amphibious, crocodile- 
like nature of the animal. Gracilisuchus, as already mentioned 
(and discussed below), is quite surely a relative of Ornithosuchus, 
and I have followed Walker's restoration of that reptile in re- 
storing Gracilisuchus as a biped. It is possible that the normal 



1972 GRACII.ISUCHUS STIPAXICICORUM 23 

pose of GrncUisuchus was a qiiadrupechil one; however, I feel 
sure that, if pressed, this animal was able to run in the l)i])ccl:il 
manner in \vhich I have restored it. 

RELATIONSHIPS 

It is quite clear, I think, that Gracilisuchus is a relati\-e of 
Ornilhosuchus of the later Elgin beds of Scotland, ably described 
by \Valker in 1964. The skull structure is closely comparable in 
most resrards, as are \arious postcranial features. Walker has 
suggested certain late Triassic saurischian genera as possible Orni- 
thosuchus relatives which might belong with this genus in a 
common family Ornithosuchidae, and Bonaparte (1969b) has 
recognized two Argentinian genera (Venalicosuchus and Rio- 
jdsuchus ) that seem quite sureh to belong in this family. Gracili- 
suchus, from the Chafiares beds — quite probably Anisian in 
age — is the oldest (and smallest) of forms that may pertain to 
this apparently common and perhaps widespread Triassic family. 
In a fe\v features (such as the partial closure of the lateral tem- 
poral fenestra) the genus Gracilisuchus is perhaps slightly aber- 
rant, l:!ut it seems quite surely close to the base of this stock. 

\\'alker argues further that Ornithosuchus is a carnosaur — a 
proper dinosaur rather than a thecodont ancestor of dinosaurs. 
I provisionally adopted this interpretation when I published my 
1966 edition of Vertebrate Paleontology. I confess, however, to 
now having reservations on this assignment (cf. Bonaparte, 
1969a). Certainly the ornithosuchids show a number of features 
that might be expected in a carnosaur ancestor. But in certain 
features Gracilisuchus surely is below a proper dinosaur ''grade" 
in structure: the apparent complete closure of the acetabulum, 
for example. Again, saurischians are completely devoid of armor, 
and I would be loath to believe that the dorsal armor seen in 
Gracilisuchus and Ornithosuchus would ha\e been developed 
and secondarily lost. I do not wish to enter the controversy over 
the evolution of tarsal structure, but the presence in ornitho- 
suchians of a "crocodilian"' type of tarsus is, to say the least, an 
argument against placing the family in the Carnosauria, although 
not necessarily debarring the group from an ancestral position. 
Certainly the ornithosuchids show a trend in de\elopment that 
is in manv wavs similar to that which led to the carnosaurs. But 
for the time, it is, I think, better to regard them as forms related 
to and paralleling the line leading to the carnosaurs rather than 
members of that group. 



24 



BREVIORA 



No. 389 



As our knowledge of thecodonts increases (as it is currently 
doing at a rapid rate), it seems clear that while crocodilians, 
pterosaurs, bird ancestors and ornithischians have struck off on 
a variety of "tangents," the saurischians show merely an im- 
provement on structural patterns already evident among the 
thecodonts. It has been generallv held that the Saurischia are 
a "natural"' order, monophyletic in origin. But recently it has 
been advocated (Charig et al., 1965) that the sauropods are 
only distantly related to the "theropods" and may have arisen 
independently from ancestral thecodonts, and it is not impossible 
that among "theropods," carnosaurs and coelurosaurs may ha\e 
had independent origins (cf. Bonaparte, 1969a). I am consti- 
tutionally allergic to unnecessary ad\'ocacy of polyphyletism, but 
I fear that we are as yet far from a solution to questions of re- 
lationships between various thecodont and saurischian groups. 



TABLE I 
Lensth of limb bones, in nun 





Holotype 


MCZ 4118 


MCZ 4116 


Hiuncius 


38 


44 




44 


Radius 


30 


— 




— 


Ulna 


31 


— 




— 


Femur 


58,60 


68 




64,62,61,60 


Tibia 


56 


59, 


64 


61,56 


Fibula 


55 


64 




47 + 



Tucuman specimen 



81 
73 
65 



REFERENCES 

Bonaparte, J. F. 1969a. Comments on early saurischians. Zool. J. Linn. 

Soc, 48: 471-4S0. 
1969b. Dos nuevos "faunas" de reptiles triasicos de Ar- 
gentina. Gondwana Stratigraphy, lUGS Symposium 1967: 283-306. 
Charig, A. J., Attridce, J., and Crompton, A. W. 1965. On the origin of 

the sauropods and the classification of the Saurischia Proc. Linn. Soc. 

London, 176: 197-220. 
RoMER, A. S. 1966. Vertebrate Paleontology. Chicago, 111.: Univ. of Chicago 

Press. 
1971. The Chanares (Argentina) Triassic reptile fauna. 

X. Two new but incompletely known long-limbed pseudosuchians. 

Breviora, No. 378: 1-10. 
1972. The Chanares (Argentina) Triassic reptile fauna. 



XIL The postcranial skeleton of the thecodont Clianarcsuchus. Breviora, 
No. 385: 1-21. 
Walker, A. D. 1964. Triassic reptiles from the Elgin area: Ornithosuchus 

and the origin of carnosaurs. Philos. Trans. R. Soc. London (B), 248: 
53-134. 



LIHRARY 

MAR 1 8 1985 

HARVARD 

BIREVIORA 

Rliuiseom of Coixiparatave Zoology 

Cambridge. Mass. August 11, 1972 Number 390 

THE GHANARES (ARGENTINA) TRIASSIG 

REPTILE FAUNA. XIV. LEWISUCHUS ADMIXTUS, 

GEN. ET SP. NOV., A FURTHER THECODONT 

FROM THE GHANARES BEDS 

Alfred Sherwood Romer 

Abstract. Incomplete remains of a new thecodont from the Chanares 
Formation. Leicisuchus admixtus. are described and figured. Incomplete 
skull remains indicate that the posterior border of the lateral temporal 
opening was nearly vertical; the basal articulation with the palate was 
freely movable. A maxilla indicates elongation of the snout. Much of the 
presacral column is present; the cervical vertebrae are somewhat elongate, 
the ribs unspecialized. Scapulocoracoids are preserved, but no pelvic mate- 
rial; limb material is incomplete and disarticulated; femur and tibia are 
slender and of approximately equal length; the pes is long and slender. A 
single row of thin dorsal scales is present. Lewisuchus is a relatively 
primitive pseudosuchian that may be related to coelurosaur ancestry. 

INTRODUCTION 

After much preparation and a general survey of the Chanares 
collection, it became apparent that six thecodonts of \arious sorts 
were present in the material; these ha\-e been described in pre- 
\ious papers in this series. Recently, ho\ve\er, Mr. Arnold Lewis, 
in preparing a concretion containing a mixed assortment of 
reptilian remains, discovered that in addition to parts of a gom- 
phodont skeleton and miscellaneous materials of small thecodonts, 
there was present a considerable fraction of a skeleton and skull 
of a thecodont of relati\ely good size which was obviously new, 
and is described below. 

I am indebted to the National Science Foundation grant GB- 
2454 for aid in the collecting of the specimen and grant GB- 
22658 for preparation and publication. 



BREVIORA 



No. 390 




Figure 1. Lateral view of the skull, restored, an, angular; d, dentary; 
;', jugal; m, maxilla; po, postorbital; q, quadrate; san, surangular; sq, squa- 
mosal. The jugal-quadratojugal suture is obscure. X 2/3. 



Lewisuchus adtnixttis, gen. et sp. nov. 

Holotype. Museo de La Plata 1964-XI-14-14, consisting of 
much of the presacral column, part of the skull and jaws, scapu- 
lororacoids and some limb material, contained in a concretion 
including also remains of a gomphodont and smaller thecodonts. 
The specimen was collected from the Ghanares Formation of 
La Rioja Province, Argentina, about 4 km north of the mouth 
of the Chaiiares River. 

The generic name is in honor of Chief Preparator .\rnold D. 
Lewis, who discovered the remains during preparation of the 
nodule containing them. The specific name refers to the mixture 
of materials in the nodule. 



DESCRIPTION 

Skull and jaws (Figs. 1-5 ) . In one portion of the nodule were 
found the "cheek" region of the skull in articulation ^vith the 
back end of the mandible, and close by, the occipital plate and 
basicranium, as well as a limited amount of other skull material. 
Separately were found an incomplete maxilla with dentition, an 
appropriate dentary, and a small segment of another maxilla 
and dentary. 

Of the skull remains, the region of the left lateral temporal 
opening is well preserved ( Fig. 1 ) . The fenestra is subquadrate 
in outline, relatively narrow anteroposteriorly, the posterior mar- 
gin descending \ertically to the region of the jaw articulation. 
The jugal forms most of the straight lower margin of the skull 
in this area, the portion preser\'ed running back from the area 



1972 LEWISUCHUS ADMIXTUS O 

below the orbit to a contact with the quadratojugal and includ- 
ing a triangular process ascending between orbit and lateral 
temporal opening to a diagonal articulation with the postorbital. 
The hitter bone extends down nearly to the posteroventral angle 
of the orbit, narrowing distally; this \'entral poition of the bone 
has a vertical groove at mid-width. At the level of the top of 
the lateral fenestra the postorbital is thickened, with a pro- 
nounced transverse ridge on its outer surface. The bone is broken 
off shortly abo\-e this point. 

The quadratojugal completes posteriorly the ventrolateral skull 
margin. Posteriorly, adjacent to the quadrate, it expands in 
triangular fashion and sends a slender process, cur\dng some- 
what forward, up along the anterior margin of the quadrate. 
This process, as preser\-ed, terminates about half-way up the 
posterior margin of the lateral fenestra. 

The quadrate is nearly complete. Below, it presents a broad 
area for jaw articulation ; this area, howe\'er, is not ^vell preser\-ed 
and the surface was perhaps cartilaginous in life. Its external 
ramus sweeps far upward behind the lateral fenestra, with a 
mildly con\-ex anterior border, and %vith a conca\-e cross section. 
At the posterior margin of this ramus there is, as in thecodonts 
generally, a sharp ridge, internal to which is the ramus articulat- 
ing ^vith the pterygoid, little de\eloped dorsally but of consider- 
able extent further \entrally. Lying above the head of the 
quadrate is a triangular piece of bone, quite surely broken off 
dorsally, with its apex directed downward along the anterior 
edge of the quadrate, its posterior margin following the cur\ed 
upper edge of the quadrate. Its cur\ed anterior margin appar- 
ently formed the posterodorsal angle of tlie lateral fenestra. This 
is surely a fragment of the squamosal. It is probal)le that in life 
squamosal and quadratojugal were in contact along the anterior 
margin of the quadrate. Abox'e the pre«er\ed portions of the 
postorbital and squamosal are se\'eral bone fragments of indeter- 
minate nature which ma\" ha\'e pertained to the missing skull 
roof. 

Internal to the temporal area described above, but not articu- 
lated with it, is a nearly complete liraincase (Figs. 2—4) mostly 
in a good state of preservation. No sutures are apparent. The 
occipital plate is complete. Above is a broad supraoccipital area, 
with a median \ertical ridge dorsally. On either side the upper 
margins cur\'e for^■vard into the otic res^ion of the braincase: 
below this, on either side, paroccipital processes, broadened at 
their tips, run outward and somewhat downward and posteriori) . 



4 



BREVIORA 



No. 390 




aiz^ 



Figure 2. Posterior view of the skull; roof restored, pt, pterygoid; q, 
quadrate; qj, quadratojugal; sq, squamosal; tbsph, basispheuoidal tubera. 
X 1. 




Figure 3. Ventral view of posterior part of skull; quadrate and dermal 
bones shown on right side of figure; ventral view of braincase on left. 
bptp, basipterygoid process of basisphenoid; /, jugal; popr, paroccipital 
process; ps, cultriform process of parasphenoid; pt, pterygoid; q, quadrate; 
qj, quadratojugal; st, stapes; XII, openings for hypoglossal nerve. X 1. 



bptp^ 



popr 




{ ] Y \ yn.jx.x.fo 



tbsph 



Figure 4. Lateral view of braincase. hptp. basipterygoid process; fo, 
foramen ovalis; popr, paroccipital process, cut olT at base; ps, cultriform 
process of parasphenoid; tbsph, basisphenoidal tubera. Roman numerals 
indicate presumed region of nerve exits. X 1. 



1972 



LEWISUCHUS ADMIXTUS 



Below the large foramen magnum is the occipital condyle. Its 
surface is broad and somewhat subdivided posteriorly, and faces 
as much ventrally as posteriorly, suggesting a head posture ap- 
propriate to a possible bipedal pose. The basicranial region is 
well preserved. Just anterior to the condyle are highly de\eloped 
basisphenoidal tubera. Broad posteriorly, they diminish in size 
anteriorly as they con\erge, with a deep longitudinal median 
grooxe between them. Anterior to the rea^ion of the bases of 
the tubera are highly developed basipterygoid processes, which 
extend strongly downward, outward, and somewhat anteriorly. 
It is obvious that at their curved articular termini there was free 
mo\-ement between the processes and the pterygoids. Extending 
forward bet\veen the base of the basipterygoid processes is the 
slender cultriform process of the parasphenoid, incomplete an- 
teriorly. 

The upper margins of the braincase stop at a point where 
presumably the roof of the brain cavity was continued by the 
dermal roofing bones. Laterally (Fig. 4), the upper margins 
curve forward on either side. The upper part of the lateral brain- 
case walls slants medially; below this, a prominent if rounded 
ridge runs forward from the anterior surface of the paroccipital 
process, separating from the upper part of the wall a lateral 
surface ^vhich lies in a \'ertical plane and ventrally turns some- 
what medially. 

Three bars or "struts" constitute the lateral braincase wall 
connecting the upper portion of the braincase with the basi- 
cranial region. The most posterior, adjacent to the foramen 
magnum, is formed by the exoccipital, presumably reinforced 
anteriorly by opisthotic ossification. Two foramina are present 
in this strut. At least one was obxiously occupied by ner\e XII; 
possibly the other served for passage of the \agus ner\'e, but the 
opening is small and I tend to belie^•e that this was a second 
hypoglossal foramen, and that the vagus emerged anterior to 
this strut. 

Between the posterior and middle struts there is a large un- 
ossified area in the side wall of the otic region; presumably in 
this area lav the exits of nerve VII, the \agus foramen and the 
fenestra ovalis. The upper portion of the inter\al between pos- 
terior and middle struts is ossified, the ossification lying deep to 
the struts concerned. The middle strut is presumably formed 
by a prootic ossification, with ner\e V emerging anteriorlv be- 
neath it. The most anterior strut, presumably a laterosphenoid 
ossification, descends to meet the basisphenoid at the base of 



6 BREVIORA No. 390 

the basipterygoid process. Dorsally the presumed laterosphenoid 
bifurcates, leaving an opening (? a fenestra epioptica) between 
its branches and the dorsal taenia marginalis. No more anterior 
ossification is preserved in the braincase. The dorsal surface of 
the cultriform process is grooved, presumably for reception of a 
cartilaginous sphenethmoidal braincase segment. 

Much of the left pterygoid is preserved, although it is some- 
what displaced. The quadrate ramus is present. Its ventral 
border is strongly ridged, continuing a ridge present more pos- 
teriorly on the quadrate. Anteriorly this ridge curves upward, 
separating a thinner posterior portion of the ramus from a thicker 
anterior area; presumably this marks the boundary of the tym- 
panic cavity. The area for articulation with the basipterygoid 
process is a recess on the medial border of the bone at the junc- 
tion of quadrate and palatal rami. The posterior portion of the 
palatal ramus is preserved; there is no evidence of palatal teeth. 
A well-developed lateral flange appears to have extended di- 
rectly laterally, with a slight ventral curvature distally. A ridged 
medial border of the ramus is preserved for a distance. From 
the tip of the lateral flange the lateral, ridged, margin of the 
palatal plate curves forward and somewhat medially and dor- 
sally; for the short distance preserved, the lateral border twists 
sharply outward and gains contact with the jugal. It is probable 
that this region pertains to an ectopterygoid, and there is some 
evidence of a line of suture between this element and the ptery- 
goid. 

On the right side a slender splint of bone, about 7 mm long, 
extends outward from the braincase region anterior to the paroc- 
cipital process. This is reasonably identified as a stapes. 

No more anterior portions of the skull are present in articula- 
tion ^^•ith, or near, the posterior skull elements so far described. 
Probably pertaining to this skull are tooth-bearing elements that 
are certainly thecodont in nature and of appropriate size. A 
tooth-bearing strip of bone, 73 mm in length, is surely an incom- 
plete right maxilla ( Fig. 1 ) . The teeth are incompletely pre- 
served, but some 18 teeth or alveoli can be counted. A maximum 
size appears to be de\eloped not far from the anterior end, pos- 
terior to which there is a gradual diminution in tooth size. Ex- 
ternally, near the anterior end there is a curved ridge that 
presumably marks the border of the depression containing the 
antorbital fenestra. Part of the upper margin of the maxilla 
appears to be a finished surface bounding the fenestra. The in- 
ternal surface of the bone is considerably swollen in the area 
in which the bases of the larger teeth were contained ; above this 



1972 LEWISUCHUS ADMIXTUS 



Figure 5. Inner view of lower jaw, restored, (iti, angular; d, dentary; 
san, surangular; sp, splenial. X 2/3. 



is a thin area that is obviously part of the maxillary extension 
upward anterior to the antorbital fenestra. It may be noted that 
the tooth row is essentially straight, without the ventral con- 
vexity noticeable in carnosaurs and e\en in ornithosuchids. 

A fragment that pertains to a left maxilla shows a series of 
teeth clearly decreasing in size posteriorly. As in thecodonts 
generally, the teeth are somewhat compressed mediolaterally, 
conical, sharp-pointed, and recur^•ed posteriorly toward their tips. 

The posterior portion of the left mandible is present in essen- 
tially natural relations with the cheek elements and quadrate 
(Figs. 1, 5). The anterior end of the preserved portion lies in 
the region of the external mandibular foramen, showing alcove 
this opening the surangular and, below^ the angular and poste- 
rior end of the splenial. Internally the jaw portion preserved 
shows the suture outline of the mandibular fossa. It is probable 
that in life the outer surface of the ramus was tilted strongly 
medially, but, even so, it would appear that the fossa faced as 
much medially as dorsally. Abo^•e the fossa the jaw, as pre- 
served, appears to show a broad horizontal shelf along the course 
of the surangular. The articular surface of the mandible is large, 
broadened lateromedially, and divided into a smaller anterior 
and larger posterior portion; the articulation is so oriented that 
the inner margin is somewhat more anteriorly placed. There is 
a well-developed retroarticular process and, in addition, a strong 
flange directed N'entromedially behind the articular area. 

Presumably belonging to this specimen is an isolated tooth- 
bearing element, obviously a dentary, 62 mm in length, which 
bears about 20 teeth or alveoli ( Figs. 1,5). As frequently in 
thecodont jaws, there is no great regional difference in tooth 
size along the series. It is probable, from the contours of the 
bone, that little is absent anteriorly. Two small foramina for 
blood \essels are present close to the front end ; internally a longi- 
tudinal meckelian groove is present, above which the bone is 
thickened for tooth roots. 



8 BREVIORA No. 390 

In Figure 1 I ha\e attempted to restore the skull in lateral 
view. Because of the relative length of the maxilla, it is ob\'ious 
that the "snout" was much longer, relatively, than in many 
thecodonts. 

Axial skeleton. Not connected with the cranial remains first 
described, but reasonably associated because of thecodont na- 
ture and proper size, is a series of 17 articulated xertebrae, be- 
ginning with the axis; part of this series is shown in Figure 6. 
Anterior to the axis are imperfect remains that appear to repre- 
sent the atlas centrum and axis intercentrum (no intercentra are 
present more posteriorly). The axis is well developed, with a 
relatively low but long neural spine, with a curved upper mar- 
gin. The cervical vertebrae are elongate as compared with the 
rest of the column; the axis centrum is 14 mm in length, and 
the more posterior cervicals are approximately similar, compared 
to an a\erage of 1 1 mm in dorsal members of the series. Neural 
spines are not well preserved in the cervical region (the axis 
apart). In vertebra 8 the spine is low, extending but 9 mm 
above the level of the zygapophyses. The back border of the 
spine is essentially vertical, but the anterior border slants strongly 
forward dorsally, so that, from a width of 5 mm across the base, 
the dorsal margin is 10 mm in extent. There is no thickening 
of the dorsal margin for armor support (such as is seen in 
Gracilisuchus) . In the dorsal region the spine bases are stouter, 
but the anterior slant of the anterior margin persists. There are 
some poorly preserved traces of scutes above the cervical verte- 
brae, apparently thin and probably in a single row. 

As preserved, the cervical centra appear to be thin, compressed 
from side to side except for prominent Ncrtical ridges at either 
end. Even here, howe\er, there appears to be no sharp ventral 
longitudinal ridge, and as we proceed posteriorly the centra be- 
come thicker and but s^entlv rounded ventrallv. In the cervical 
region the sides of the centra show a longitudinal depression; 
farther back, with major de\"elopment of the transxerse processes, 
this depression is part of a more expanded excavation, bounded 
above by the roots of the transverse processes. The articular 
area for the rib capitulum lies, in the cervical region, at the base 
of the ridge forming the anterior margin of the centrum; it is, 
however, little marked in the column as preserx-ed. Presumably 
this articular area ascended toward or to the arch in the posterior 
part of the series but there is little evidence preserved in the 
specimen. In the column as preserved the diapophxsis is not 




Figure 6. Right lateral vie^v of cen'ical and anterior dorsal vertebrae 
and ribs, h, head of right humerus and incomplete left humerus; sccor, 
inner view of right scapulocoracoid. X 1. 



prominently developed in the cenical \ertebrae; in the dorsal 
region, howexer, the trans\erse processes become prominent, ex- 
tendina: strona:lv outward and somewhat downward and back- 
ward. The process is supported ventrally by ridges extending 
upward to its base from both anterior and posterior margins of 
the centrum ; above, a stout ridge connects the base of the process 
with the anterior zygapophysis, and a less dexeloped ridge ex- 
tends to the region of the postzygapophysis. 

Close to the posterior end of the articulated series are three 
further \ertebrae of appropriate size for Lewisuchus. Details are 
not well preser\ed, l^ut presumably these xertebrae were from 



10 BREVIORA No. 390 

the "lumbar" or sacral region. Several further isolated vertebrae 
are present in the concretion, not well preserved except for one 
which is clearly a posterior dorsal. 

Ribs are in general incompletely preserved (Fig. 6). Close to 
the base of the axis centrum is a short rodlike structure, expanded 
at one end, which is presumably an incomplete axial rib; ad- 
jacent to it is a slender rod, 14 mm long, which may be an 
atlantal rib. There are no ribs preserved associated with cervicals 
3-6; ribs are present, although incomplete, with vertebrae 7-10. 
All are markedly two-headed ; there is no evidence of the develop- 
ment of accessory processes of the rib heads described in a num- 
ber of other thecodonts. The preser\ed portions of these ribs, 
measured from the tubercula, are 20, 15, 45, and 37 mm long. 
Although none of the four is complete, the first two appear to 
be slender and close to their termini where broken off; the last 
two are more stoutly built and seem surely to have been true 
dorsal ribs. A few further incomplete rib segments are present 
close to the articulated rib series. In another part of the nodule 
are several articulated dorsal ribs of a size appropriate for Lewi- 
suchus; one has a length of about 80 mm. The curvature of 
these ribs suggests a deep but narrow trunk. Close to the three 
vertebrae mentioned above as possible lumbars or sacrals are 
two structures with triangular outlines and with a dexeloped 
articular area at the narrow end; these may be sacral ribs. 

Appendicular elements. Close to, and quite surely pertaining 
to the articulated vertebral series are two scapulocoracoids (Fig. 
7 ) , the left seen from the outer surface, the right from the inner 
side. Scapula and coracoid are well fused, with no apparent 
suture. The scapular blade is unusually tall and slender. Dis- 
tally it expands somewhat; the distal margin is taller posteriorly, 
slanting downward toward the anterior margin. The lower part 
of the scapula expands both anteriorly and posteriorly, with a 
thickened acromial ridge anteriorly and a somewhat comparable 
posterior ridge buttressing the upper glenoid rim. The \entral 
margins of the coracoids are imperfectly preserxed, but the bone 
seems to be primitixe in structure, with no evidence of a crocodi- 
loid posteroventral extension, and without evidence of any ante- 
rior "incision" in the plate. 

There is no material interpretable as being a clavicle or inter- 
cla\icle of this specimen. And nothing found in this concretion 
can be identified as pertaining to the pelvic girdle of Lewisuchus. 

There is a considerable amount of limb material strewn 
through the concretion, but most appears referable to a gompho- 



1972 



LEWISUCHUS ADMIXTUS 



11 




Figure 7. Left, left scapulocoracoid; right, incomplete left humerus in 
ventral view. X 1. 



dont or to one or more thecodonts of smaller size. Only a small 
amount of material appears to be of a size and nature appropri- 
ate to Lewisuchus. 

Close to the right scapulocoracoid is the head of a humerus 
ob\iously of slender build and a second imperfect humerus Fig. 
7) ; its head seems comparable to that just mentioned. The shaft 
is slender; the distal end is missing, but if extrapolated from 
the part present, on the analogy of Hesperosuchus (Colbert, 
1952, fig. 22), the length in life must have been on the order 
of 70 to 75 mm. I find no elements that are interpretable as 
radius or ulna. A femur of appropriate size (Fig. 8) measures 
105 mm in length. The bone is badly crushed proximally; it 
ne\'ertheless shows a well-developed greater trochanter and an 
apparently spherical head turned in sharply from the shaft. The 
bone is slender, the shaft having a diameter of but 7 mm. Un- 
fortunately the distal end is imperfect. 

A slender tibia, represented mainly by an impression in the 
matrix, is 106 mm in length. Near the shoulder region (but also 
close to the vertebrae mentioned as perhaps being in the lumbo- 
sacral region) is a group of podial elements (Fig. 8B). Two 
slender bones, measuring 31 and 30 mm in length, are surely 
metapodials; semi-articulated with the second are three pha- 
langes, the terminal one clawed. Near the first of the two meta- 
podials is a relati\ely long phalanx, and beyond it a second series 
of three articulated phalanges, terminating in a clawed element. 



12 



BREVIORA 



No. 390 





Figure 8. A. Right femur in dorsal view (incomplete distally) ; B, ele- 
ments of second and third toes of right pes; C, the same, articulated. X 1. 



It is most reasonable to interpret the two metapodials and the 
seemingly associated phalanges as toes II and III of the right 
pes. When articulated (Fig. 8C), it is obvious that digit II is 
stronger than III, and nearly as long, suggesting comparison 
with the proterochampsid type of foot (Romer, 1972a). 



DISCUSSION 

Because of the inadequacy of the material, I have refrained 
from attempting a skeletal restoration of Lewisuchus. In default 
of good skull material, allocation of Lewisuchus to a definite 
position in the order Thecodontia is difficult. The modest trend 
toward strengthening of the inner toes immediately suggests com- 
parison with Chanaresuchus (Romer, 1971; 1972a) and the 
proterochampsids, an assignment with which facial elongation is 
compatible. But strengthening of the inner toes is not confined 
to the proterochampsids, and the subquadrate configuration of 
the lateral temporal opening is sufficient to debar Lewisuchus 
from the Proterochampsidae, and indeed, from the suborder 
Proterosuchia, in which the lateral opening is relati\ely long 



1972 LEVVISUCHUS ADMIXTUS 13 

with the quadrate slanting backward ventrally. Lewisuchus is 
thus probably assignable to the Pseudosuchia. 

Lewisuchus retains certain primiti\e features, such as the free 
basal articulation between palate and braincase, and the nearly 
straight back margin of the lateral temporal fenestra, which lacks 
the V-shape here found in many advanced forms. On the other 
hand, the long slender tibia, about equal to the femur in length, 
and the long metapodials, suggest a strong advance towards a 
truly bipedal gait. 

Lewisuchus does not appear to be closely comparable to other 
described pseudosuchians. It differs in many obvious features 
from the ornithosuchids (cf. Romer, 1972b). It is possibly re- 
lated to LIcsperosuchus of Colbert ( 1952 ), but this form is poorly 
known and, furthermore, appears to difTer in certain features, 
such as the incipient de\'elopment of accessory rib flanges, absent 
in Lewisuchus. Elongation of cervical vertebrae suggests com- 
parison with Teleocrater, an incompletely known form from the 
Alanda beds, under description by Charig (1957). Lewisuchus 
may possibly be a form leading toward the coelurosaurs. Not 
improbably Lewisuchus may exentually merit being made the 
type of a family of its own. But for the time being it is perhaps 
best left as a pseudosuchian in cert ac sedis. 

LITERATURE CITED 

Charig, A. J. 1957. New Triassic archosaurs from Tanganyika including 
Mandasuchus and Teleocrater. Abstr. Diss. Univ. Cambridge, 1955-56: 
28-29. 

Colbert, E. H. 1952. A pseudosuchian reptile from Arizona. Bull. Amer. 
Mus. Nat. Hist., 99: 566-592. 

Romer, A. S. 1971. The Chanares (Argentina) Triassic reptile fauna. 

XI. Two new long-snouted thecodonts, Chanayesuchus and Gualosui hus. 
Breviora, No. 379: 1-22. 

1972a. The Chanares (.Argentina) Triassic reptile fauna. 

XII. The postcranial skeleton of the thecodont C/ianaresuc/ius. Breviora, 
No. 385: 1-21. 

1972b. The Chanares (Argentina) Triassic reptile fauna. 

XIII. An early ornithosuchid pseudosuchian, GracUisuehits st'ipanici- 
corum, gen. et sp. nov. Breviora, No. 389: 1-24. 



ML'S. CO MP. ZOOL 

LinnARY 



B R E V I O R A 

I^Iuseom of Comparative Zoology 

Cambridge, Mass. August 11, 1972 Number 391 

THE RELATIONSHIP OF ISLAND AREA AND 

ISOLATION TO COLOR 

POLYMORPHISM IN LIGUUS FASCIATUS 

(PULMONATA, BULIMULIDAE) 

Michael A. Rex 



Abstract. Liyuus fasciatus is a highly polymorphic arboreal pulnionate 
living on small islands called hammocks in the Florida Everglades. Color 
variation in this snail has in other studies been assumed to be random and 
of no selective significance. Multiple regression analysis showed that 
hammock area was a statistically significant predictor of color morph di- 
versity, and that measures of isolation were either insignificant or significant 
but subordinate to the effect of area. From this result it was inferred that 
color variation in Liguus was regulated by available ecological diversity 
(hammock area) even when interhammock migration appeared to be ex- 
tensive. Color polymorphism in this snail might be imposed by visual 
predation, larger hammocks with more heterogeneous floral composition 
having more morphs because they offer more possibilities for cryptic or 
aposematic associations to avoid predators. 

INTRODUCTION 

Conchologists have long considered color \ariation in the re- 
markably polymorphic Liguus fasciatus to be random and of no 
adaptive significance. "On the whole it is not apparent that 
any form of selection has been a major factor in the exolution 
of Florida Liguus" (Pilsbry 1946: 47). Ecological uniformity 
of habitats led Pilsbry ''1912, 1946) to attribute differences in 
color and form between monomorphic colonies to isolation 
alone, and to explain variation in mixed colonies by random 
mutation and recombination of h\pothetical Mendelian alterna- 
tives. 

However, recent studies indicate that random genetic e\ents 
do not, as once thought, constitute the most important mode of 
evolution in land snails. Crampton's (1916, 1925, 1932) con- 
tention that intraspecific variation in Partula was due to genetic 



2 BREVIORA No. 391 

drift has been contested by selectionists both on the basis of his 
collecting methodology (Mayr 1942; Cain and Sheppard 1950; 
Ford 1964) and on reexamination of his own data (Bailey 
1956). Gulick (1873, 1905) was unable to correlate variation in 
Achatinella with ecological diversity, but Welch (1938, 1942, 
1958) found changes in both color and form with akitude in 
achatinellids when temperature and moisture gradients were 
pronounced. Diver (1940) and Lamotte (1952) proposed that 
color variation in Cepaea nemoralis resulted from genetic drift, 
though it is now evident that relative frequencies of color and 
banding patterns in Cepaea are regulated to some extent by 
visual predation (review in Cain and Sheppard, 1954) and cli- 
matic factors (Lamotte 1959, 1966; Wolda 1967; Arnold 1968, 
1969). Furthermore, Murray (1964) demonstrated that 
hermaphroditism, sperm storage, and multiple matings in C. 
nemoralis function to maximize effective population size, thereby 
reducing the probability of genetic drift in small colonies. Gould 
(1969) disclosed the adaptive significance of both color and form 
in his study of temporal character variation in Poecilozonites. 

In the present paper I show that the number of color morphs 
in populations of Liguus jasciatus living on small islands can be 
predicted by island size and isolation. It is suggested that poly- 
morphism in this snail is maintained by a selective balance of 
phenotypes. 

Liguus is particularly suitable for investigating efTects of 
environmental factors on phenotypic diversity. The snail li%^es 
on small islands called hammocks in the Florida Everglades. 
Hammocks are slightly elevated remnants of Pleistocene coral 
reefs that support a tropical hardwood vegetation. They are 
surrounded by sparse pine woods or swamp, both of which are 
inhospitable to Liguus. Thus, habitat size and distance between 
individual populations are well defined and easily measured. 

LIGUUS MATERIAL 

Liguus colonies, once abundant in Florida, have now been 
decimated to near extinction by land clearing, glade fires, ama- 
teur collectors, and introduced rats. Fortunately, a few careful 
investigations were carried out before the onslaught. Material 
used in this study came from extensive collections made by W. 
J. Clench and W. S. Schevill during February and March, 
1931, from an assemblage of hammocks on Long Pine Key, 
Dade County, Florida ( Fig. 1 ) . The hammocks ranged in size 



1972 



COLOR POLYMORPHIS^r 




^ c 



>;^ 

^ o 

^ ± 

n 
M > 
c n 
o 



u 

o 
P 

E 

n 



C 
o 



3 



S T3 
^ u 

E 
.2: J= 



c 






3 



4 BREVIORA No. 391 

from 0.34 to 43.80 acres. According to Clench (personal com- 
munication), an endeavor was made to collect samples of all the 
varieties present in each hammock \isited, but not necessarily 
in proportion to their natural frequency distribution (non- 
random ) . The material, comprising several thousand shells from 
48 hammocks, was sorted by Clench as to varieties and placed 
in the mollusk collections of the Museum of Comparative 
Zoology, Harvard University. 

I have considered the Long Pine Key material to include nine 
color morphs, which are described briefly below. The morphs 
clearly have some genetic integrity. By examining several clutches 
of eggs, Pilsbry ( 1 946 ) showed that separate patterns segregated 
out, but the exact extent to which this occurs is unknown. 
Pilsbry (1946) classified the morphs of L. jasciatus into sub- 
species, forms, and varieties. The subspecific designations repre- 
sented different color patterns and had no geographic connota- 
tion. The forms were based on the colors present in the patterns 
and the presence or absence of pink coloration on the early 
whorls and columellar callus. The varieties were usually named 
stages in the continuous variation present in the forms. His sys- 
tem for sorting out morphs is conx'cnient and I have largely 
adhered to it here. But the genetic criteria on which his hierarchy 
was based are unacceptable. For instance, the rare variety 
deckerti (see below) found in three widely separate localities in 
Florida was thought to have arisen from the form castaneo- 
zonatus through mutational "loss" of the factor for pink colora- 
tion (1946: 70). Its white apex and columellar callus could, 
however, be the result of a number of conceivable schemes 
involving matings with morphs having white apices, dominance 
interactions, recombination, and close linkage. The genetic 
mechanisms governing coloration in these snails are probably 
complex and I have chosen not to make any underlying assump- 
tions about them. For purposes of this paper it is sufficient to 
recognize that L. jasciatus is a highly variable species in which 
more or less disjunct color morphs can be discerned and that 
the number of morphs present in a population is some indica- 
tion of its genetic variance. 

MORPHS PRESENT IN LONG PINE KEY 

Names listed are those used by Pilsbry ( 1 946 ) and the pat- 
terns refer to Figure 2. 
1. ebiirneus: shell completely white. 



1972 



COLOR POLYMORPHISM 




Figure 2. Basic color patterns for Litjuus fasciatus present on Long Pine 
Key, Florida (from Piishry, 19+6). See text for explanation. 



2. 



cingulatus : 



faint yellow spiral zones (Pattern A) on a white 
background; apex and columellar callus white. 

3. roseatus: yellow spiral zones (Pattern A) on a white back- 
ground; apex and columellar callus pink. 

4. castaneozonatus: splotchy to solid brown spiral zones (Pat- 
tern A) on a white background; apex and columellar callus 
pink. 

5. deckerti: brown spiral zones (Pattern A) ; apex and colu- 
mellar callus white. 

6. luteus: uniform faint yellow coloration (Pattern B) ; apex 
and columellar callus white. 

7. ornntus: uniform yellow coloration (Pattern B) intensifying 
to amber near the lip; apex and columellar callus pink. 

8. testudineus: variegated and banded (Pattern C or D), with 
brown coloration on white to yellowish background; apex 
pink with pink or \vhite columellar callus. 

9. marmoratus: \ariegated and banded (Pattern C or D) 
with brown coloration on a white background; apex and 
columellar callus white. 

Patterns C and D are endpoints of continuous variation, the 
dense variegation or "flames" of C gradually superimposing the 
bands and lightly variegated pattern in D. There appear to be 
no good objective criteria for discriminating various stages in 
this continuous variation. Thus, the distinction between morphs 
9 and 10 is one of apical coloration. 



METHODS 

Computerized stepwise multiple regression analysis using pro- 
gram BMD02R (Dixon, 1968) was performed to determine 
whether the degree of polymorphism in populations from dif- 



6 BREVIORA No. 391 

ferent hammocks correlated with any of several independent 
variables measuring area or isolation. 

No general discussion of the estimating equations of multiple 
regression is presented here as they ha\e now been so extensively 
employed in ecological studies (cf. Ebeling et al., 1970; Hamil- 
ton et al., 1963; Hamilton et al., 1964; Hamilton and Rubinoff 
1963, 1964, 1967; Johnson et al, 1968; Pomeroy, 1967; Vuil- 
leumier, 1970). 

The methodologv allows one to determine both whether a 
statistically significant prediction of the dependent variable can 
be made using some or all of the independent variables, and 
which independent variables are most important. The square 
of the multiple correlation coefficient R" estimates the amount 
of variance in the dependent variable explained by the combined 
effects of the independent variables. The significance of the en- 
tire regression was tested by the variance ratio F, which is the 
mean square due to regression di\'ided by residual mean square. 
The significance of the contribution of each independent vari- 
able to the total explained variance in the dependent variable 
(R") was tested by using the t- ratio, which equals the partial 
regression coefficient divided by the standard error (Snedecor 
and Cochran, 1967). Three versions of the regression were car- 
ried out: a linear model, a semi-log model using log-trans- 
formed independent variables, and a log-log model. 

The dependent variable was simply the number of morphs 
present on each hammock. Independent variables were restricted 
to those which could be measured from a large map of the 
collecting site prepared by W. S. Schevill (Fig. 1). Previous 
studies in island biogeography, in which multiple regression 
models were used to predict species diversity (cf. esp. Hamilton 
and Rubinoff 1963, 1964, 1967; Vuilleumier 1970), suggested 
several independent variables. The following were tested for 
their ability to predict the number of morphs: (1) Hammock 
Area. This variable is commonly used as an index of ecological 
diversity. Larger hammocks were expected to have more micro- 
habitats; i.e., more plant species, more possibilities for cryptic 
associations, greater foliage height diversity, etc. (2) Distance 
to Hammock 27. Hammock 27 (Fig. 1) was the largest island 
and may have been an effectixe dispersal center for the entire 
group of hammocks even though it was peripherally located. It 
also had the fortuitous advantage of being close to Paradise Key, 
an enormous hammock which probably exerted influence over 



1972 COLOR POLYMORPHISM 7 

the morph diversity of Long Pine Key hammocks, hut which 
was not included in this study because data on the number of 
color morphs present were unavailable. This variable tested 
efTecti\eness of isolation from a possible source area of high 
phenotypic variability. Hammock 27 had eight of the nine color 
morphs. Three additional variables determined the effects of 
isolation in a more localized sense. (3) Distance to the nearest 
large hammock, a large hammock being defined as any of 
hammocks 54, 23, 26, 8, 34, 27. (4) Distance to the nearest 
hammock, (5) Size of the nearest hammock. 

RESULTS 

The relative influence of each independent variable and an 
anahsis of variance are given in Table 1. In all three models 
hammock area was the most significant (P < .001) predictor 
of the number of morphs present. Area made a higher contri- 
bution to R~ in the semi-log model (.39) than in the linear 
model (.30), an indication that the data are somewhat curvi- 
linear with respect to area. 

Variables measuring isolation, with the exception of distance 
to hammock 27, proved to be insignificant. Distance to ham- 
mock 27 made a significant contribution to R~ in the linear (P 
< .05) and semi-log (P < .025) models. This suggested that 
a slight phenotypic diversity gradient extended westward from 
hammock 27. If hammock 27 was, or was near, the effective 
population center, then perhaps the population was simply ex- 
panding westward. There was some evidence to support this 
contention in the distribution of individual morphs. Two 
morphs, ornatus and marmoratus, did not extend farther west 
than hammocks 34 and 56 respectively and deckerti was con- 
fined to hammocks 55 and 26. They were the least frequent in 
occurrence and were found predominantly on large hammocks. 
These morphs might still have been in the process of dispersing 
westward. When thev were removed from the analysis, the 
number of morphs still correlated significantly with hammock 
size (.01 ^ P ^ .001). But distance measures correlated so 
poorlv when added to the multiple regression models that the 
significance of the entire regression was reduced to P < .05 
in the linear and log-log \ersions. Otherwise, the morphs showed 
no evident propensity for east or west. Even though distance 
to hammock 27 made a significant contribution to R", its effect 
was clearly subordinate to that of area. 



8 



BREVIORA 



No. 391 



c 

.2 

re 

3 
<u 
C . 

.1= ^ 

c« Q 

^ e 
^ C 
M c 

<LI 

.fg 

E.S 

> 



> ■^ 
a 

O.I 

a 

C fn 

rt C 

O 

■a 'Z 



.2 c 

CU 

"i 

-a c 
c 

B. O 






1-1 

< 



^ 3 
-g c - 

•5 t o 
tL c ■" 



a. 



o 5 4- 



3 

Co 



« 



* 



o 



« h o 
^ c *^ 
« o 



> 

c 

-a 
c 
u 
D. 

C 

















o 


o 


— , 


o 


O 




'"^ 


o 


""J 


m 


\r\ 


v/1 




O 


V 


A 


A 


A 


A 




V 

CI, 


en 


o 


o- 


u-i 


wn 




, " 


r^ 


u-1 


so 


O 


.— < 




^ 


o 


•* 


in 


CM 


•+ 






>o 


en 


CNJ 


m 


>+ 




•^^ 


en 












II 


u-i 


r^ 


l-H 


1— » 


"*■ 


00 




o 


ON 


.-H 


^.^ 


en 


-t 


* 


00 


CM 


o 


O 


o 


rg 


-^ 


M 


o 


O' 


■^ 


o 


en 


tn 


1— t 


\^ 












o 


(N 


o 


o 


o 




1-H 




O 


VTN 


u-^ 


vn 




O 


V 


V 


A 


A 


A 




V 


-+ 


-I- 


CM 


'I- 


r*-i 




rsf 


en 


oo 


oc 


^H 


u-i 




rj 


C7\ 




t^ 


CN) 


O 




~' 


\o 


lJ-1 


en 


SO 


U-l 




C<0 


•t 


CNJ 










II 


VO 


tn 


t^ 


so 


1-H 


ly^ 


CI 


t^ 


CM 


.— q 


en 


CO 


OO 




oo 


■ ; ON 


O 


O 


o 


oo 


\o 


en 


p 


o 


O 


o 


•+ 


u.'^ 
















o 


\r\ 


c 


o 


o 






o 


O 


CM 


vo 


u^ 




c 


V 


V 


A 


A 


A 




V 


m 


•+• 


On 


oo 


-t- 




rj 


o 


CM 


CM 


y-* 


o 




r^ 


t~^ 


ON 


t^ 


CM 


so 




en 


U-) 


CM 




u-i 


CSJ 




U~t 


-h 


CM 










1 


en 


CM 


SO 


O 


o 


.— 




1— ( 


SO 


so 


u-i 


^— ( 


C' 


lO* 


O 


so 




O 


o 


ON 


en 


o 


o 


o 


o 


en 


fc"' 






^ 














o 





























p 








t-l 














O 






c 


.1^ 


^ 




<-i-i 






r3 


o 


o 












o 


o 




!> 


C3 


CM 




E 

£ 


E 








o w 














4-1 

aj cr. 


4J 




> 3 


O 


u E 


u u 


O oj 


V4_ K 




O '^ 


g 


si 

4- « 


C 1- 


C 1- 


o !- 






£ 


c8 eg 


03 tT! 

4- <u 




(is 


•7.? 


5 


.2 J= 

Q 


Q 




N C 


ra 




>i 




ft 


X 




o 


C 
< 





J>^' 




X 


4-1 


CM 


c 


M-t 


Qj 


O CM 


OJ 


— o 


u. 


o o 


Cl 


U-1 


u 


oo 1 



X 



u 


, W) 


yj 


L> O 


-Q 




3 
C/5 


bCCM 




•'—CNJ 


4^ 


X <>J ^ 

C: so o 


1) 




so '■^ 


c 


CM d 1 


"* 


O 1 =1 


T3 


1 X 


X 


+ X ^o" 



o 

X 






X bC 


-5 






r^ — CM 

CM -^^ 


X 






-- ■=> o 


O 






O CM O 


XJ 






'■^ 


E 






1 o-f 


>-i 








X 






c-f- .. 


u 






X ^^x 


a; 






so X f'^ 








(N.I ^ 


o 






— btoo 


— 






°^d 


< 






-J- '~° , 
r^ 1 


«J 






■' o 


c 






en X 

so 1 


«4M 








E 






^X" 


o 






1 ^'^ 


u 






MCM 


M-i 






O "^ 
-—CM 


^ 






X vo d 


3 

X 






iM oo-r 


u 






eo 


X' 






't " -H 


OJ 






-I- "^ 
+ OO 


C3 






" d 


> 






VJ3 -t- ,, 


-a 






s;:^!! 


C3 






n""^ 


"^ 






_o 






IL b£> 


r! 






^ _o 


u 




.. 


>" __- 






^1 


OJ ^- 


n 

4.1 


E 
o 


1-1 
« 

X 

C 


SEi 


>, 


"O 


_o 


- M 


."ti 


0; 


•w 


. O M 


IS 




C3 

3 


C E M 


.D 


"J.^ 


u 


.— qj O 


O 

0, 


O 


s 


k4 C/5 hJ 


X 


.2 






1) 


'x 





II M t 



bfi 



1972 COLOR POLYMORPHISM 9 

It is worth noting that in three cases independent variables 
measuring isolation had significant simple correlation coefficients 
with the number of morphs : 1 ) distance to the nearest large 
hammock, r =^ -.407, P < .01; 2) distance to hammock 27, 
r = —.398, P < .01, both in the semi-log model; and 3) dis- 
tance to the nearest large hammock, r = -.310, P < .05 in 
the log-log model. These isolation variables were de-emphasized 
in the multiple regression analyses because they did not correlate 
with the dependent \ariable as significantly as area (r = .549, 
.538, .623, P < .001, in the linear, log-log, and semilog \'ersions 
respectively ) , did not explain as much of the variance in the 
dependent xariable as area, and were partially redundant with 
one another. 

Associations between intraspecific variability, interspecific 
competition, faunal dixersity, etc. and various physical or bio- 
logical envaronmental parameters are often obscure. The use- 
fulness of multiple regression lies in the ability to discern a 
hierarchy of importance among independent variables relating 
to some dependent variable from a matrix of correlations be- 
tween all \'ariables. In this particular study it indicated that 
hammock area was more important than isolation in predicting 
the amount of color variation in populations of L. fasciatiis. 
It is possible, however, that the distance measurements and 
transformations used may not have been the relevant ones for 
predictive purposes. 

MacArthur and Wilson (1967: 132-133) concluded on the 
basis of theoretical considerations that in colonization "stepping 
stones" should be less important to passive dispersers than to 
acti\'e dispersers. Snails probably disperse passix^ely and one 
would expect direct distance measurements to be most appro- 
priate in this kind of study. When hammock-hopping distances^ 
were tried in place of variables 2 and 3, however, the results 
remained \'ery nearly the same as those presented in Table 1 . 
R" for the linear, semi-log, and log-log model was .39, .51 and 
.34 respecti\'ely, and probability levels for significant variables 
were identical, with the exception of \'ariable 2 in the semi-log 
model which was significant at P < .005 instead of P < .025 

'Hammock-hopping was measured by using intervening hammocks as 
stepping stones rather than using direct distance. Also, measurements 
were made through the pine wood or to the narrowest passage through 
glade if glade must he crossed, the assumption being that it would be 
difficult for an actively dispersing pulmonate to cross bodies of water. 



10 BREVIORA No. 391 

(a difference of less than .02 in the contribution of variable 
2 to R-). 

Distinction in mode of dispersal that might otherwise be indi- 
cated by using either direct or hammock-hopping distances 
could be blurred by the small size of the study area as a whole. 
Predatory birds and high winds are likely agents of dispersal 
and they probably transport snails a highly variable range of 
distances. Simpson (1929) occasionally found live Liguus far 
into the open pine wood which he thought were propagules on 
their way to colonize new hammocks. He gives a delightfully 
anthropocentric account of Liguus "obeying an instinct for 
founding new colonies." It seems more likely that the snails 
he found were dropped by birds. 

An analysis of variance for the entire regression for each 
model is given in Table 1 . All three models provided significant 
(P < .01) results. The best predictive value (highest R~) was 
afforded by the semi-log model. In all models, however, at least 
half of the variance in the number of morphs was left un- 
accounted for. This "error" (E = 1 -R") was due to several 
factors. First, there were undoubtedly important variables that 
were not considered simply because the data were unavailable. 
For instance, more precise measurements of ecological diversity 
such as the number and kind of plant species occurring on each 
hammock may have been important. Secondly, there were errors 
in measuring the independent variables used in the appraisal of 
polymorphism. Some morphs must have been occasionally over- 
looked during collecting. Finally, there is the certainty that not 
all of the variance could ever be attributed to ecological para- 
meters, part of it being due to genetic mechanisms such as 
balanced polymorphism ( heterozygote superiority) and part of 
it actually being random. 

DISCUSSION 

T infer from the inability of isolation to predict the number 
of color morphs and the widespread distribution of the morphs 
in general that interhammock migration, however it occurs, is 
a fairly frequent event, but that its contribution to maintaining 
polymorphism in Liguus is strongly mediated by hammock area. 
The question now arises: Why is hammock area a good pre- 
dictor of color variation? A speculative hypothesis is the 
following. 



1972 COLOR POLYMORPHISM 11 

Boettger (1931) first suggested, without giving supporting 
data, that the degree of color polymorphism in Cepaea is reg- 
ulated by background heterogeneity through selection by visual 
predation. The quantitative studies of Cain and Sheppard see 
re\ie\v 1954) corroborated Boettger completely. More diverse 
backgrounds such a-s hedgerows, rough herbage, and mixed 
deciduous woods supported more variable populations of 
Cepaea than the more uniform beechwoods or short turf. Fur- 
ther, cryptic associations between color morphs and background 
color appeared to confer a selective advantage in eluding pred- 
ators. For instance, in a study of Cepaea in \Vytham Woods, 
Sheppard ( 1 95 1 ) found that brown and pink morphs were at 
an advantage in mid-April when the forest floor was brow^n, 
the more conspicuous yellow morphs being selectively predated 
by thrushes. In May when the forest became green the rexerse 
situation obtained, yellow then being at an ad\antage. 

A similar mechanism might regulate morph diversity in 
Liguus. Larger hammocks with greater foliage height diversity 
and a greater number of plant species might provide more pos- 
sibilities for cryptic associations to avoid predators. Natural 
predators of Liguus include the opossum (Pilsbry, 1946), the 
crow, and "other large birds" (Simpson, 1929). Colonies on 
small hammocks having limited floral heterogeneity might con- 
tinually have morph di\ersity depleted by predators. 

Eisner and \Vilson (1970) recently suggested the obverse 
of this hypothesis: that coloration in Liguus is aposematic 
I rather than cryptic ) . ^Vhen disturbed the snail withdraws into 
its shell and discharges large quantities of liquid. Simpson 
(1929) thought this fluid secretion enabled the snail to slip 
from the beaks of predatory birds, but Eisner and Wilson be- 
lieved it to be a chemical defense mechanism. The success of 
cryptic or aposematic coloration depends on suitable back- 
ground color. Since none of the wide variety of morphs will 
appear conspicuous in all situations, perhaps the snail employs 
a mixed strategy involving both kinds of coloration. Further 
observations are needed. 

The possibility of low morph di\ersity on small hammocks 
being due to genetic drift is not discounted, but seems unlikely 
for reasons presented against Pilsbry's argument below. Other 
hypotheses, such as regarding polymorphism as a manifestation 
of niche subdivision (Van Valen 1965), are concei\able, but 
lack substantiating evidence. 



12 " BREVIORA No. 391 

An explanation for the pre\iously noted cunilinearity of the 
data with respect to area might be that the number of cryptic 
associations utilized by this snail is attained at intermediate ham- 
mock size, and the probability of adding a new and different 
association with further increase in area is negligible. 

Pilsbr\'s argument that \'ariation in Liouus is random seems 
untenable for the following reason: Interhammock migration, 
as already mentioned, appears to be quite extensive. Since the 
snail is hermaphroditic and has multiple matings a single fer- 
tilized immigrant can introduce considerable \ariation into a 
colony (Pilsbry, 1912, found at least three separate morphs 
in one clutch of nine eggs from a single adult ) . In the early 
stages of colonization, larger hammocks might be expected to 
ha\'e populations with greater morph diversity simply because 
they represent larger target areas for propagules; this might be 
the case with ornatus, ?narmoratus, and deckerti. The great 
abundance of Liguus at Long Pine Key and further north in 
Florida suggests, however, that the snail has populated southcn 
Florida for a rather long period of time, possibly since the last 
glaciation, and that it has essentially achieved a steady state. 
This being the case, and if variation occurred randomly and 
were unaffected by selection, then there would be no reason to 
expect populations living on larger hammocks to be more vari- 
able than those li\ing on smaller ones. The highly significant 
positi\e correlation of hammock area and the number of morphs 
(P < .001) indicates, however, that morph diversity on smaller 
hammocks is maintained at a low level. Visual predation might 
be an important selective agent reducing variation on small 
hammocks. Visual predators are known for Liguus and a similar 
mechanism has been shown to operate with another polymorphic 
pulmonate. 

ACKNOWLEDGMENTS 

I am grateful to S. J. Gould, K. J. Boss, T. W. Schoener, 
A. Schoener, and E. O. \Vilson for critically reading the manu- 
script. A. Clark, R. Bullock, R. D. Turner, R. I. Johnson, 
W. J. Clench, and R. A. Howard offered helpful suggestions or 
read parts of the manuscript. AV. Bossert and F, Vuilleumier 
pro\'ided advice on statistics. C. Jones drafted Figure 1. This 
research received partial support from NSF Grant GB- 19922 
to Harvard University (R. C. Rollins, principal investigator) 
and GZ-1615 to Harvard University. 



1972 COLOR POLYMORPHISM 13 

LITERATURE CITED 

Arnold, R. \V. 196S. Climatic selection in Crpaia nrmoralis (L.) in 
the Pyrenees. Phil. ']"rans Roy. Soc. London, B, 253: 549-593 

. 1969. The effects of selection by climate on the land- 
snail Ccpaca nemoral'is (L.). Evolution, 23: 370-378. 

Baii.ey, D. \V. 1956. Reexamination of the diversity in Pariula tacn'iata. 
Evolution, 10: 360-366. 

BoETTGER, C. R. 1931. Die Entstehung von Populationen mit bestimmter 
Variantenzahl bei der Landschneckengattung Crpaca Held. Zeit. 
ind. Abst. Ver., 58: 295-316. 

Cain, A. J., and P. M. Sheppard. 1950. Selection in the polymorphic 
land snail Ccpaca nemoral'is. Heredity, 4: 275-294. 

, and 1954. Natural selection in Ccpaca. 

Genetics, 39: 89-116. 

Crampton, H. E. 1916. Studies on the variation, distribution and evo- 
lution of the genus Pariula. The species inhabiting Tahiti. Cam. 
Inst. Washington Publ., 228: 1-311. 

1925. Studies on the variation, distribution and evo- 
lution of the genus Pariula. The species of the Mariana Islands, 
Guam and Saipan. Carn. Inst. Washington Publ., 228a: 1-116. 
. 1932. Studies on the variation, distribution and evo- 



lutiiin of the genus Pariula. The species inhabiting Moorea. Carn. 

Inst. Washington Publ., 410: 1-335. 
Diver, C. 1940. The problem of closely related species living in the 

same area, p. 303-328. In J. S. Huxley [ed.]. The New Systematics. 

Oxford: Clarendon Press, 583 pp. 
Dixon, W. J. [ed.]. 1968. Biomedical computer programs. University of 

California Publications in Automatic Computation, No. 2. Berkeley: 

Univ. California Press. 600 pp. 
Ebeling, a. W., R. M. Ibara, R. J. Lavenberg, and F. J. Rohlf. 1970. 

Ecological groups of deep-sea animals off southern California. Bull. 

Los Angeles Co. Mus. Nat. Hist, 6: 1-43. 
Eisner, T., and E. O. Wilson. 1970. Defensive liquid discharge in 

Florida tree snails (L'ujuus fasciaius). Nautilus, 84: 14-15. 
Ford, E. B. 1964. Ecological Genetics. London: Methuen. 335 pp. 
Gould, S. J. 1969. An evolutionary microcosm: Pleistocene and Recent 

history of the land snail P. (Poccilozoniics) in Bermuda. Bull. Mus. 

Comp. Zool, 138: 407-532. 
GuliCK, J. T. 1873. On diversity of evolution under one set of ex- 
ternal conditions. J. Linn. Soc. London, 11: 496-505. 
. 1905. Evolution, racial and habitudinal. Cam. Inst. 

Washington Publ., 25: 1-269. 
Hamilton, T. H., R. H. Barth, Jr., and I. Rubinoff. 1964. The en- 
vironmental control of insular variation in bird species numbers. 

Proc. U. S. Nat. Acad. Sci., 52: 132-140. 



14 BREVIORA No. 391 
, AND I. RUBINOFF. 1963. Isolation, endemism, and mul- 



tiplication of species in Darwin finches. Evolution, 17: 388-403. 

AND 1964. On models predicting abund- 



ance of species and endemics for the Darwin finches in the Gala- 
pagos Archipelago. Evolution, 18: 339-342. 
, AND 1967. On predicting insular vari- 



ation in endemism and sympatry for the Darwin finches in the Gala- 
pagos Archipelago. Amer. Natur., 101: 161-171. 
, , R. H. Barth, Jr., and G. Bush. 1963. 



Species abundance: natural regulation of insular variation. Science, 
142: 1575-1577. 

Johnson, M. P., L. G. Mason, and P. H. Raven. 1968. Ecological param- 
eters and plant species diversity. Amer. Natur., 102: 297-306. 

Lamotte, M. 1952. Le role des fluctuations fortuites dans la diversite 
des populations naturelles de Ccpaca ncmoralts (L.). Heredity, 6: 
333-343. 

1959. Polymorphism of natural populations of Cepaea 

nemoralis. Cold Spring Harbor Symp. Quant. Biol., 24: 65-86. 

1966. Les facteurs de la diversite du polymorphisme dans 



les populations naturelles de Cepaea nemoralis (L.) Lavori della 
Societa Malacologica Italiana, 3: 33-73. 

MacArthur, R. H., and E. O. Wilson. 1967. The theory of island bio- 
geography. Monographs in Population Biology No. 1. Princeton, N. J.: 
Princeton Univ. Press. 203 pp. 

Mayr, E. 1942. Systematics and the Origin of Species. New York: 
Columbia Univ. Press. 334 pp. 

Murray, J. 1964. Multiple mating and effective population size in 
Cepaea nemoralis. Evolution, 18: 283-291. 

PiLSBRY, H. a. 1912. A study of variation and zoogeography of Liguus 
in Florida. J. Acad. Nat. Sci. Philadelphia, 15: 429-471. 

1946. Land Mollusca of North America (north of Mex- 
ico). Acad. Nat. Sci. Philadelphia Monograph No. 3. Vol. 2, pt. 1 
pp. 1-520. 

POMEROY, D. E. 1967. The influence of environment on two species of 
iand-snaijs in South Australia. Trans. Roy. Soc. S. Australia, 91: 
181-186. 

Sheppard, p. M. 1951. Fluctuations in the selective value of certain 
phenotypes in the polymorphic land snail Cepaea nemoralis (L.). 
Heredity, 5: 125-134. 

Simpson, C. T. 1929. The Florida tree snails of the genus Liguus. Proc. 
U. S. Nat. Mus., 73: 1-44. 

Snedecor, G. W., and W. G. Cochran. 1967. Statistical methods. Ames: 
Iowa State Univ. Press. 593 pp. 

Van Valen, L. 1965. Morphological variation and the width of ecologi- 
cal niche. Amer. Natur., 99: 377-390. 

VuiLLEUMiER, F. 1970. Insular biogeography in continental regions. I. 
The Northern Andes of South America. Amer. Natur., 104: 373-388. 



1972 COLOR POLYMORPHISM 15 

Welch, D. A. 1938. Distribution and variation of Achat'incUa mustelina 

Mighels in the Waianae Mountains, Oahu. Bull. Bernice P. Bishop 

Mus., 152: 1-164. 
. 1942. Distribution and variation of the Hawaiian tree 

snail Achaiinclla apfxjulva Dixon in the Koolau Range. Oahu. Smith. 

Misc. Coll., 103: 1-236. 
. 1958. Distribution and variation of the Hawaiian tree 



snail Achatlnclla hiilimoidcs Swainson on the windward slope of the 
Koolau Range, Oahu. Proc. Acad. Nat. Sci. Philadelphia. 110: 123- 
211. 
WoLDA, H. 1967. The effect of temperature on reproduction in some 
morphs of the land snail Cepaea ncmoralis (L.). Evolution, 21: 
117-129. 



Lir.nARY 
MAR 1 8 19B5 

nARVARD 
nRSlTY> 



B R E V I O K A 

IMuiseum of Comparative Zoology 

Cambridge. Mass. August 11, 1972 Number 392 



A FOSSIL PELOMEDUSID TURTLE FROM 
PUERTO RICO 

Roger Coxaxt Wood* 

Abstract. A fossil turtle shell, too incompletely preserved to merit a 
formal binomial taxonomic designation, from the Oligocene San Sebastian 
Formation of Puerto Rico is here described. It represents the first record 
of a truly endemic pelomedusid in the Caribbean region. The specimen 
appears to be more closely related to South American pelomedusids than 
to any of the North American members of the family, which are known 
only as fossils. Presumably the Puerto Rican pelomedusid was derived 
either from South American ancestors or from some progenitor that gave 
rise to both it and the South American forms. 

Fossil turtles from the Caribbean region are few and frag- 
mentary. Our knowledge of these has most recently been 
reviewed by Williams ( 1 950 ) and then further extended in a 
subsequent publication ^Villiams, 1952). In his earlier paper, 
Williams p. 8) briefly mentioned a specimen from Puerto 
Rico ". . . represented by numerous fragments of plastron, pel- 
vis, and of the buttress region of the carapace . . ." which had 
been donated to the American Museum of Natural History in 
1924 by Seiior. Narcisco Rabell Cabrero. \Villiams noted that 
"the exact locality and circumstances of the find are uncertain" 
and that it "was unfortunately so incomplete as to afford no 
generic characters." Nevertheless, an accompanying map (^\'il- 
liams, 1950, fig. 1), in which the specimen is identified simply 
as an emydine, indicates that it was found in the southeastern 
quadrant of the island. 

The fos<il turtle to which Williams alluded, however, is not 
an emydine nor w^as it collected in southeastern Puerto Rico. 
It is, instead, a pelomedusid from the northwestern part of the 
island. Apparently this specimen was once part of a larger col- 
lection of \ertebrate remains from the San Sebastian Forma- 

*Stockton State College, Pomona. X. J. 08240 



2 BREVIORA No. 392 

tion, which outcrops in the \icinity of the city of that same 

name\ 

To the best of my knowledge, none of this material has ever 

been described, nor, except for the specimen that forms the 

subject of this paper, can its present whereabouts be established. 

The San Sebastian Formation is generally considered to be 
of middle to late Oligocene age (Schuchert, 1935: 467-468; 
Weyl, 1966, table 15; Khudoley and MeyerhofT, 1971, fig. 27), 
although there is some possibility that at least the upper part 
of it may be early Miocene in age (Khudoley and Meyerhoff, 
1971: 145-146). This formation is part of a larger series of 
middle Tertiary limestones and shales that form an extensi\'e 
east-west trending belt more than 100 kilometers long across 
northwestern Puerto Rico. The sediments in this belt are gen- 
erally belie\ed to represent a widespread mid-Tertiary marine 
transgression over this sector of the island. Shale is the pre- 
dominant sediment within the San Sebastian Formation, 
although a wide variety of other lithofacies also occur — 
gravels, sands, lignitic clays, limestones, and chalks. Fossils 
have been found throughout this formation, which is in the 
neighborhood of 700 feet thick; marine in\ertebrates are 
abundant, as are a \ariety of fossil land plants belie\ed to 
represent a tropical flora typical of that now found bordering 
on lagoons and estuaries. From this evidence it would appear 
that the sediments of which the formation is comprised were 
deposited partly under brackish and partly under near-shore 
marine conditions. The few fragments of matrix still adheriniz; 
to the turtle shell here described indicate that the specimen was 
recovered from a limestone facies. 

The shell ( # 1836 in the collection of the department of verte- 
brate paleontology at the American Museum of Natural History ) 
is badly fragmented, and its partial reconstruction has only been 
possible as the result of much patient piecing together by a 
number of persons (Drs. Walter Auffenberg, Eugene Gaffney, 

' Correspondence (now on file in the department of vertebrate paleontology 
at the American Museum of Natural History) from Professor James F. 
Kemp, of the department of geology at Columbia University, to Walter 
Granger mentions the existence of a number of specimens in addition to 
the fossil turtle. These were said to include teeth, jaws, and vertebrae. 
According to the son of Seiior Rabell Cabrero. who still lives in San Se- 
bastian, the specimens that his father collected probably came from 
limestone exposures along the main road from San Sebastian to Lares 
(personal communication, Mr. and Mrs. Alan Patterson). 



1972 FOSSIL PELOMEDUSID TURTLE 3 

and the author) over a period of se\cral years. Much of the left 
side of the shell has been preser\ed so that, although it is 
far from complete ( Figs. 1 and 2 ) , man\ of its essential fea- 
tures can be determined. Save for the epi-and entoplastra, most 
of the left side of the plastron remains. Much less of the cara- 
pace, unfortunately, is known; only the posterior neurals as well 
as some of the left pleurals and peripherals still exist. In addition, 
the pehis is largeh intact. Moreo\er, cleanly broken surfaces on 
many of the isolated bone fragments that I ha\e not been suc- 
cessful in reassembling indicate that at the time of its discovery 
the shell was somewhat more complete than at present, but how 
much of it actually has been subsequently lost during the nearly 
fifty inter\ening years is conjectural. 

Originally the carapace must ha\e measured well over half a 
meter in length. In trans\erse section, it is low-arched, with a 
strong lateral carina at the bridge. Nearly all of the last two 
neurals and the left side of the one immediately preceding these 
ha\"e been preser\ed. They are arranged in an uninterrupted 
series. The last is pentagonal, while the preceding two are hex- 
agonal ("coffin shaped"). Their dorsal surfaces are unkeeled. 
The neurals become progressi\ely shorter posteriorly (5.95 cm, 
4.45 cm, and 2.95 cm in length from front to back). Although 
it is impossible to determine the exact widths of any of the three 
neurals, it is clear that the last was broader than long, the next- 
to-last was of approximately the same width as length, and the 
most anterior was somewhat longer than broad. These propor- 
tions are typical of most pelomedusids. Some of the scute sulci 
that were superimposed on the dorsal surface can still be de- 
tected ; these are shown in Figure 1 . The hour-glass configura- 
tion at the junction between the fourth and fifth \ertebrals is 
similarly de\eloped elsewhere among the pelomedusids onl\- in 
the African species Pclusios sinuatus. to the be>t of my knowledge. 
The external surface of the shell (plastron as well as carapace) 
is essentially smooth, without the concentric striations or vermic- 
ulated sculpture characteristic of some pelomedusids. Rough- 
ened areas, posterolateral to the last neural on the \-isceral 
surface, indicate where the pelvis was fused to the carapace. The 
axial buttresses appear to ha\-e been robust, but the extent to 
which the inguinal buttresses may ha\e been de\'eloped cannot 
be ascertained, owing to the incomplete preservation of the cara- 
pace. 

As Dreserved, the midline length of the plastron is 35.7 centi- 
meters. I estimate that its total length must ha\e been somewhat 
greater than 50 centimeters. The posterior plastral lobe was 



BREVIORA 



No. 392 




Figure 1. External view of what remains of the carapace of the Puerto 
Rican pelomedusid (AMNH 1836) ; the anterior end is toward the top of 
the page. The positions on the visceral surface of the left axial buttress as 
well as of the iliac scars of the pelvis are indicated by broken lines. Those 
scute sulci that can still be discerned are represented by parallel dotted 
lines. Missing portions of the shell are indicated by shaded areas. 



1972 



FOSSIL PELOMEDUSID TURTLE 




Figure 2. External view of what remains of the plastron of the Puerto 
Rican pelomedusid (AMNH 1S36); the anterior end is toward the top of 
the page. The positions of the pelvic scars on the visceral surface are in- 
dicated by broken lines. The scute sulci and missing portions of the shell 
are depicted as in Figure L 



6 BREVIORA No. 392 

evidently slightly wider than the anterior one; its width to the 
midline at the inguinal notch is 13.7 centimeters, whereas the 
corresponding width from the axial notch to the midline is 
approximately 12 centimeters. The sides of the posterior lobe 
are straight rather than curved and converge toward the rear. 
The anal notch is broad. The external surface of the posterior 
lobe is flat rather than depressed inward, which suggests (but 
does not conclusively demonstrate) that the specimen may have 
been a female. The bone forming the posterior lobe becomes 
gradually and uniformly thinner toward the rear. Pelvic scars 
are present on the visceral surface of the plastron. These are 
disposed in essentially the same manner as in any of the living 
pelomedusids. Only the left mesoplastron has been preserved; 
it is large and hexagonal in shape, but unlike most pelomedusids 
it is not more or less equidimensional. Instead, it is appreciably 
longer (9.9 cm) in its anteroposterior axis than in its tranverse 
one (8.1 cm). The pectoral-abdominal scute sulcus curves back- 
ward laterally to cross the front portion of the mesoplastron. 
Otherwise, there is nothing noteworthy about the remainder of 
the plastral scute sulci that can still be discerned. 

The pelvis is the best preserved part of the specimen. Never- 
theless, parts of the pubes and ischia are missing on both sides, 
although somewhat more remains of the right side than of the 
left ( Plate 1 ) . In its general appearance, the pelvis is similar to 
the structure typical of most living and fossil pelomedusids. 

There can be no question that the specimen here described 
represents a pelomedusid; the presence of mesoplastra together 
with a pelvis that is fused to the shell assures this determination. 
More precise placement of this Puerto Rican pelomedusid with- 
in the family, however, is difficult. In the New World, only two 
clearly definable pelomedusid genera — Bothremys (GafTney 
and Zangerl, 1968) and Taphrosphys (Hay, 1908) — have 
been reported outside the confines of South America', and these 
are known only as fossils. The Caribbean form does not belong 
to either of these genera. Bothremys differs from it in having 
mesoplastra that are broader than long (Schmidt, 1940: 7 and 
fig. 5; Zangerl, 1948: 38 and figs. 3, 15), as well as an anterior 
plastral lobe that is slightly broader than the posterior one at 

'Two other North American fossil genera, Amhiypeza and Naladochclys, 
have been described on the basis of extremely fragmentary material and 
referred to the Pelomedusidae (Hay, 1908: 122, 125), but I regard both 
of these as nomina vana (Wood, MS). 



1972 



FOSSIL PELOMEDUSID TURTLE 




S 

u 




o 




o. 






• 




01 




Ui 


< 


-4-1 




lU 


• • 


s 


■^ 


4^ 


X 


C 


3 


i> 


T3 


u 


u 




E 


4> 

> 


_o 


« 


u 




ex 


o 


c 

rs 


^ 


o 


hb 


cS 








o 






cs 


;- 




OJ 




3 


X 


CL, 


c 




V 


0^ 


X 








c 


•-^ 


1> 


o 


u 


X 


V 






> 


rt 




X 


u 
^ 




<+-( 




o 




V 


0) 


JO 




X 


'> 


,J_, 


^^ 


J= 


ea 


_M 


•5 


*u 


u 


4J 


E 
o 


H 


V 




C 


, 


es 


'-" 






U 


« 








(li 


• *• 




% 




u 



8 BREVIORA No. 392 

its maximum width (Zangerl, 1948: 38). Taphrosphys differs 
principally in the shape and position of the pelvic scars on the 
internal surface of the plastron (,Hay, 1908, fig. 118) in addition 
to having its anterior plastral lobe broader than the posterior 
one." 

Representatives of three pelomedusid genera are known in the 
South American paleontological record." Two of these — 
Apodichelys (Price, 1954a) and Roxochelys (Price, 1954b) — 
are known only as fossils from Brazil, while the third - — - Podoc- 
neniis — still has living species in South America as well as 
in Madagascar. Apodichelys is based on a single specimen of 
very small size, the internal mold of a shell. Its anterior plastral 
lobe was considerably broader than the posterior one, and for 
this reason it is unlikely to be related to the Caribbean form. The 
type and only described specimen of Roxochelys consists of a 
small portion from the front of a carapace and an anterior plas- 
tral lobe. Comparable parts of the Caribbean specimen ha\'e 
not been preserved, so that possible relationships between the 
Brazilian and Puerto Rican forms cannot yet be evaluated. For- 
tunately, several nearly complete shells of Roxochelys have 
recently been discovered, but until they are described meaning- 
ful comparisons cannot be undertaken. Wood and Gamero 
(1971) recently have summarized our knowledge of the South 
American fossil forms that have been referred to Podocnemis, 
while Siebenrock (1902) and Williams (1954) have reviewed 
the living species of the genus. Except for the proportions of its 
mesoplastron and the outlines of the last two vertebral scutes, 
the Caribbean specimen falls within the limits of morphological 
variability known to occur in Podocnemis. 

From the foregoing discussion, it seems reasonable to state 
that the Caribbean specimen is probably more closely related to 
South x'Xmerican pelomedusids than to those from North Ameri- 
ca. But difficulties arise in trying to determine the most 
appropriate way to express this relationship. The shape of the 

"This information is based on a recently discovered specimen, #18706 
in the collections of Princeton University, that is much more complete than 
any that were available to Hay. Dr. E. S. Gaflney has studied but not yet 
described this specimen, and I appreciate his courtesy in permitting me to 
use the information mentioned here. 

^Staesche (1929) has referred some fragments from the late Cretaceous 
or early Tertiary of Argentina to a fourth genus, Na'iadochelys, but such 
an attribution is highly doubtful in view of the fact that the type material 
from North America cannot be adequately characterized at present. 



1972 FOSSIL PELOMEDUSID TURTLE 9 

mesoplastron and the \ertebral scute pattern might be considered 
sufficiently distincti\e for the recognition of either a new species 
of Podocnemis or perhaps even of a new genus. Numerous fossil 
pelomedusid taxa have been named and described in the past 
on the basis of an equally small list of purportedly diagnostic 
features. In view of the fact that the shell is so incompletelv 
known, however, I prefer to refrain from giN'ing it any kind of 
formal taxonomic designation at either the generic or the specific 
level, for at least three possibilities exist: the shell may belong 
to either Podocneynis. Roxochelys, or a third previously unrecog- 
nized genus. At present I do not beliexe that currently a\ailable 
e\idence is adequate to discriminate among these choices. 
Consequently I will merely designate the Puerto Rican specimen 
as Pelomedusidae gen. et sp. indet. 

Outside of South America, only one other fossil pelomedusid 
of Tertiary age has previously been described in the New World. 
This is the specimen, consisting solely of the anterior half of the 
plastron, that Collins and Lynn (1936) originally described as 
Taphrosph\s iniocenica and that GafTney and Zangerl (1968) 
later (and correctly, I believe) referred to Bothremys. This 
specimen, now lost (GafTney and Zangerl, 1968: 209), is of 
Miocene age. Collins and Lynn (1936: 162) assumed that B. 
miocenica must have been a fresh water form because all living 
pelomedusids occur in fresh water habitats. They further as- 
sumed that their specimen had been carried out to sea before 
being covered by sediment. In \iew of the fact that many pelo- 
medusids in the past (including, I suspect, all other previously 
described North American fossil pelomedusids) were evidently 
adapted to marine conditions (Wood, MS), as well as that the 
bulk of fossils associated with B. yyiiocenica include strictly 
oceanic cheloniid turtles, cetaceans, and marine inxertebrates, it 
seems reasonable to postulate that this species was probably a 
marine form that was not transported any particularh great dis- 
tance before burial. 

Circumstantial evidence indicates that the Puerto Rican pelo- 
medusid w'as probably also a marine form. As already mentioned, 
the fauna that was reportedly discovered in association with it 
cannot now be located, thus eliminating a potentially important 
source of environmental information. Fortunately, however, one 
other fossil vertebrate previously has been described from the San 
Sebastian Formation and this species — the dugongid sirenian 
Caribosiren turneri (Reinhart, 1959) — may be helpful in 
determining the depositional environment in which the remains 



10 BREVIORA No. 392 

of the Puerto Rican pelomedusid were buried. Caribosiren was 
found in a limestone outcrop along the road between the towns 
of San Sebastian and Lares (Reinhart, 1959: 8); consequently, 
it seems probable that this species was recovered from the same 
general area as the locality where Senor Rabell Cabrero discov- 
ered his fossil turtle. The close proximity, if not coincidence, of 
the localities where these fossil vertebrates were found strongly 
suggests that they lived in the same aquatic habitat. The lime- 
stones from which Caribosiren was disinterred contained 
abundant foraminifera and molluscs and these indicate a shallow- 
water marine habitat (Reinhart, 1959: 16-17). Living 
dugongids, furthermore, are confined to marine waters (Walker, 
1968: 1332). Therefore, it seems reasonable to suppose thai 
Caribosiren was also fully adapted to a marine existence. All 
these considerations lead me to believe, by inference, that the 
Puerto Rican pelomedusid was in all probablity a truly marine 
turtle. 

Other pelomedusids, although not fossil ones, previously have 
been reported from the Caribbean region. Palacky (1897: 14) 
cursorily mentioned the occurrence of Podocnemis durneriliana 
on Guadeloupe without specifying any further details. The va- 
lidity of this record is dubious, however, for, as Barbour ( 1934: 
111) has noted: ". . . apparently material from a host of locali- 
ties around the Caribbean basin, at one time or another, was 
assembled at one of these French islands [Guadeloupe or Marti- 
nique] before being shipped to Paris, where the material now 
bears a locahty label indicating the point of shipment only." 
Barbour further quoted a publication dating from 1862 which 
stated that tortoises " '. . . are brought for sale to the islands 
from the coast (Venezuela) . . .' " This reference specifically 
singles out an area where P. durneriliana naturally occurs. Either 
one of these explanations could account for the reported presence 
of this species on Guadeloupe. In view of the fact that no one 
has subsequently found P. durneriliana on this island, it seems 
reasonable to regard Palacky's comment on its distribution as 
erroneous. 

A different pelomedusid genus does, however, actually occur 
on Guadeloupe. Pinchon (1967: 561) has recorded its presence 
there: "C'est une Tortue originaire d'Afrique tropicale, Pelusios 
subniger; elle fut certainement introduite par I'homme mais on 
ignore absolument quand et comment." According to Pinchon 
(personal communication), representatives of this species can 
be found only on the eastern, nonmountainous half of the island 



1972 FOSSIL PELOMEDUSID TURTLE 11 

( Grand Terre ) , particularly in ponds in the vicinity of the town 
of Moule, and even here they are not particularly abundant. 
x\lthough no examples of Pelusios from Guadeloupe have yet 
been carefully examined, there are no readily apparent differ- 
ences serving to distinguish this population from African ones of 
the same species (which is, it may be worth noting, widespread 
in sub-Saharan Africa). It is for this reason, in addition to the 
fact that Pelusios does not occur in the fossil record except in 
Africa, that it seems reasonable to suppose that the Guadeloupe 
population has been introduced to the island within historic 
times. 

Of the two reported occurrences of living pelomedusids in the 
Caribbean region, therefore, one is erroneous and the other 
represents what is almost certainly a relatively recently introduced 
population that has achieved limited success at best. Conse- 
quently, the Puerto Rican specimen is the first record of a truly 
endemic pelomedusid in the Caribbean. On the basis of morpho- 
logical similarities, furthermore, it seems likely that the Puerto 
Rican form was derived either from South American ancestors 
or from some progenitor that gave rise to both it and at least 
some of the South American forms. 

ACKNOWLEDGMENTS 

I am indebted to Dr. E. S. GafTney of the American Museum 
of Natural History for his kindness in permitting me to study the 
fossil turtle described in this paper. Miss Charlotte Holton, also 
of the American Museum, was of great assistance in my efforts 
to locate correspondence relating to the specimen. Shelter from 
the hazards of the New York subway system was amicably pro- 
vided by Mr. and Mrs. R. H. Wadsworth during several visits 
to the American Museum for research on the specimen here 
described. Dr. Frank Whitmore, vertebrate paleontologist for 
the United States Geological Survey at Washington, D. C, was 
most helpful in directing my attention to Reinhart's publication 
on Caribosiren. Mr. and Mrs. Alan Patterson kindly interviewed 
the son of Senor Rabell Cabrero on my behalf during their visit 
to San Sebastian in late December, 1971; the information they 
were able to uncox^er proxed to be of great interest. I am in 
addition much obliged to Pere R. Pinchon of Fort de France, 
Martinique, and Dr. P. C. H. Pritchard, of the University of 
Florida, for information concerning the occurrence of Pelusios 
subniger on Guadeloupe. Prof. E. E. \\'illiams of Harvard Uni- 



12 BREVIORA No. 392 

versity first prompted my interest in the Puerto Rican fossil 
turtle and then critically read the resulting manuscript. Miss 
Victoria Kohler and Mr. Don Crowe assisted in the preparation 
of the photographic plate. Finally, I would lilce to express my 
gratitude to the National Geographic Society for its generous 
support of my research on pelomedusid turtles. 

LITERATURE CITED 

Barbour, T. 1934. Observations on Antillean tortoises and terrapins. Co- 

peia, (3) : 111-113. 
Collins, R. L., and W. G. Lynn. 1936. Fossil turtles from Maryland. 

Proc. Amer. Phil. Soc, 76(2): 151-173. 
Gaffney, E. S., and R. Zangerl. 1968. A revision of the chelonian genus 
Bothremys (Pleurodira: Pelomedusidae). Fieldiana, Geology, 16(7): 
193-239. 

Hay, O. p. 1908. The fossil turtles of North America. Carnegie Inst., 
Publ. no. 75, 568 pp. 

Khudoley, K. M., and a. A. Meyerhoff. 1971. Paleogeography and geo- 
logical history of Greater Antilles. Geol. Soc. America, Memoir 129: 
1-199. 

Palacky, J. 1897. La distribution geographique des cheloniens. Bull. In- 
ternational, Acad, des Sciences de I'Empereur Frangois Joseph I (Ceska 
Akademie Cisare Frantiska Josefa I, Prague), 4: 9-14. 

PiNCHON, R. 1967. Quelques aspects de la nature aux Antilles. Fort de 
France, Martinique. Printed for private circulation by the author. 
254 pp. 

Price, L. I. 1954a. Um quelonio pleurodiro no calcario da serie Apodi, 
Cretaceo do estado do Rio Grande do Norte. Notas Preliminares e 
Estudias, Divisao de Geologia e Mineralogia, nr. 85: 1-12. 

. 1954b. Os quelonios da formagao Bauru, Cretaceo terrestre 

do Brasil meridional. Divisao de Geologia e Mineralogia (Brasil), 
Bol. 147: 1-34. 

Reinhart, R. H. 1959. A review of the Sirenia and Desmostylia. Univ. 
Calif. Publ. Geol. Sciences, 36(1): 1-146. 

Schmidt, K. P. 1940. A new turtle of the genus Podocnemis from the 
Cretaceous of Alabama. Geol. Ser., Field Mus. Nat. Hist., 8(1): 1-12. 

Schuchert, C. 1935. Historical Geology of the Antillean-Caribbean Region. 
New York: John Wiley and Son, Inc. xxvi and 811 pp. 

Siebenrock, F. 1902. Zur Systematik der Schildkroten-Gattung Podocnemis 
Wagl. Sitz. Kaiserl. Akad. Wiss. Wien (Math.-Naturw. Classe), 
111(1): 1-14. 

Staesche, K. 1929. Schildkrotenreste aus der Oberen Kreide Patagoniens. 
Palaeontographica, 72: 103-123. 

Walker, E. P. 1968. Mammals of the World (2nd edition), vol. 2. Balti- 
more: The Johns Hopkins Press, pp. 647-1500. 



1972 FOSSIL PELOMEDUSID TURTLE 13 

Weyl, R. 1966. Geologic der Antillen. Beitriige zur Regionalen Geologic 

der Erde, vol. 4. Berlin-Nikolassee: Gebriider Borntraeger. vii and 

410 pp. 
Williams, E. E. 1950. Trs/uJo cubnisis and the evolution of western 

hemisphere tortoises. Bull. Amer. Miis. Nat. Hist., 95 (art. 1): 1-36. 
. 1952. A new fossil tortoise from Mona Island, West Indies, 

and a tentative arrangement of the tortoises of the world. Bull. Amer. 

Mus. Nat. Hist., 99(art. 9): 541-560. 

1954. A key and description of the living species of the 



genus Podocnemis [scnsu Boulenger) (Testudines, Pelomedusidae). 
Bull. Mus. Comp. Zool., 111(8): 279-295. 

Wood, R. C, .\nd M. L. Diaz de Gamero. 1971. Podocnemis vcnezuelen- 
sis, a new fossil pelomedusid (Testudines, Pleurodira) from the Plio- 
cene of Venezuela and a review of the history of Podocnemis in South 
America. Breviora, No. 376: 1-23. 

Zangerl, R. 1948. The vertebrate fauna of the Selma Formation of Ala- 
bama. Part II. The pleurodiran turtles. Fieldiana: Geol. Mem., 3(2): 
19-56. 



in u.' •i^. v^-v-' ■ 



Linr.ARY 
MAR 1 8 1985 

BREVIORA 

Miuiseeni of Comparative Zoology 

Cambridoe, Mass. August 11, 1972 Number 393 

MINUTE JAMAICAN PROSOBRANGH 

GASTROPODS: 

STOASTOMA AND ITS GONGENERS 

Kenneth J. Boss 

Abstract. Several generic level taxa have been allied to the minute 
terrestrial helicinid Stoastoma (Mollusca; Gastropoda; Prosobranchia). 
The type-species of each taxon is illustrated for the first time and the 
status of each is discussed. The nominal congeners are four recognizable 
genera that are virtually restricted to Jamaica: fVilkinsoriura, Leii'is'ia, 
Stoastoma with its synonym Hcmicyclostoma, and Fadycnia with its synonyms 
Metcalfcia, Pctitia, Lindslcya, and Blaiidia. 

INTRODUCTION 

Among the terrestrial prosobranch gastropods, the family 
Helicinidae is exceptionally rich in species and is developed in 
two primary regions of the world, southeastern Asia and the 
West Indies. In the latter are several radiations that have been 
accorded subfamilial and generic rank (Keen, 1960). One 
genus, Stoasto7na C. B. Adams, consists of species with minute 
shells, usually in the size range of 1-3 mm in diameter. Of the 
New World helicinids, it has been almost totally neglected, be- 
cause of nomenclatorial problems and the tiny size of the shells. 
This paper reviews the generic units of Stoastoma and provides 
the first illustrations of their type-species. 

In 1849 C. B. Adams described the genus Stoastoma. Subse- 
quent to Adams' death in 1853, Edward Chitty, an amateur 
conchologist, English lawyer, and judge who lived in Jamaica, 
took up the task of describing land snails from this Antillean 
island. In 1857, after his return to England, Chitty published 
an overzealous paper on Stoastoma in which he recognized the 
group as a separate family, the Stoastomidae, and proposed 
seven new genera, sixty-one new species, and two new varieties 
from Jamaica. Although these descriptions pay considerable 
attention to conchological detail, none of the species was illus- 



2 , BREVIORA No. 393 

trated and there is some confusion as to the type-species of the 
subgenera. 

Clench and Turner (1950) provided a list of Chitty's taxa 
but did not figure any Stoastoma. Baker (1922, 1934) and 
Wenz (1938) treated some aspects of the taxonomy of 
Stoastoma, while Keen (1960), though listing the taxa and 
synonymizing some of the subgenera, did not figure any type- 
species. Herein are illustrated the type-species of the generic 
taxa described by Adams and Chitty, along with corrected type 
designations. 

Type designations have been given bv Baker (1922), Wenz 
(1938), Clench and Turner [\9bQi), and Keen (1960) but 
most of Chitty's taxa, in accordance with the International 
Code of Zoological Nomenclature (Article 68 [C], Recom- 
mendation 69 B [2] ) , ha\e their type-species by virtual tau- 
tonymy. Pfeiffer (1858, 1865, 1876) also provided lists of 
species. 

C. B. Adams frequently did not give any specific type locality 
for his species; the localities provided herein are those of the 
lectotypes here designated. Principle citations are also pro\ided. 

ACKNOWLEDGMENTS 

Mr. John Peake of the British Museum (Natural History) 
kindly arranged for the loan of specimens from the Chitty Col- 
lection. The manuscript was critically read by Dr. R. D. 
Turner and Messrs. R. I. Johnson and M. K. Jacobson. 

LIST OF THE TAXA 

Blandia Chitty 1857, PZSL, pp. 167 & 169 [type-species, by virtual tau- 
tonymy, Stoastoma blandianum C. B. Adams 18+9, lectotype here selected 
and figured (Figs. 20-22), MCZ 177243; type-locality. Pease River, Man- 
chester, Jamaica]; Pfeiffer, 1865: 207; 1876: 237; Baker, 1922: 57; Pilsbry, 
1930: 230; Baker, 1934: 14; Wenz, 1938: 446; Clench and Turner, 1950: 9; 
Keen, 1960: 288. 

Fadyenia Chitty 1857, PZSL, pp. 167 & 168 [type-species, by virtual tau- 
tonymy, Stoastoma fadyrnianum C. B. Adams 1849, lectotype here selected 
and figured (Figs. 23-25), MCZ 177233; type-locality, hills SW of Port 
Henderson, Jamaica] ; Pfeiffer, 1865: 190; 1876: 237; Baker, 1922: 57; Baker, 
1934: 3; Wenz, 1938: 446; Clench and Turner, 1950: 9; Keen, 1960: 288. 

Hemicyclostoma Chitty 1857, PZSL, p. 163 {nomfn riudum) ; Pfeiffer, 
1865: 191 [type-species, by original designation, Stoastoma pisum C. B. 
Adams 1849, lectotype here selected and figured (Figs. 1-3), MCZ 177234], 
non Grav 1840. 



1972 STOASTOMA 3 

Lridsia Chitty 1857, PZSL, pp. 167 i- 168 [type-species, by subsequent 
designation (Baivcr, 1922: 57), Stoastoma philippianum C. B. A(]ains 1851, 
lectotype here selected and figured (Figs. 10-12), MCZ 177241; type- 
locality. Burnt Hill near Ackendown, Westmoreland, Jamaica]; Pfeiffer, 
1865: 185; 1876: 237; Baker, 1934: 14; Wenz, 1938: 447 (incorrectly 
designated Stoastoma ayassiziannm C. B. Adams 1849 as tyjie-species ) ; 
Clench and Turner, 1950: 10 (incorrectly designated Stoastoma agassizianum 
C. B. Adams 1849 as type-species); Keen, 1960: 228. 

Lindslfya Chitty 1857, PZSL, pp. 167 & 169 [type-species, by virtual 
tautonymy, Stoastoma lindslcyanum C. B. Adams 1849, lectotype here 
selected and figured (Figs. 4—6), MCZ 177238; type-locality, Manchester, 
Jamaica]; PfeiflFer, 1865: 200; 1876: 237; Baker, 1922: 57; Baker, 1934: 3; 
Wenz, 1938: 446; Clench and Turner, 1950: 10; Keen, 1960: 288. 

Metcalfeia Chitty 1857, PZSL, pp. 167 & 168 [type-species, Metcalfcia 
mrtcalfe'iana Chitty 1857, by virtual tautonymy, lectotype here selected and 
figured (Figs. 7-9), BMNH 57.12.1.218; type-locality, Hanover, Jamaica]; 
Pfeiffer, 1865: 192; 1876: 237; Baker, 1922: 57 (incorrectly designated 
Stoastoma ch'ittyanum C. B. Adams 1849 as type-species); Baker, 1934: 3; 
Wenz, 1938: 446; Clench and Turner, 1950: 10 (designated M. mrtcalf- 
e'iana Chitty 1857 as type-species); Keen, 1960: 288. 

Pctitia Chitty 1857, PZSL, pp. 167 & 168 [type-species, by virtual tau- 
tonymy, Stoastoma pctitianum C. B. Adams 1851, lectotype here selected 
and figured (Figs. 17-19), MCZ 177237; type-locality. Pease River, Man- 
chester, Jamaica]; Pfeiffer, 1865: 196; 1876: 237; Baker, 1922: 57: 1934: 
14; Wenz, 1938: 446; Clench and Turner, 1950: 11; Keen, 1960: 288. 

Stoastoma C. B. Adams 1849 (September), Monograph of Stoastoma, a 
new genus of new operculated land shells, pp. 1—16, Amherst, Massachu- 
setts; 1849 (September), Contributions to Conchology, no. 1, p. 16 [type- 
species, by subsequent designation, Chitty 1857, PZSL, p. 167, Stoastoma 
pisum Adams 1849, lectotype here selected and figured (Figs. 1-3), MCZ 
177234; type-locality, Manchester, Jamaica]; Pfeiffer, 1858: 170; 1865: 184; 
1876: 237; Baker, 1922: 57; Wenz, 1938: 446; Keen, 1960: 287. 

tVilkinsonara Chitty 1857, PZSL, pp. 167 &: 168 [type-species, by virtual 
tautonymy, Stoastoma vilkittsoniae C. B. Adams 1851'; lectotype here se- 
lected and figured (Figs. 13-16), MCZ 177251; type-locality, Vallah's Hill, 
Jamaica]; Pfeiffer, 1865: 186; 1876: 237; Baker, 1922: 57 as Wilkmso7iia 
ivilkinsoriiar; 1934: 14; U'enz, 1938: 446 (incorrectly designated Stoastoma 
goiihi'iantim C. B. Adams 1849 as type-species) ; Clench and Turner, 1950: 
12. 



'The spelling of this specific name was altered to '■■ivilkirisonarana' by 
Chittv, 1857: 168, to 'ii-ilkinsoniae' by Baker, 1922: 57, and to '-vHko/iso/iiar' 
by Clench and Turner, 1950: 12. 



4 BREVIORA No. 393 

DISCUSSION 

Pilsbry and Brown (1910) noted the occurrence of Stoastoma 
pisum, a terrestrial form living on stones, in the vicinity of 
Mandeville, Jamaica. They figured the radula and stated that 
it confirmed the position of the genus in the family Helicinidae. 
That is, the radula of S. pisum exhibits the essential features 
of a helicinid radula (Troschel, 1857) : a single rachidian 
tooth, flanked by A, B and C centrals, a lateral complex con- 
sisting of a comb-lateral and an accessory plate, and a marginal 
complex consisting of numerous teeth or uncini. This gives the 
formula (MC) (LC) CBARABC (LC) (MC). As illustrated 
by Baker (1922), its specific features include an A central with 
a single large outer hook (resembling Lucidella) , a B central 
with three heavy lobes, and a C central with four. The comb 
lateral has three large cusps and the inner marginals (first 14) 
are unicuspid, whereas the other six are bicuspid. 

In his extensive study on the radula of the Helicinidae, Baker 
(1922) retained Stoastoma, provisionally placing most of the 
Chitty subgenera into its synonymy and claiming that it forms 
a connecting link between the subfamilies Vianinae and Heli- 
cininae. He pointed out further (1922: 58) that, "Since Chitty's 
description of the many species and 'genera,' the group seems to 
have received little attention." He also tentatively included 
Lindsleva in the synonymy of Lucidella because of a similarity 
in radular structure. 

Subsequently, Baker (1934) again dealt with some of Chitty's 
generic taxa. He separated Stoasto?na s. s. and considered 
Fadyenia at the generic level, allocating Lewisia, Wilkinsonaea, 
Blandia, Petitia, Metcalfeia and Lindsleya either into synonymy 
or to unclearly defined groups. For example, Lewisia was recog- 
nized as a subgenus of Fadyenia and Wilkinsonaea tentatively 
synonymized. Likewise, Blandia was assigned a subgeneric status 
with Petitia as a synonym. And lastly, Fadyenia s. s. included 
both Metcalfeia and Lindsleya. 

Keen ( 1960) also considered these taxa. Stoastoma s. s., with 
its exact synonym Hemicyclostoma, was referred to the nominate 
subfamily. Fadyenia, placed in the Ceratodiscinae, was divided 
into three subgenera: Fadyenia s. s. with Lindsleya and Metcal- 
feia as synonyms, Blandia with Petitia and Wilkinsonaea as 
synonyms, and Lewisia. 

The study of the type-species of Chitty's genera indicates that 
there should be some rearrangement of the taxa. Based on the 



1972 STOASTOMA 5 

structure and morphological features of the shell, four distinct 
groups can be recognized and these characteristics are such as 
to be accorded generic rank. 

As in other helicinids, the internal portion of the shell is 
absorbed in Sloasloma, a feature long ago pointed out by Bland 
(1858j. The shells of most of the species described by Chitty 
and Adams are uniqueh tiny and rather strongly sculptured. 
Accentuated surface features are apparently rather rare in the 
Helicinidae, having been noted in some E?noda-\ike forms which 
are accorded generic rank by Clench and Jacobson (1971). 
Among all the groups, Stoastoma s. s. is distinct in the size of 
the shell alone, being two or three times larger than any of the 
forms described by Chitty. Additionally there is a spur-like 
flaring of the base of the aperture (Fig. 3). The shell is heavy, 
strong, and globose. All of Chitty"s other generic taxa differ 
in being smaller and more delicate. 

Lewisia vFigs- 10-12) is distinguished by its lack of sculpture. 
Its smooth discoid or lens-like shell is distincti\e. Additionally 
it is imperforate with the umbilicus sealed and with a spiral 
ridge extending to the base of the aperture. Chitty (1857) 
characterized Lewisia as being "Quasi double-mouthed." I 
imagine that he was referring specifically to a teratological 
specimen of Lewisia agassiziana C. B. Adams. The type-lot of 
this species ( BMNH 54.4.24.473-474) contains tw^o specimens, 
one of which has an unusual sac-like growth over the umbilical 
area. A similar structure also occurs in the unique (single) 
specimen of L. macandrewiana Chitt\, according to his de- 
scription. The dominant features of the type-species of Lewisia, 
i.e., its smooth, imperforate shell, are apparently not shared by 
the species that Chitty included in Lewisia, namely agassiziana 
C. B. Adams, woodwardiana Chitty and macandrewiana Chitty. 
These nominate species are sculptured; Lewisia may then be 
monotypic. 

Likewise, WUkinsonaea is most unusual ''Boss, 1972). Of a 
depressed sub-discoidal shape, it possesses markedly strong spiral 
sculpturing or carinae in the form of ridges which terminate 
as bulbous, digitiform extensions of the aperture. With an open 
umbilicus, it preserves its unique nature by possessing a \-ery 
unusual operculum, which is deeply conca\e, finely punctate 
and auriculate at both extremities. Also the last whorl is extra- 
ordinarily produced and detached from the body whorl. 

All of Chitty's other generic taxa are synonymous and can be 
placed in Fadyenia, since it has page priority and Baker (1934) 



6 BREVIORA No. 393 

utilized it preferentially. The shells vary in size from the very 
small subdiscoidal Stoastoma blandianum with only 3/2 whorls 
to the larger globose S. lindsleyanum with 5^2 whorls. How- 
ever, conchological features indicating a basic similarity among 
all the type-species include: a \ariably developed, but distinct 
sculpture consisting of nvmierous spiral striations; a detached 
portion of the last whorl; an irregularly semicircular aperture 
with the central margin smooth and more or less straight and 
the external margin frequently formed into a recurv^ed lip and 
variously scalloped by the terminations of the spiral carinae; a 
relatively deep, open umbilicus. The taxa synonymized under 
Fadyenia include Metcaljeia, Petitia, Lindsleya, and Blandia. 

In addition to the species of Stoastoma, Wilkinsonaea, Lewisia, 
and Fadyenia described by Chitty and C. B. Adams from 
Jamaica, several other West Indian forms have been ascribed 
to Stoastoma. Pfeiffer (1857: 51) described S. portoricense 
from the island of that name collected by Hjalmarson on a 
plantation near Manati. Crosse (1892: 46) remarked how 
this form links Puerto Rico and Jamaica zoogeographically. 
Although Dall and Simpson (1901: 447) cited its occurrence, 
van de^r Schalie (1948: 22-23, pi. 1, fig. 7) was the first to 
figure the species and provide a list of several localities. Ap- 
parently the species is not a marked calciphile, occurs at eleva- 
tions to 3000 feet, and is predominantly distributed in the 
northern and western parts of the island. Although nominally 
placed in Fadyenia by van der Schalie, the shell is distinctly more 
highly spired than Jamaican species. Its small size, somewhat 
detached outer whorl, and umbilication show some affinity with 
Jamaican Fadyenia. 

From Haiti, Weinland (1862) described Stoastoma haitianum, 
which was collected some distance from Jeremie in the south- 
west of the island; he stated that the species was related to 
S. philippianiwi C. B. Adams. Pfeiffer (1862), in his com- 
ments on Weinland's paper, remarked that this was an im- 
portant find, considering the habitat. He related the species 
to Metcaljeia, particularly S. chittyanum C. B. Adams. Crosse 
(1891) also noted that with the occurrence of a species of 
Stoastoma on Haiti, the islands of Puerto Rico and Haiti were 
zoogeographically related to Jamaica. A second species from 
Hispaniola was described as S. domingensis by Vanatta (1920: 
206, pi. 6, figs. 4, 5) from one mile NW of Sanchez, Santo 
Domingo. Although he did not mention Weinland's species, 
he compared the form with S. leanum C. B. Adams from 



1972 STOASTOMA 7 

Jamaica, a species that Chitty (1857) had placed in Lindsleya. 
S. doyningensis is very similar to S. portoricense and can be 
referred to Fadyenia on the same basis and with the same reser- 
vation that portoricense is. 

An exceptionally tiny shell (0.7 mm in height) was described 
by Pilsbry as S. atomus (1930: 230, pi. 15, figs. 4, 4a, 4b). 
It was taken near George Town, Grand Cayman Island. Pilsbry 
placed the species in Blandia and stated that it was comparable 
to S. blandianum C. B. Adams. Indeed, a comparison of his 
figures with those of S. blandianum (Figs. 20-22) shows that 
the species are very similar, though S. atomus is decidedly smaller. 
Pilsbry also commented that though Stoastoma and its relatives 
belonged to the Hclicinidae on the basis of the radula, they were 
unique in their "peculiar gait." In his list of Cayman Island 
land mollusks. Clench (1964) also included S. atomus. 

Lastly Baker (1922) described a Lucidella venezuelensis 
from Palma Sola, Estado Falcon, \^enezuela, which he referred 
to Fadyenia [in litt. to M. K. Jacobson, 17 Nov. 1949). Pro- 
viding a figure of the shell and relating the species to S. dom- 
ingensis, he recounted the species more thoroughly in 1923. 

The three species, venezuelensis, domingensis, and portoricense 
are all quite similar to one another. Though tentatively placed 
in Fadyenia, they are distinguished from the Jamaican forms 
by being distinctly higher or more conic in shape, with a char- 
acteristically detached and somewhat depressed outer whorl. In 
contrast, Pilsbry's species from the Caymans is virtually con- 
specific with blandianum. 

In summary, of the generic le\el taxa attributed to the so- 
called Stoastomidae by Chitty (1857), four genera, which are 
xirtualh' restricted to Jamaica, are recognizable and retained 
in the Helicinidae: Wilkinsonaea, Lewisia, Stoastoma with its 
synonym Hemicyclostoyyia, and Fadyenia with its synonyms 
Metcaljeia, Petitia, Lindsleya and Blandia. 



8 BREVIORA No. 393 

LITERATURE CITED 

Adam3, C. B. 1849. Monograph of Stoastoma, a new genus of new opercu- 
lated shells. Privately printed. Amherst, Massachusetts. 16 pp. 

Baker, H. B. 1922. Notes on the radula of the Helicinidae. Proc. Acad. 
Nat. Sci. Philadelphia, 74: 29-67, pis. 3-7. 

1923. The Mollusca collected by the University of Michi- 
gan-Williamson expedition in Venezuela. Occ. Papers Mus. Zool., Univ. 
Mich., No. 137. 58 pp., 5 pis. 

1934. Jamaican land snails. Nautilus, 48: 1-14, 60-67. 



Bland, T. 1858. On the absorption of parts of the internal structure of 
their shells by the animals of Stoastoma, LucidcUa, Trochatella, Hclicina 
and Proserpina. Ann. Lyceum Nat. Hist. New York, 6: 75-77. 

Boss, K. J. 1972. Chitty Chitty Bang Bang. Bull. Amer. Malac. Union 
for 1971, p. 11, pi. 2. 

CHiTTi', E. 1857. On Stoastomidae as a family and on seven proposed 
new genera, sixty-one new species, and two new varieties from Jamaica. 
Proc. Zool. Soc. London, pp. 162—200. 

Clench, W. J. 1964. Land and freshwater Mollusca of the Cayman 
Islands, West Indies. Occ. Papers Moll., Harvard Univ., 2(31): 345- 
380, pis. 61-63. 

, AND M. K. Jacorson. 1971. Monograph of the Cuban 

genera Emoda and Glyptcmoda (Mollusca: Archaeogastropoda : Heli- 
cinidae). Bull. Mus. Comp. Zool., 141(3): 99-130, 7 pis. 

, AND R. D. Turner. 1950. Edward Chitty, with a bibliog- 



raphy and a catalogue of his species of Jamaica land moUusks. Occ. 

Papers Mus. Inst. Jamaica, Kingston, No. 1. 12 pp., 1 pi. 
Crosse, H. 1891. Faune malacologique terrestre et fluviatile d'ile de Saint- 

Domingue. Jour. Conch., 39: 69—211. 
1892. Faune malacologique terrestre et fluviatile d'ile de 

Portorico. Jour. Conch., 49: 5—71. 
Dall, W. H., AND C. T. Simpson. 1901. The Mollusca of Porto Rico. U. S. 

Fish. Comm., Bull., 20(1): 351-524. 
Keen, A. M. 1960. In Moore, R. C. (ed.). Treatise on Invertebrate Pale- 
ontology. Mollusca, Helicinidae, 1: 285-288, figs. 186-187. University of 

Kansas. Lawrence, Kansas. 
Pfeiffer, L. 1852. Monographia Pneumonopomorum Viventium. Cassell. 



439 pp. 



43-52. 



1857. Diagnosen neuer Landschnecken. Malak. Blat., 3: 



1858. Monographia Pneumonopomorum Viventium. Supple- 

mentum primum. Cassell. 249 pp. 
1862. Bemerkungen zu den beschriebenen Arten. Malak. 



Blat, 9: 199-202. 
1865. Monographia Pneumonopomorum Viventium. Supple- 



mentum secundum. Cassell. 284 pp. 
1876. Monographia Pneumonopomorum Viventium. Supple- 



mentum tertlum. Cassell. 479 pp. 



1972 STOASTOMA 9 

PiLSBRY, H. A. 1930. Results of the Pinchot South Sea Expedition. 1. 
Land moliusks of the Caribbean Islands, Grand Cayman, Swan, Old 
Providence and St. Andrew. Proc. Acad. Nat. Sci. Philadelphia, 82: 
221-261, pis. 15-19. 

ScH.ALiE, H. VAN DER. 1948. The land and fresh-water moliusks of Puerto 
Rico. Misc. Publ., Mus. Zool., Univ. Michigan, No. 70. 134 pp., 14 pis. 

Troschel, F. H. 1857. Das Gebiss der Schnecken. Berlin. Vol. 1: 80-91, 
figs. 57—67 (Helicinidae). 

Vanatta, E. G. 1920. New land shells. Proc. Acad. Nat. Sci. Philadelphia, 
72: 203-206, pi. 6. 

Weinland, D. F. 1862. Diagnosen einiger neuen westindischen Land- 
schnecken. Malak. Bliit., 9: 194-199. 

Wenz, W. 1938. Gastropoda, hi Handbuch der Palaozoologie. Helicini- 
dae, 6: pt. 1, pp. 435-448, figs. 1071-1118. Berlin. 



10 BREVIORA No. 393 



Plate I. 
Figures 1-3. The type-species of Stoastoma s. s. Stoastoma pisum C. B. 
Adams 1849, Monograph of Stoastoma, p. 11, lectotype, here selected, 
MCZ 177234, dia. = 3.7 mm, ht. = 3.3 mm; dorsal, apertural, and ventral 



views. 



Figures 4—6. The type-species of Lindsleya Chitty. Stoastoma lindsleyanum 
C. B. Adams 1849, Monograph of Stoastoma, p. 12, lectotype, here selected, 
MCZ 177238, dia. = 2.6 mm, ht. ^ 2.2 mm; dorsal, apertural, and ventral 
views. 

Figures 7—9. The type-species of Mctcalfeia Chitty. Metcalfeia metcal- 
feiana Chitty 1857, p. 179, lectotype, here selected BMNH 57.12.1.218, dia. 
= 2.6 mm, ht. =^ 2.2 mm; dorsal, apertural, and ventral views. 



1972 



STOASTOMA 



11 




12 BREVIORA No. 393 



Plate II. 

Figures 10-12. The type-species of Lcivisin Chitty. Stoastoma ph'tlip- 
pianiim C. B. Adams 18 51, Contributions to Conchology, 9: 158, lectotype, 
here selected, MCZ 177241, dia. = 2.2 mm, ht. ^ 1.1 mm; dorsal, apertural, 
and ventral views. 

Figures 13-16. The type-species of Jf'ilkiiisoriara Chitty. Stoastoma 
vilk'insoniac C. B. Adams 1851, Contributions to Conchology, 8: 148, lecto- 
type, here selected, MCZ 177251 (lost after photographs made), about 1.8 
mm in diameter and 0.9 mm in height; dorsal, two apertural (showing 
protrusion of digitate aperture), and ventral views. 

Figures 17-19. The type-species of Pctitia Chitty. Stoastoma pctitianum 
C. B. Adams 1851, Contributions to Conchology, 8: 151, lectotype, here 
selected, MCZ Ml 111 , dia. =: 1.7 mm, ht. =: 1.1 mm; dorsal, apertural, 
and ventral views. 

Figures 20-22. The type-species of Blandia Chitty. Stoastoma blandianum 
C. B. Adams 1849, Monograph of Stoastoma, p. 6, lectotype here selected, 
MCZ 177243, dia. = 1.3 mm, ht. = 0.7 mm; dorsal, apertural, and ventral 
views. 

Figures 23-25. The type-species of Fadyenia Chitty. Stoastoma fadycnla- 
num C. B. Adams 1849, Monograph of Stoastoma, p. 7, lectotype, here 
selected, MCZ 177233, dia. = 2.1 mm, ht. = 1.7 mm; dorsal, apertural, 
and ventral vaews. 

All specimens were coated with magnesium before being photographed. 



1972 



5TOASTO>rA 



13 







MAR 1 8 198^ 

B R E V I O R A.* 

Muiseom of Comparative Zoology 

Camrridge. Mass. August 11, 1972 Number 394 



THE CHANARES (ARGENTINA) 

TRIASSIC REPTILE FAUNA. 

XV. FURTHER REMAINS OF THE 

THECODONTS LAGERPETON AND LAGOSUCHUS 

Alfred Sherwuud Romer 

Abstract. Further fragmentary postcranial remains in the Chanares 
material attributable to the pseudosuchians Lagcrpcton chanarcnsis and 
Lagosuchiis talampayensis are described and figured. A sepcimen including 
a peculiar pelvis, hind leg, and vertebrae is described as Lagosuchus 
liUocnsis sp. nov. 

In addition to material definitely attributable to the seven 
thecodonts previously described in this series of papers on the 
Chanares fauna, various nodules collected in 1964-65 include 
numerous fragmentary remains of thecodonts, mainh- of small 
size. I ha\e gone o\-er this series of specimens and attempted 
to identify such materials. Some apparently represent immature 
specimens of Gracllhuchus and Chanaresuchus; some remain 
problematical; still others seem certainly to pertain to the 
small thecodonts that I hax'e described as Lagerpeton chayia- 
rensis and Lagosuchus talampayensis (Romer, 1971). 

Lagerpeton chanarensis. I have pre\iously described the type 
specimen, consisting of the slender hind leg of a small pseudo- 
suchian in which the elongate pes differs from the typical 
pseudosuchian pattern in that digit 4 is longer than digit 3, 
and digit 1 is unusuallv short; I also mentioned the presence 
in the Instituto Lillo collections of an almost identical specimen. 
I may note that in the type the lengths of femur, tibia, and 
metapodials 1-4 are, in order: 77, 92, 8, 24, 45, and 48 mm. 

Further materials apparently belonging to Lagerpeton are 
present in se\eral Chaiiares nodules. In one nodule \vhich, in 
addition to much of a gomphodont skeleton, included the t\ pe of 
Lewisuchus admix tus f Romer, 1972) and a few bones of Lago- 
suchus, there are materials that appear to pertain to Lagerpeton 



BREVIORA 



No. 394 




Figure 1. Laacrpcxon chanarcnsis. Material in nodule with the Lcwi- 
suchus type. A, Left femur in dorsal view, head restored from right femur; 
B, the same in ventral view; C, fibula; D, partial fibula; E, partial tibia; 
F, right ilium; G, dorsal vertebrae; H, presumed sacrals, in ventral view. 
XI. 



(MCZ 4121), but represent an individual somewhat smaller 
than the two pre\'iously described ( Fig. 1 ) . Lying close together 
and presumably associated, they include two femora, with lengths 
of 58 and 62 mm, a fibula 84 mm long, proximal parts of a 
tibia and of a second fibula, a right ilium and part of the left, 
and a number of vertebrae. The limb elements compare well 
in structure with those of the type. The femur has a strongly 
developed sigmoid cur\'ature, the head is well set off from the 
shaft, and the longitudinal ridge representing the fourth tro- 
chanter is well developed. The fibula is expanded distallv for 
articulation with the tarsals. The acetabular cavity of the ilium, 
deeply incised, extends unusually far posterodorsally, and the in- 
ner surface of the bone is deeply grooxed for the sacral ribs. There 



1972 



FURTHER THECOnOXT REMAIN'S 




Figure 2. Left scapulocoracoid of Laijosuchus lalampnyoisls. X +/3- 



are several short series of \ertebrae, rather imperfecth' preserved. 
I figure se\eral elements of a series of 1 2 dorsals, with mean 
central lengths of 7.5 mm. Near the right ilium are three \-erte- 
brae seen in \-entral \ie\v, with central lengths of 6.5 mm; these 
bear short stout ribs, and are apparenth sacrals. In the nodule, 
but somewhat separated, are remains of two scapulocoracoids, 
the better one of which is shown in Figure 2. This measures 
42 mm in height. The scapular blade is relati\ely tall and 
narrow at the base. There is a modestly developed acromial 
ridge. Scapula and coracoid are firmly united, as preser\-ed. The 
coracoid is deep dorso\entrally and narrow anteroposteriorly. 
The glenoid appears to face more directly backward than in 
most pseudosuchians. 

MCZ 3691 is a slab mainlv showing the confused remains of 
two partially articulated gomphodont skeletons. Present, how- 
e\-er, are two pairs of slender tibiae and fibulae; none are com- 
pletely preser\-ed and exoosed, but one tibia has, as far as \-isible, 
a length of 80 mm. These elements quite surely belong to 
Lagerpeton. 

Lagosuchus talajnpayensis. The type remains of this form 
(Fig. 3) were contained on a slab (La Plata Museum 64-XI- 
14-11) that also includes the holotype of Gracilisuchus stipani- 
cicorum fRomer, 1972). Present on the slab are incomplete 
remains of both hind legs and an incomplete humerus with 
articulated radius and ulna. Lengths of femur, tibia, and 



4 



BREVIORA 



No. 394 



A 




<<l 





Figure 3. Holotype of Latjosiichus talampaynisis. A, Tibia, fibula, and 
pes; B, femur and incomplete tibia and fibula; C, incomplete humerus, 
radius, and ulna. X +/3. 



metatarsals 1-4 are, in order: 38-39, 47-48, 15, 23, 26, and 
25 mm. In the type only a few phalanges are preserved, but it 
was obx'ious from the metapodial lengths that digit 4 would 
ha\e been slightly shorter than digit 3; metapodial 1, it will be 
noted, is relati\'ely longer than in Lagerpeton. The foot restora- 
tion given in Figure 2 of my earlier paper (1971) was com- 
pleted from the problematical Instituto Lillo specimen discussed 
below. Vertebrae present on this slab, which I pre\'iously thought 
pertained to this form, appear on closer study to be caudals of 
Gracilisuchus. 

MCZ 4116 is a slab on which are numerous bones of 
Gracilisuchus. At one corner, however, are a few remains of 
Lagosuchus. These include: a femur 37 mm long closely re- 
sembling the type, articulated with which is the proximal end 
of a tibia : a series of dorsal vertebrae ( Fig. 4 ) ; fragmentary 
remains of the pehis. The vertebrae have central lengths of 
slightly over 5 mm. 

In the Tucuman collections there is a hind leg that corresponds 
closely with the type in pattern. Measurements of femur, tibia, 



1972 FURTHER THECODONT REMAINS 




Figure 4. Lagosuchus /(ilarnpnyinsis, MCZ 4116. A series of dorsal 
vertebrae. X 2. 

and metatarsals 2-4 are, in sequence: 42, 48, 24, 28, and 27 
mm, thus closely approximating the holotype in size. 

Lagosuchus Ulloensis sp. nov. 

This species is based on a specimen in the Institute) Lillo, 
Tucuman, collected by Sr. Bonaparte from the Chaiiares beds, 
and kindly loaned to me for study. Included are a complete 
right limb f Fig. 5 ) , the left limb complete except for phalanges, 
a series of vertebrae including the sacral region and the proximal 
part of the tail (Fig. 6), and much of the pelvis. Except for 
larger size, the hind leg elements closely resemble those of the 
type. Length measurements of femur, tibia, and metatarsals 
1-4 are, in sequence: 55, 71, 23, 35, 39, and 38 mm. The 
specimen is almost exacth' 50 per cent larger than the holotype. 
If the lengths of the comparable elements of that specimen be 
multiplied^ 3/2, we obtain: 57, 72, 22.5, 34.5, 39, and 37.5 
mm. In \iew of this close correspondence in proportions, I felt 
justified in concluding that the two were .specifically identical, 
used the foot of this Tucuman specimen in restoring the foot 
(Romer, 1971: fig. 2), and assumed that the small holotype 
was presumably a young indi\idual, the present specimen an 
adult. Now, however, we ha\"e three specimens of femur of 
holotype size, only one larger. It is highly improbable that in 
a collection there would be three immature forms against onlv 
one adult. The differences are too great to be considered as 
sexual or due to indixidual \-ariation. I am therefore forced to 
the conclusion that the present specimen repre'^ents a distinct 
species of Lagosuchus, which can be at present differentiated 
from the genotypic species only on the basis of its 50 per cent 
larger .size. 

The \'ertebral .series preser\-ed appears to be of normal the- 
codont type; the .second and third of the series appear to be 
sacrals, with imperfectlv preserved ribs and with central lengths 
of 6.5 mm; the caudals average 7 mm in central length. The 



BREVIORA 



No. 394 




Figure 5. Lagosuchus lillofiisis, holotype. Left hind leg, complete except 
for unguals. X 1- 



FURTHER THECODONT REMAINS 




Figure 6. I.agosuchus ViUocnsis, holotype. A series of vertebrae including 
sacrum of proximal part of the tail. X 3/2. 



first chexron present is between presumed caudals 3 and 4. 
Much of the pehis is preserved. Interpretation of its peculiar 
structure should best be delayed until a more complete specimen 
is found. 



REFERENCES CITED 

ROMER, A. S. 1971. The Chanares (Argentina) Triassic reptile fauna. 

X. Two new hut incompletely known long-limbed pseudosuchians. 

Breviora, No. 37S: 1-10. 
1972. The Chafiares (Argentina) Triassic reptile fauna. 

XIV. Lrivisuc//us aJmixtits, gen. et sp. nov., a further thecodont from 

the Chafiares beds. Breviora, No. 390: 1-13. 



^^'^. ZCWIP. ZOOL 

BREVIORA 

U.NI/ VARSITY 

IMiuiseom of Comparative Zoology 

Cambridge, Mass. Xovkmber 6. 1972 Xumber 395 



THE CHANARES (ARGENTINA) 

TRIASSIC REPTILE FAUNA. 

XVI. THECODONT CLASSIFICATION 

Alfred Sherwood Romer 

Abstract. The various non-phytosaurian genera of the order Thecodontia 
and Triassic crocodilians are reviewed and assembled in a classification. 
Four suborders of thecodonts are recognized: Proterosuchia, Parasuchia, 

Aetosauria, and Pseudosuchia. 

Although it has lons^ been aa^reed that the order Thecodontia 
is the basic stock from which arose the important later archo- 
saur groups — dinosaurs, pterosaurs, crocodiles, and bird an- 
cestors — we suffered until recent decades from a \erv limited 
knowledge of the membership of the order. Apart from the 
abundant phxtosaurs of the late Triassic, we knew only a score 
or so genera of thecodonts, mainly late sur\i\ors. Today the 
picture is greatly changed, mainly owing to work in South 
America and Africa, so that we ha\e about three times as many 
named non-phytosaurian thecodont genera (many, unfortu- 
nately, kno\vn from \"ery incomplete materials). Further in- 
creases are certain to occur in the near future, as a result of 
recent and current work in a \ariety of regions: East Africa 
(Attridge et al., 1964), Australia .Bartholomai and Howie, 
1970), Antarctica (Kitching et al., 1972), India (under the 
auspices of the Indian Statistical Institute ) , and North America. 
In consequence of this recent acti\ity, a number of workers have 
been led to discuss the classification and phylogeny of the or- 
der. Summaries have been made by Rozhdest\enskii 1964) 
for the Russian treatise on paleontology, and by Hoffstetter 
(1955) for the French treatise. Huene has reviewed the group 
on numerous occasions (most recenth in 1956 and 1962); 
Hughes (1963) has summarized postcranial patterns in baiic 
members of the group; Walker has discussed the classification 
of thecodonts and primitive crocodilians in se\'eral papers (par- 



2 BREVIORA No. 395 

ticularly 1968, 1970). My own recent studies of seven new 
Chanares genera (work supported by grants from the National 
Science Foundation) force me likewise to review the group. 
Which I do herewith. 

The origin of the archosaurs is far from clear. Because they 
are diapsids they are frequently associated with the groups now 
generally termed Lepidosauria (Eosuchia, Squamata, Rhyncho- 
cephalia ) , but apart from the temporal region there is little 
resemblance between even the earliest representatixes of the two 
groups. Except for a fragmentary late Permian specimen from 
Russia, there is no trace of an archosaur earlier than the Lys- 
trosaurus zone at the base of the Triassic. Reig (1967, 1970) 
has advanced the theory that they were derived from the pely- 
cosaur family \^aranopsidae, closely related to the forms that 
gave rise to the therapsids; but as I ha\'e pointed out ( 197 Id), 
nearly all the supposed points of resemblance that Reig cites 
are merely primitix'e reptilian features found in pelycosaurs and 
inherited by them (and by archosaurs) from primitive capto- 
rhinomorph cotylosaurs. Petrolacosaurus of the late Carboni- 
ferous (Peabody, 1952) \'ery probably has a diapsid temporal 
region, and may possibly be related to the ancestry of either 
the lepidosaurs or archosaurs or both; but in other regards it is 
very primitive, differing little from romeriid captorhinomorphs, 
or their relatives among the pelycosaurs. There is a gap of a 
full period between the appearance of Petrolacosaurus and of 
either of the major diapsid groups. 

It was long customary to divide the Thecodontia into two 
suborders, the Parasuchia, for the phytosaurs (relatively prom- 
inent in the group as known in older times), and the Pseudo- 
suchia, in which were "lumped" all other members of the order. 
In recent decades it has become apparent that certain early and 
primitive types may well be separated from the more typical 
pseudosuchians to form a suborder Proterosuchia (cf. Reig, 
1970), and some recent essays in classification have advocated 
still further subdivisions. 

It was long believed by many workers (including myself) 
that there was from the beginning a strong trend toward bi- 
pedal progression in the archosaurs, and that quadrupedal mem- 
bers of the group were in general descended from bipedal 
ancestors. However, increasing knowledge of pre-Keuper theco- 
donts has weakened this assumption; the disproportionate de- 
velopment of the hind legs and the powerful tail may be 
interpreted as adaptations for an amphibious life ( Charig, 1 966 ; 



1972 THECODONT CLASSIFICATION 3 

Charig, Attridge, and Crompton, 1965; Charig, 1972). {Meso- 
saurus shows a parallel structure.) Certainly there was a strong 
trend toward bipedalisni in advanced thecodonts, leading to 
bipedal dinosaurs, to birds and to pterosaurs, but \arious theco- 
donts were definitely quadrupeds, and rather surely quadrupedal 
progression in crocodilians and very probably in sauropod dino- 
saurs is primiti\'e.^ 

Primitive thecodont characters. Hughes, as mentioned above, 
reviewed the postcranial features expected in primitive archo- 
saurs, and Reig in 1970 gave an excellent account of primitive 
archosaur structure, which I summarize below. 

The maxilla excluded from the external naris by the pre- 
maxilla; no septomaxilla; an antorbital fenestra of moderate 
size; postfrontal and postparietal present but no supratemporal 
or tabular; parietal foramen at best small (and generallv lost) ; 
two temporal openings; the quadrate extends upward to be 
socketed beneath the squamosal; the lateral temporal opening 
with a "straight" posterior margin slanting backward ventrally 
and bounded by a union of squamosal and quadratojugal; the 
jaw articulation behind the le\el of the posterior margin of the 
lateral fenestra. The basal articulation of the palate mox'eable, 
and an interpterygoid vacuity persistent; palatal teeth reduced 
(and soon lost) ; an external mandibular fenestra almost al- 
ways present; teeth subthecodont. The vertebrae never noto- 
chordal, and generally no more than mildly amphicoelous. Ribs 
double headed, the tuberculum arising from the trans\erse proc- 
ess, which becomes prominent in the dorsal region; the capitu- 
lum anteriorly arising low down in the centrum, but rising pos- 
teriorly to close in on the tuberculum and to mo\e onto the base 
of the transverse process. A powerful tail. 

The coracoid a simple single element, with a roimded border. 
The pubis always turned downward more than in the oldest 
reptiles, but primiti\ely with retention of a considerable area of 
puboischiadic junction. The acetabulum closed. Lack of strong 
inturning of femoral head, suggesting that the proximal limb 
segments were still directed somewhat laterally rather than in 
a plane parallel to the body. As in primitixe reptiles generally, 
the front legs rather shorter than the hind. The gait quadru- 
pedal. The tibia shorter than the femur, the hind foot short, 
the toes perhaps somewhat spreading. Astragalus and calca- 

'Ceratopsians are rather surely secondary (iiiadrupeds, and this, I be- 
lieve, was probably the case with other ornithischian (juadrupeds. 



4 BREVIORA No. 395 

neum both well de\'eloped, the primitive nature of the joint be- 
tween leg and foot uncertain. 

Advanced thecodont characters. Within the order Theco- 
dontia there are obviously many advances from the primitive 
structural situation, and since the thecodonts are by definition 
the stem group from which the various dinosaur types, the 
crocodilians, pterosaurs, and bird ancestors have been derived, 
we would expect a wide variety of diverse patterns in the or- 
der, were it fully known. As yet we have no clear indication of 
lines leading toward pre-avians, pterosaurs, or ornithischians; 
but saurischian dinosaurs are merely logical outgrowths of pat- 
terns obvious amongst thecodonts, and there are numerous in- 
dications of the development of crocodilians and pre-crocodilians 
among Triassic archosaurs. Some of the advanced or aberrant 
trends seen among thecodonts leading toward and to true sau- 
rischians or crocodilians may be listed. 

Variations in snout length; entry of maxilla into narial bor- 
der; backward movement of external nares dorsally; variations 
in development of antorbital vacuity; reduction of upper tem- 
poral opening; modification of lateral temporal opening, by 
forward angulation of quadratojugal and quadrate; forward 
movement of jaw articulation; loss of palatal teeth; loss of mo- 
bility of basipterygoid articulation; closure of palatal vacuities; 
modification of teeth from the simple primitive pattern to a less 
predaceous type; general loss of intercentra behind the axis; 
increase in sacrals; development of proximal accessory processes 
on ribs; increased down-turning of pubis and loss of continuity 
of puboischiadic plate; increasing length and slenderness of 
limbs; increased disproportion of hind limb over fore; excess of 
length of tibia over femur; elongation of metatarsals; trend to- 
ward an avian type of hind foot. A few of these "advances" 
are specializations, but many represent a trend in progressive 
forms toward a saurischian condition. Certain characters are 
indicative of conditions leading toward and to the Crocodilia, 
including reduction of antorbital vacuities, further modification 
of the lateral temporal region to produce the crocodilian type 
of otic notch, development of a secondary palate, posteroven- 
tral development of coracoid, lengthening of proximal carpals, 
and exclusion of pubis from the acetabulum. 

Below, I review briefly the known genera of non-phytosau- 
rian thecodonts and Triassic "pre-crocodiles," arranging most 
of them in provisional family groups and higher categories. 

Proterosuchidae. If one wishes to sort out a basic group of 



1972 THECODONT CLASSIFICATION 5 

Thecodontia as a suborder Protcrosuchia, it is Proterosuchus 
(of which Chasmatosaurus Haughton [1924a] is the better 
known synonym)^ to which one turns for a primitive form. 
Proterosuchus has long been known from the early Triassic Lys- 
trosaurus Zone of South Africa (Broom, 1903) ; a good account 
of the skull is that of Broili and Schroeder ('1934), and Cruick- 
shank (1972) has just described an excellent skeleton. The 
genus is also present in India (w^here fragmentary remains were 
early described as Ankistrodon and Epicampodon [Huene, 
19421), and in China (Young, 1936, 1958, 1963). Elaphro- 
suchus (Broom, 1946; Brink, 1955) is a close South African 
relative. Except for the o\erhang of the premaxilla and the 
somewhat excessi\-e length of the snout, Proterosuchus is almost 
a diagrammatic primiti\-e archosaur, and in the retention of 
dorsal intercentra is even e\'en more primiti\'e than other pro- 
terosuchian t) pes. It lacks, however, the almost universal archo- 
saur external mandibular fenestra (present, however, in 
Elaphrosuchus) . Presumably related to Proterosuchus, but 
known only from very incomplete material, are Chasmatosuchus 
(Huene, 1940) from the early Triassic of Russia and, notably, 
Archosaurus (Tatarinov, 1960) from the late Permian of that 
region - — the oldest known archosaur. 

Proterochampsidae. I ha\'e recently discussed this family 
(Romer, 1971c, 1972a). Chanciresuchus is an adequately 
known form; Gualosuchus, also from the Chaiiares beds, is 
similar. Cerritosaurus from the Santa Maria beds (Price, 1946), 
known only from the skull, is obviously closely related; possibly 
Huene's (1944) genera Procerosuchus and Rhadinosuchus (cf. 
HofTstetter, 1955) represent postcranial material of the same 
form. Protcrochampsa (Reig, 1959; Sill, 1967) from Ischigua- 
lasto, has been described from relatively poor material, but is 
certainly a close relative. Protcrochampsa was claimed by Reig 
and Sill to be a pre-crocodilian, but thought by Walker (1968) 
to be a pre-phytosaur ; Walker (1968, 1970) suggests Cerrito- 
saurus as a possible pre-crocodilian (although he is puzzled by 
the nostril position). There is, howe\'er, no real indication of 
crocodilian relationships in any member of the family, and only 
dorsal migration of the nostrils is positixely suggestive of phyto- 
saur relationship (see below). The proterochampsids are long- 
snouted amphibious forms which ha\'e advanced little beyond 



'Recent rediscovery of the lost type of Proterosuchus has definitely con- 
firmed this long-questioned synonymy (Cruickshank, 1972). 



6 BREVIORA No. 395 

the proterosuchian grade of organization and may well be placed 
in the Proterosuchia. 

Erythrosuchidae. Apart from the genus Proterosuchus the only 
well known form which it is universally agreed pertains to the 
Proterosuchia is Erythrosuchus, from the Cynognathus beds of 
the Karroo, first described by Broom (1906), but more exhaus- 
tively treated by Huene (1911; cf. Brink, 1955). Erythrosuchus 
in a number of regards is as primitive as Proterosuchus, but a 
bit more advanced in, for example, loss of dorsal intercentra 
and loss of palatal teeth. Although the skull and postcranial 
skeleton of this reptile are incompletely known, we are obviously 
dealing here with quite a different sort of animal. The skull is 
short and massively built, the jaw articulation is still placed well 
posteriorly, and the lateral temporal opening is primitively built. 
The body is short and heavily built, the limbs short and stout. 
The dentition is carnivorous in type; dicynodonts were the ob- 
vious prey. 

In recent years several genera, imperfectly known, have been 
described from the early Triassic of Russia. Tatarinov (1961) 
believes them to be generically identical with Erythrosuchus, 
including Dongusia (Huene, 1940, founded on a single verte- 
bra), Garjainia (Otschev, 1958) and Vjushkovia (Huene, 
1960). Included here also by some writers are genera that I 
discuss below under the rubric Prestosuchidae. 

Prestosuchidae (Rauisuchidae). As knowledge of Middle 
Triassic thecodonts increased, it gradually became apparent that 
there continued throughout much of the Triassic period a series 
of relatively primitive thecodonts of large size, with large skulls, 
short legs, and a persistently quadrupedal posture. This was 
early realized by Huene, who tended to confuse such forms with 
larger meml^ers of the aetosauroid group, but ne\ertheless erected 
the family Rauisuchidae for certain thecodonts of this sort. The 
first able discussion of such a group was that of Reig ( 1 96 1 ) ; 
Young ( 1 964 ) expressed similar views, and the evidence for the 
existence of such a group is increasing. Rauisuchus (Huene, 
1944) is poorly known and doubts as to its position have been 
expressed; in consequence, the term Prestosuchidae (Charig, 
1957), based on a better known Santa Maria genus Presto- 
suchus (Huene, 1944), is preferable, Saurosuchus from Ischi- 
gualasto, first described by Reig (1959) and now being 
restudied by Sill, is reasonably placed in the same group, as is 
the fragmentary Luperosuchus skull from the Chanares beds 
(Romer, 1971a). A beast of similar nature, if not referable to 



1972 THECODONT CLASSIFICATION 7 

the Erythrosuchidae, is Shansisuchus of China (Youn^, 1S)64). 
The headless thecodont skeleton from the Cacheuta basin of 
Argentina, first described by Rusconi ( 1 95 1 ) and incongruously 
associated by him with the skull of a flat-headed labyrinthodont, 
is obviously in the same general category and has been named 
Cuyosuchus by Reig (1961). Quite surely a member of the 
group is Ticinosuchus (Krebs, 1965), the only good skeleton of a 
thecodont from the Middle Triassic of Europe. Hoplitosaurus 
[Hoplitosuchus] (Huene, 1944) is represented by poor material 
from the Santa Maria beds, but probably belongs in this fam- 
ily. So very hkely do Mandasuchus (Charig, 1957), Stagono- 
suchus (Huene, 1938), and Pallisteria (now being described by 
Charig ) , represented by incomplete materials from the Manda 
beds of East Africa, as well as Fenhosuchus (Young, 1964) 
from China. Spondylosoyna from the Santa Maria (discussed 
below) may belong here (Charig, 1957). 

Putting together data from the various genera mentioned, we 
may characterize the prestosuchids as being the largest theco- 
donts of (roughly) Middle Triassic times. In the massixely 
built skull the premaxilla sends a strong flange upward poste- 
rior to the naris, broadly excluding the maxilla from that open- 
ing. In Saurosuchus and, apparently, in Luperosuchus, a slit is 
present here between premaxilla and maxilla, and in the skull 
of Shansisuchus, as restored by Young, this slit appears to have 
developed into a good-sized \acuity. The jaw articulation is 
apparently nearly directly below the plane of the posterior mar- 
gin of the lateral temporal fenestra, which is of a primitive 
nature, with Httle or no trend for dcNelopment of an otic notch. 
The palate is primiti\e, with a mobile basal articulation, but 
palatal teeth ha\e been lost. The \ertebrae are short throughout 
the presacral column. The limb girdles are relati\ely primitive, 
with some retention of a puboischiadic plate, although the pubis 
is well turned \^entrally. The limlis are short, although the hind 
limbs are somewhat longer than the fore; the femur is distinctly 
longer than the tibia, the humerus longer than the radius. The 
pose was quadrupedal; the main joint between limb and hind 
foot lay between astragalus and calcaneum. The foot was 
plantigrade, with the toes short and spreading, and little or no 
digital reduction. Large dorsal osteoderms were present. 

This attempt at definition is to some degree composite, but 
will apply, as far as known, to the forms already mentioned. 
The prestosuchids, as here assembled, may be a rather hetero- 
geneous group. It seems certain that they are of proterosuchian, 



8 BREVIORA No. 395 

and presumably erythrosuchid, derivation, and some of the gen- 
era noted above may actually prove to be members of the 
Erythrosuchidae; certainly they should be included in the Pro- 
terosuchia. 

Huene (1944) points out, as does Krebs (1966), that the 
prestosuchid t)pe of foot is of the sort that could have made 
the "C heir other ium" footprints abundant in the early Triassic 
(Soergel, 1925), and not improbably all footprints of this sort 
were due to prestosuchids or their erythrosuchid ancestors. 

Were the prestosuchids a sterile group? They surely have 
nothing to do with the ancestry of ornithischians, crocodilians, 
pterosaurs and birds, or theropod dinosaurs. Under the once 
popular hypothesis that sauropods were of bipedal ancestry, re- 
lationship with these forms too would be ruled out. But Charig 
et al. ( 1 965 ) have presented a strong case for a continuous 
quadrupedal ancestry for the sauropods. The picture is still a 
cloudy one; but it is not impossible that it ma)' one day prove 
that the erythrosuchid-prestosuchid line plodded on, with change, 
through the Triassic to become ancestors of the sauropods. 

Phytosauridae (Belodontidae). The phytosaurs, or Parasuchia, 
were the first thecodonts to be at all well known, and they have 
retained prominence in the order. This is mainly due to the 
fact that they are confined (with one possible exception) to the 
late Triassic redbeds of Europe, North America, India, and 
China and it is from these beds in Europe and North America 
that, until recently, nearly all thecodont material was derived. 
More than 30 genera have been described, and despite work by 
Gregory (1962), Westphal (1963) and, finally by Gregory and 
Westphal combined (1969), the generic situation is still none 
too clear. Their general structure, in which there is relatively 
little variation, can be clearly seen in such works as those of 
McGregor (1906), Camp (1930), and Colbert (1947). Their 
general proportions (and probable habits) were similar to those 
of crocodilians of later periods. The body is low-slung, the limbs 
short (with, as in thecodonts generally, the front limbs shorter 
than the hind ) , the gait, when out of the water, definitely quad- 
rupedal. The limb girdles were persistently primitive in nature 
(except for an anterior incisure in the coracoids) and the pubois- 
chium still retains much of its primitive platelike condition. 
Heavy armor was developed. The palate was persistently primi- 
tive, but the anterior portion of the skull was greatly specialized ; 
there was great snout elongation, but with breathing facilitated 
by movement of the nares far back along the skull roof. 



1972 THECODONT CLASSIFICATION 9 

The phytosaurs flourished greatly in the northern continents 
during the late Triassic, only to be superseded by the croco- 
dilians, already e\ol\-ing at that time. Despite our considerable 
knowledge of late Triassic faunas in South Africa and Argentina, 
no positixe e\idence of the presence of phytosaurs is known in 
the "Gondwanaland"' regions except for peninsular India. Iso- 
lated dermal plates from Madagascar ha\e been compared with 
those of phytosaurs Guth, 1963; Westphal, 1970), but they may 
equalK well pertain to prestosuchids. The discovery by Jaekel 
(1910 J of a phytosaur, Mesorhinus. in the early Triassic Bunter 
has been disturbing, and doul)ts ha\e l^een cast upon its age 
and phytosaurian nature (Gregory, 1962). But recent work 
appears to ha\e confirmed the stratigraphy, and in view of the 
wide radiation of archosaurian types which we now see to ha\-e 
taken place \\ell before the close of the Triassic, the appear- 
ance of a possibly primiti\e ph)tosaur by the end of Bunter 
times seems reasonable. 

Aelosauridae (Stagonolepidae) . The aetosaurs are a well-de- 
fined group; Aetosaurus from the German Keuper ( Fraas, 
1877) "has long been well known, and \Valker (1961) has re- 
cently gi\en an excellent description of Stagonolepis from the 
Elgin Triassic. Other late members of the family which appear 
to be relati\ely unspecialized are Neoaetosauroides (Bonaparte, 
1969a, 1972) from the Los Colorados of Argentina and, appar- 
endy, Stegomus fjepsen, 1948; Walker, 1961) from the New- 
ark series. Earlier is Aetosauroides from Ischigualasto, with 
which Argentinosuchiis, described also by Casamiquela ('I960, 
1961, 1967), may be synonymous (Bonaparte, 1972). In the 
American late Triassic are large aetosaurs with exaggerated 
armor — Typothorax [Episcoposaurus, ? Acorn psosaurus] and 
Desynatosuchus ' Sawin, 1947; Gregory, 1953). The diagnostic 
features of the aetosaurs are ob\ious — nearly complete armor 
sheathing; a short-legged quadrupedal gait; a relati\eh' primi- 
tive pelvic structure; a toothless, rather pigHke snout; large 
nares; deeply incised antorbital vacuities; retention of at least 
some basipterygoid mo\ement; teeth modified from the general 
thecodont pattern. The lateral temporal opening is of interest. 
It is merely a small triangular affair, behind which the quadrato- 
jugal extends broadly upward and markedly forward, to come 
close to a contact with the postorbital. This situation suggests 
that in this regard the temporal region has gone through an 
e\olutionarv process similar to that xvhich I ha\-e suggested for 
GracUisuchus Romer, 1972b). 



10 BREVIORA No. 395 

There are no known intermediates between aetosaurids and 
other thecodonts, although the group cannot be traced back of 
the stratigraphic lexel of the Ischigualasto beds. The heavy 
armor suggests some possible relationship to other types in which 
armor is present (pre-crocodilians, phytosaurs, ornithosuchids) 
but no close relationship is apparent. The "advanced" nature 
of the lateral temporal region suggests some relationship to forms 
in which the original, essentially quadrilateral shape of the lat- 
eral fenestra had become modified; but this feature may have 
evolved in parallel fashion. The strictly quadrupedal nature of 
aetosaurids suggests that they branched off from other theco- 
donts at a \-ery early stage, possibly directly from proterosuchian 
ancestors. But evidence is lacking. 

The aetosaurs are generally included in the Pseudosuchia. 
But It is, I think, preferable to restrict this term to forms of ad- 
vanced nature, tending toward and to a bipedal condition, and 
consider, as Reig (1970) has done, that these aberrant forms 
constitute a separate suborder Aetosauria. 

Euparkeriidae. Euparkeria [Broivniella] of the Cynognathus 
zone of South Africa (Broom, 1913), recently thoroughly re- 
studied by Ewer ( 1965), is one of the most interesting of theco- 
donts. As pointed out by Hughes (1963) and Reig (1970) as 
well as Ewer, Euparkeria retains a large number of primitive 
proterosuchian characters — palatal teeth and dorsal intercen- 
tra, for example. But, on the other hand, this lightly built little 
reptile is a reasonable ancestor for the true pseudosuchians of 
more advanced nature which flourished throughout the later 
stages of the Triassic. Early authors in\'ariably placed Eupar- 
keria among the pseudosuchians. Hughes, Ewer, and Reig 
would include it in a "horizontaF' type of classification in the 
basic proterosuchian group. This is perfectly acceptable; but 
I tend to class it as a primitive but true pseudosuchian, asso- 
ciating it with the more advanced forms to which it, or its close 
relatives, gave rise. 

No further form has been placed with certainty in the Eupar- 
keriidae. Young (1964) would assign here Wangisuchus from 
the Sinokannemeyeria beds of China. The remains are frag- 
mentary, but indicate the presence of a long-legged pseudo- 
suchian at an early Triassic stage, and the assignment may well 
be correct. 

Ornithosuchidae. Ornithosuchus from the Elgin beds (with 
which Walker belie\'es Dasygnathus to be synonymous) has 
been excellently redescribed by Walker (1964). Recently 



1972 THECODONT CLASSIFICATION 11 

(1972b) I have described a small early ornithosuchid, Gracili- 
suchus, from the Chanares beds of Argentina; Bonaparte 
(1969a) has described as an ornithosuchid Riojasuchus from 
the Los Colorados, and an Ischigualasto genus, Venaliconu hus 
(Bonaparte, 1972). 

As best seen in little Gracilisuchus and the larger and later 
Ornithosuchus, we are dealing with gracefulh' built archosaurs, 
in \\hich the slender hind legs and long slender hind toes sug- 
gest that these forms were at least partially bipedal. The skull 
is rather ad\-anced in nature with a large and deeply recessed 
antorbital vacuity. It has a palate in which motility on the 
braincase has been lost and the pterygoids reach the midhne; 
the posterior border of the lateral temporal fenestra is pro- 
nouncedly \'-shaped. Notable is the presence of stout paired 
armor plates firmly sheathing the dorsal surface of the verte- 
brae. Walker has ably argued for considering Ornithosuchus as 
a primiti\'e carnosaurian saurischian dinosaur. 1 have elsewhere, 
howe\-er, expressed doubts about this (as do Bonaparte, 1969b; 
Reig, 1970) and currently tend to believe that the ornitho- 
suchids are more properly to be considered as one of several 
probable lineages amongst advanced thecodonts which w^re ap- 
proaching the theropod le\el of organization, but were not direct 
theropod ancestors. Ornithosuchus itself is too late in time to 
be ancestral to essentially contemporary theropods; further, I 
think it is doubtful that the characteristic dorsal plating, ac- 
quired early in the Triassic, would be abandoned. 

\^arious other genera have been suggested as members of this 
group. Walker, on the assumption that Ornithosuchus is a true 
dinosaur, suggested that the late Triassic carnosaurs Teratosau- 
rus and Sinosaurus were members of this family; this seems ver\- 
doubtful. Parringtonia fHuene, 1939) of the Manda is known 
only from ver)- fragmentary material, but in what we know of 
it, it is comparable with the nearly contemporaneous Gracili- 
suchus. A form which may belong here is Dyoplax from the 
Keuper (Fraas, 1867). This is known mainly from a poorly 
preser\'ed skull and a series of paired dorsal dermal plates. Be- 
cause of the plates, it has frequently been assigned to the aeto- 
saurid group. But there is no evidence of other armor; the 
plates are comparable to those of omithosuchids; and signifi- 
cantly, the cer\ical ribs are of exactly the specialized oxerlapping 
type seen in Gracilisuchus. 

Scleromochlidae. Far remo\ed from most other pseudo- 
suchians is tiny Scleromochlus, known onlv from bones and bone 



12 BREVIORA No. 395 

impressions in nodules from Elgin; first described by Smith 
Woodward in 1907, it was later restudied (with a restoration) 
by Huene (1914a). The trunk and neck are short; in contrast, 
the legs are extremely long ■ — even the front legs, despite the 
fact that there is the usual disparity in length between front and 
hind legs. Huene's skull restoration is mainly hypothetical, as is 
that of the feet (except for the definite close apposition of the 
four long metatarsals). It is tantalizing — but at present use- 
less — to speculate as to the possible relationships of this little 
reptile to the ancestry of the birds or the pterosaurs. 

It is not impossible that Lager peton (Romer, 1971b) is re- 
lated; it is almost equally small, lightly built, and very long- 
legged. 

Miscellaneous pseudosuchians. Among pseudosuchians, in a 
narrow sense of that term — that is, thecodonts progressing to- 
ward or to a bipedal condition — we have discussed the Orni- 
thosuchidae and the very difTerent form Scleromochlus. These 
are clearly defined types, meriting distinction at the family level. 
In addition, however, are a number of forms which do not be- 
long to either of these families, but show varied advances over 
the primitive thecodonts toward bipedality and may well repre- 
sent a series of separate families. Family names have been given 
to such forms in several instances, but at the moment it seems 
inad\'isable to recognize such families, and pro tern they may 
be simply listed as Pseudosuchia incertae sedis. 

Erpetosuchus [Herpetosuchus], known only from a single 
specimen from Elgin (Newton, 1894; Walker, 1970), is quite 
distinctive. The antorbital fenestra is highly developed; the 
teeth are restricted to the anterior part of the jaws. Most notable 
is the temporal region; the lateral temporal opening is reduced 
to a ventral triangle, which is quite unrelated to the crocodilian 
type of modification but rather to the type of emphasis on the 
forward-turned V-shaped modification seen in major develop- 
ment in GracUisuchus. 

Lewisurhus, although represented only by a fraction of a skull 
and incomplete skeleton (Romer, 1972c), is distincti\'e. It is 
primiti\'e in such features as a persistently moveable basiptery- 
goid articulation and a lateral temporal opening of primitive 
nature; it is nevertheless well advanced in a bipedal direction 
and is perhaps tending toward the coelurosaurs. Teleocrater 
from the Manda, which Charig (1957) suggests as a possible 
coelurosaur ancestor, may be related. 

Lagosuchus from the Chahares beds (Romer, 1971b), known 



1972 THECODONT CLASSIFICATION 13 

from little but the hind leg, appears to be an ad\anced pseudo- 
suchian of some sort. 

Triassolestes from the Ischigualasto beds was thought by Reig 
(1963, 1970) to be a dinosaur. There is no proof of this, and 
placing this form in the Thecodontia seems more reasonable. 
Bonaparte (1972) states that nearly complete skeletal remains 
are known. Bonaparte notes that there are elongate proximal 
carpals and hence he is inclined to ally it with Sphenosuchus 
and other pre-crocodilians. But there are no further postcranial 
similarities to the crocodiles, and the skull is quite uncrocodilian 
in build. 

Hesperosuchus (Colbert, 1952) is poorl\- known cranially, 
but is ob\ iously a somewhat adxanced pseudosuchian not closely 
related to the genera already mentioned. Bonaparte (1969a, 
1972) and Walker (1970) believe it to be related to Pseud- 
hesperosuchiis, and to pre-crocodilians. 

Saltoposuchus (Huene, 1921) is shown by Huene in restora- 
tion as if completely known; as may be seen, however, from 
Huene's text, this pseudosuchian is quite inadequately known and 
its relationships essentially indeterminate. It has dorsal armor 
but apparently not of ornithosuchid type. Walker (1970) sug- 
gests crocodilian relationships, but does so without adequate 
reason for this assignment. 

Strigosuchus and Dibothrosuchus are names gi\-en by Simmons 
(1965) to poorly known forms from the late Triassic of China 
which are presumably pseudosuchians of some sort. 

Thecodontia in cert ae sedis. There exists a \'ariety of named 
Triassic genera, represented mainly by highly incomplete mate- 
rials, which are probably thecodont in nature but cannot be 
satisfactorily placed in any specific group. Thus, for example, 
Seemnnnia f Huene, 1958) and Crenelosaurus fOrtlam, 1967) 
ha\e been f regrettably) based on isolated teeth from the Bunter, 
and Rusconi M947b) has given the name Ocoyuntaia to a 
tooth from the Cacheuta beds of Argentina. This last author 
( 1947a) has given the name Typothorax? punctulatus to a good- 
sized dermal plate from the Cacheuta region; ver\- likely tooth 
and plate pertain to prestosuchids. ^Vcllcs (1947) has described 
as Arizonasaurus an archosaurian maxilla from the Moenkopi 
Formation of Arizona; with this he tentati\ely associated other 
materials, some possibly archosaurian, one definitely not. Clar- 
encea (Brink, 1959) was founded on an imperfect skull, appar- 
ently of peculiar type, from the late Triassic of South Africa; 
it is probably a thecodont, but of uncertain relationships. 



14 BREVIORA No. 395 

Ctenosauriscus [Ctenosaurus] from the Bunter ( Huene, 1914c) 
consists of a vertebral column with tall spines; it has often been 
compared with pelycosaurs, but Krebs ( 1 969 ) points out that 
it is archosaurian in nature, not pelycosaurian ; not as yet fully 
described is a comparable vertebra, termed Hypselorhachis 
(Charig, in Harland et al. 1967), from the Manda beds. A 
trend toward spine elongation is not uncommon in archosaurs, 
as seen in the dinosaurs Spinosuchiis from the late Triassic of 
Texas and S pinosaurus from the Egyptian Cretaceous. 

Huene (1944; cf. Bonaparte, 1972) applied the generic name 
Spondylosoyna to an assortment of bones, mainly fragmentary, 
from the Brazilian Santa Maria beds, and assigned the genus to 
the Saurischia. There is no guarantee that all the assortment 
belong to one type of animal (certainly not to one individual), 
and it is not improbable that we are dealing with thecodont 
material. Huene has on several occasions described fragmentary 
materials from the Muschelkalk which he belived to be coeluro- 
saurian or pelycosaurian ; considering their age, they may equally 
well pertain to thecodonts. 

Elachistosuchus (Janensch, 1949) was described as a theco- 
dont, but is now recognized to be a rhynchocephalian (Walker, 
1966). Anisodontosaurus (Welles, 1947) from the Moenkopi 
is sometimes classed among the thecodonts but there is no evi- 
dence to warrant this. 

I feel incompetent to discuss the probable systematic position 
of Podopteryx, based on a partial skeleton from the Mesozoic 
of Siberia which appears to have had membranes between the 
hind legs and tail and which Sharov (1971) assigns to the 
Pseudosuchia, or of Longisquartia^ with long scaly "plumes" 
projecting from its back which Sharov (1970) also considers a 
pseudosuchian. 

Primitive crocodile relatives are frequently confused with 
thecodonts, are difficult in some cases to separate from them, 
and may be discussed here. 

Protosuchidae (Stegojnosuchidae). Apart from forms in which 
crocodilian characters are less marked or of doubtful signifi- 
cance, we find in the later Triassic of most continents forms 
that are unquestionably members of the Crocodilia, although 
primitive in some features. Typical skull features are seen in 
the African genera Notochampsa, and the closely related if not 
identical Erythrochampsa (Broom, 1904, 1927; Haughton, 
1924a, 1924b), and Orthosuchus (Nash, \9&^). Protosuchus 
( Colbert and Mook, 1 95 1 ) is represented by a nearly complete 



1972 THECODONT CLASSIFICATION 15 

skeleton, and a rcintcrpretation of the skull roof Walker, 1968, 
1970) shows excellent agreement with the African genera. 
Walker 1968) has restudied the difTiciilt material of Ste^omo- 
suchus from the Newark series, and bcliexes this pre\iously 
problematical genus to belong to this group. Platyognathus 
from Clhina Young, 1944; Simmons, 1965), although imper- 
fectly known, seems to follow the same pattern and hardly 
merits the erection of a separate family for its inclusion. Alicro- 
champsa Young, 1951; Simmons, 1965) is a small form repre- 
sented only by \ertebrae and armor plating, but as far as can be 
seen is reasonably placed here. 

As far as the material permits, all these protosuchids show a 
long series of crocodilian characters: postcranialh", elongation 
of the coracoid, elongation of proximal carpals, exclusion (or 
near exclusion) of the pubis from the acetabulum, and good de- 
\elopment of dermal armor; in the skull, a flattened table, 
presence of supraorbital bones, small antorbital fenestrae, con- 
siderable de\elopment of a secondary palate; and, most notably, 
the forward mo\ement of the upper end of the quadrate (and 
quadratojugal), deep to the surface, to form the typical croco- 
dilian otic notch, closed behind in more advanced members of 
the order. 

Although "priority,"" as \Valker points out, would insist on 
the use of Stegomosuchidae as the famih name, common sense 
and usage speak for retention of Protosuchidae. The family 
certainly stands in an ancestral position to the more familiar 
crocodilians and is reasonably regarded as forming a primitive 
suborder (or infraorder) of the Crocodilia as the Protosuchia. 

Sphenosuchidae (Pedeticosauridae). Apart from the forms 
just discussed, w'hich are quite surely primiti\e crocodilians, 
there are a number of other Triassic genera which exhibit croco- 
dilian features in less positive fashion. Best known and most 
important here is Sphenosuchiis Haughton, 1915, 1924b; 
Huene, 1925; Broom, 1927; Walker, 1970) from the Upper 
Triassic of South Africa. The skull shows the forward inclina- 
tion of the quadrate expected in a crocodile ancestor; the quad- 
ratojugal is reduced (rather prematureh) ; the antorbital \acuity 
is small; the postfrontal is lost; the basipterygoid articulation is 
fused; there is some de\elopment of a secondary palate; the 
postcranial skeleton is incomplete, but the coracoid is croco- 
dilian, and Walker, in litteris, tells me of further features that 
indicate crocodilian relationships. 

^Vith this genus Walker associates Pedeticosaurus, likewise 



16 BREVIORA No. 395 

from the late Triassic of South Africa (Van Hoepen, 1915; 
Haughton, 1924b). This is known only from a single imperfect 
skeleton, in which the incompletelv preserved lateral temporal 
opening resembles that of Sphenosuchus. A further clue as to 
relationships lies in the fact that in both genera there is an en- 
larged mandibular tooth at about the level of the premaxillary- 
maxillary suture. Hemiprotosuchus (Bonaparte, 1969a, 1972) 
from the Los Colorados of Argentina has a similar skull, and 
like the last two genera has an enlarged lower "canine." For 
this family Walker uses the term Pedeticosauridae ; it seems un- 
fortunate to base the family on this poorly known form rather 
than on the much better known Sphenosuchus. 

Walker (1970) and Bonaparte (1969a, 1972) have sug- 
gested crocodiloid relationships for several other Triassic genera 
such as Hesperosuchus, Saltoposuchus, and Triassolestes. As 
noted earlier, there seems to be little evidence for such assign- 
ment of these genera. A much better case can be made 
out for Pseud hesperosuchus (Bonaparte, 1969a, 1972) from 
the Los Colorados beds, because of the presence of an elongate 
coracoid and elongate proximal carpals. The skull, however, is 
not at all crocodilian. 

To treat forms that have crocodiloid tendencies, but are not 
advanced to the typical protosuchian level. Walker (1970) di- 
vides the order, which he terms the Crocodylomorpha, into two 
suborders, one, Crocodilia, including proper crocodiles plus 
Protosuchia, and a second suborder, Paracrocodylia, in which 
are placed Sphenosuchus and Pedeticosaurus and their supposed 
allies, plus Baurosuchus and Hallopus. Discussion of these last 
two forms is beyond the scope of the present review, but pro- 
visionally I think we may treat other forms that are trending 
from the typical thecodont pattern toward that of the crocodiles 
as members of the Protosuchia in an expanded use of that term. 

A summary of the above discussion may be made in the form 
of a tentatixe outline of classification of thecodonts and primi- 
tive crocodilians: 

Order Thecodontia 

Suborder Proterosuchia 

Family Proterosuchidae (Chasmatosauridae) : Archosau- 
rus, Elaphrosuchus, Chasmatosuchus, Proterosuchus 
[Ankistrodon, Epicampodon, Chasmatosaurus]. 
Family Proterochampsidae : Cerritosaurus, Gualosuchus, 
Chanaresuchus Proterochampsa, ?Procerosuchus, 
?Rhadinosuchus. 



1972 THECODONT CLASSIFICATION 17 

Family Erythrosuchidae : Erythrosuchus [Donnusia, 
Garjainia, Vjushkovia] . 

Famih' Prestosiichidae (Raiiisuchidae) : Prestosuchus, 
Rauisuchus, Saurosuchus, Luperosiichus, Ticinosiichus, 
Alandasuchus, Cuyosuchus, Shansisuchus, HopUtosau- 
rus [Hoplitosuchus], Fenhosuchus, Stagonosuchus, Pal- 
listeria, ISpondylosoma. 

Suborder Parasuchia (Phytosauria) 

Family Phytosauridae (Belodontidae) : Mesorhinus of 
probably early Triassic age and \arious Upper Triassic. 
genera. 

Suborder Aetosauria 

Family Aetosauridae (Stagonolepidae) : Aetosaurus, 
Aetosauroides [?Argentinosuchus], Desmatosuchus, 
Neoaetosauroides, Stagonolepis, Stegomus, Typothorax 
[Episcoposaurus, ?Acompsosaurus]. 

Suborder Pseudosuchia 

Family Euparkeriidae : Euparkeria [Browniella], 
?Wangisuchus. 

Family Ornithosuchidae: Gracilisuchus, VenaticosuchuSy 
Riojasuchus, ?P(irringtonia, Oryiithosuchus 
[?Dasygnathiis, Dasygnathoides], ?Dyoplax. 

Family Scleromochlidae : Scleromochlus, ?Lagerpeton. 

Pseudosuchia, presumably representing a number of dis- 
tinct families: Lagosiichus, Hesperosuchus, Lewi- 
suchus, Salfoposuchus, Strigosuchus, Dibothrosuchus, 
Teleocrater, Erpetosuchus [Herpetosuchus], Triasso- 
lestes. 

Possible Thecodontia, incertae sedis: Seemannia, Cteno- 
sauriscus [Ctenosaurus], Spondylosoma, Arizonasau- 
rus, Ocoyuntaia, "Typothorax punctulatus," Crenelo- 
saunis, Hypselorhachis, Podopteryx, Clarcncea. 

Order Crocodilia 

Suborder Protosuchia 

Family Protosuchidae (Stegomosuchidae) : Protosuchus 
[Archaeosuchus], Notochampsa, Erythrochatnpsa, 
Orthosuchus, ?Platyognafhus, Stegomosuchus, 
?Microchampsa. 
Family Sphenosuchidae (Pedeticosauridae) : Spheno- 

siichus, ?Pedeticosaurus, Heyniprotosuchus. 
Protosuchia?, incertae sedis: Pseudhes perosuc hus . 



18 BREVIORA No. 395 

POSTSCRIPT 

While this paper was in page proof, I received a copy of an 
excellent paper by Bonaparte describing reptiles, mainly theco- 
donts, from the Upper Triassic Los Colorados Formation of 
Argentina. It is mainly devoted to thecodonts. It will be noted 
that his paper went to press before he had received my recent 
papers on Chaiiares thecodonts in this series. For the most part 
our conclusions are in essential agreement. He gives an account 
of the skeleton of Riojasuchus which he assigns to the family 
Ornithosuchidae, as well as Venaticosuchus and possibly Par- 
ringtonia. Of Pseudhesperosuchus, previously known only from 
a preliminary description, he gives an account of the skull and 
much of the postcranial skeleton. He definitely assigns this 
genus to the Sphenosuchidae (which he retains in the Pseudo- 
suchia), as he does Hesperosuchus, whereas I have considered 
Pseudhesperosuchus as doubtfully allied to the Sphenosuchidae 
and have kept Hesperosuchus in the Pseudosuchia in a narrower 
sense and see no reason to ally these two forms. The description 
of the skull of H emiprotosuchus shows clearly that it is a primi- 
tive "crocodiloid," but whereas I included it in the Spheno- 
suchidae, Bonaparte would place it in the somewhat more 
more advanced family Protosuchidae. 

REFERENCE 

Bonaparte, J. F. 1972. Los tetrapodos del sector superior de la fonnacion 
Los Colorados, La Rioja, Argentina. Opera Lilloana, XXII: 1—183. 

LITERATURE CITED 

Attridge, J., H. W. Ball, A. J. Charig, and C. B. Cox. 1964. The Brit- 
ish Museum (Natural History) — University of London joint palaeon- 
tological expedition to Northern Rhodesia and Tanganyika, 1963. 
Nature (London), 201: 445-449. 

Bartholomai, a., and a. Howie. 1970. Vertebrate fauna from the Lower 
Trias of Australia. Nature (London), 225: 1063. 

Bonaparte, J. F. 1969a. Dos nuevas "faunas'" de reptiles Triasicos de 
Argentina. Gondwana Stratigraphy, lUGS Symposium, Buenos Aires, 
1-15 October 1967, UNESCO, pp. 283-284. 

1969b. Comments on early saurischians. Zool. J. Linn. 

Soc, 48: 471-480. 

1972. Annotated list of the South American Triassic 



tetrapods. Proc. Pap. II Gondwana Symposium, South Africa, 1970, 
Pretoria, pp. 665-682. 



1972 THECODONT CLASSIFICATION 19 

Brink, A. S. 1955. Notes on some thecodonts. Navors. Nas. Mu>. Hloeiii- 

fontein, 1 : 141-148. 
1959. A new small thecodont from the red beds of the 

Stormberg series. Palaeont. Africa, 6: 109—115. 
Broili, F., and J. Schroder. 1934. Beobachtungen an Wirbeltieren der 

Karrooformation. \'. Ueber Chasmatosaurus van hocpcni Haughton. 

Sitzungsber. Bayer. Akad. Wiss., 1934, No. 3: 225-264. 
Broo.vi, R. 1903. On a new reptile ( Protcrosuchus fcrgusi) from the 

Karoo beds of Tarkastad, South Africa. Ann. S. Afr. Mus., 4: 159- 

163. 
1904. On a new crocodilian genus (Notochampsa) from the 

Upper Stormberg beds of South Africa. Geol. Mag., ser. 5, 1 : 582- 

584. 
1906. On the remains of Erylhrosuclnts afrlcanus Broom. 



Ann. S. Afr. Mus., 5: 187-196. 
1913. On the South African pseudosuchian Euparkcria and 



allied genera. Proc. Zool. Soc. London, 1913: 613—633. 
1927. On Splicnnsuchus and the origin of the crocodiles. 



Proc. Zool. Soc. London, 1927: 359-370. 
1932. On some South African pseudosuchians. Ann. Natal 



Mus., 7: 5S-S9. 
1946. A new primitive proterosuchid reptile. Ann. Trans- 



vaal Mus., 20: 343-346. 
Camp, C. L. 1930. A study of the phytosaurs, with description of new 

material from North America. Mem. LTniv. California, 10: 1-174. 
Casamiquel.a, R. M. 1960. Noticia preliminar sobre dos nuevos estagono- 

lepoideos Argentines. Ameghiniana, 2: 3-9. 
1961. Dos nuev^os estagonolepoideos Argentines (de 

Ischigualasto, San Juan). Rev. Asoc. Geol. Argentina, 16: 143-203. 
1967. Materiales adicionales y reinterpretation de 



Actosauroidcs scagliai. Rev. Mus. La Plata, New Ser., Paleont., 5: 

173-196. 
Charig, a. J. 1957. New Triassic archosaurs from Tanganyika including 

Mandasuchus and Tcleocratcr. Abstr. Diss. Univ. Cambridge, 1955- 

56: 28-29. 
1966. Stance and gait in the archosaur reptiles. Brit. 

Assoc. Adv. Sci., 22(103): 537. 
1972. The evolution of the archosaur pelvis and hind- 



limb: an explanation in functional terms. In Joysey. K. A., and 
T. S. Kemp, (Eds.), Studies in Vertebrate Evolution (essays pre- 
sented to Dr. F. R. Parrington, F.R.S.). Edinburgh: Oliver and Boyd, 
pp. 121-155. 

J. Attridge, and a. W. Crompton. 1965. On the origin 



of the sauropods and the classification of the Saurischia. Proc. Linn. 
Soc. London, 176: 197-221. 
Colbert, E. H. 1947. Studies of the phytosaurs Machacroprosopus and 
Rutiodon. Bull. Amer. Mus. Nat. Hist., 88: 53-96. 



20 BREvioRA No. 395 

1952. A pseudosuchian reptile from Arizona. Bull. Amer. 

Mus. Nat. Hist, 99: 561-592. 
, AND C. C. MooK. 1951. The ancestral crocodilian Pro- 



tosiichus. Bull. Amer. Mus. Nat. Hist, 97: 143-182. 

CRUICKSH.4NK, A. R. I. 1972. The proterosuchian thecodonts. In Joysey, 
K. A., and T. S. Kemp (Eds.), Studies in Vertebrate Evolution (es- 
says presented to Dr. F. R. Parrington, F. R. S.). Edinburgh: Oliver 
and Boyd, pp. 89-119. 

Ewer, R. F. 1965. The anatomy of the thecodont reptile Euparkeria 
capensis Broom. Phil. Trans. Roy. Soc. London, B 248: 379-435. 

Fraas, O. 1867. Dyoplax arenacciis, ein neuer Stuttgarter Keuper-Sau- 
rier. Jahresh. Ver. Naturk. Wiirttemberg, 23: 109-112. 

1877. Aetosaurus fcrratus Fr. Die gepanzerte Vogel-Echse 

aus dem Stubensandstein bei Stuttgart. Jahresh. Ver. Naturk. Wiirt- 
temberg, 33: 1-22. 

Gregory, J. T. 1953. Typoiliorax and Dcsmalosuchus. Postilla, Yale 
Peabody Mus., No. 16: 1-27. 

1962. The genera of phytosaurs. Amer. Jour. Sci., 260: 

652-690. 

, AND F. Westphal. 1969. Remarks on the phytosaur gen- 



era of the European Trias. Jour. Paleont., 43 : 1296—1298. 
GUTH, C. 1963. Au sujet de restes de reptiles de Madagascar. C. R. 

Acad. Sci. Paris, 256: 2661-2663. 
Harland, W. B. ct al. . 1967. The Fossil Record. A Symposium with 

Documentation. Geol. Soc. London. 827 pp. 
Haughton, S. H. 1915. A new type of thecodont from the Stormberg 

beds {Sphc7iosuchus acutus, gen. et sp. nov.). Ann S. Afr. Mus., 12: 

98-106. 
1924a. On a new type of thecodont from the Mid- 
dle Beaufort beds. Ann. Transvaal Mus., 11: 93-97. 
-_ 1924b. The fauna and stratigraphy of the Stormberg 



Series. Ann. S. Afr. Mus., 12: 323-497. 
Hoffman, A. C. 1965. On the discovery of a new thecodont from the 

Middle Beaufort beds. Navors. Nas. Mus. Bloemfontein, 2: 33-40. 
HoFFSTETTER, R. 195 5. Thecodontia. In Traite de Paleontologie, 5: 665- 

694. 
HuENE, F. V. 1911. Ueber Erythrosuclnis, Vertreter der neuen Reptil- 

Ordnung Pelycosimia. Geol. Palaeont., Abhandl., New Ser., 10: 1- 

60. 
1914a. Beitrage zur Geschichte der Archosaurier. Geol. 

Palaeont Abhandl., New Ser., 13: 1-53. 
1914b. Coelurosaurier-Reste aus dem unteren Muschel- 



kalk. Centralbl. Min. Geol. Pal., 1914: 670-672. 
1914c. Neue Beschreibung von Ctenosatirus aus dem 



Gottinger Buntsandstein. Centralbl. Min. Geol. Pal., 1914: 496-499. 
1920a. Bemerkungen zur Systematik und Stammesge- 



1972 THECODONT CLASSIFICATION 21 

schichte einiger Reptilieii. Z. Indukt. Ahstamm.-\"ererbungsl., 22: 209- 
212. 
1920b. Ein Parasuchier aus dem obeit-n Muschelkalk von 



Bayreiith. Senckenbergiana, 2: 143-145. 

1921. Neue Pseudosuchier und Coelurosaurier aus detii 



Wiirttembergischen Keuper. Acta Zool., 2: 329-403. 
1925. Die Bedeutung der Sp/irnosuc/ius CJruiipe fiir den 



Ursprung der Krokodile. Z. Indukt. Ah>tainm.-Vererbungsl., 38: 

307-320. 
1936. The constitution of the Thecodontia. Amer. Jour. 



Sci., Ser. 5. 32: 207-217. 
1938. Ein grosser Stagonolepide aus der jiingeren Trias 



Ostafrikas. Neues Jahrb. Min. Geol. Pal., Beilage-Bd., 80: 264-278. 
1939. Ein kleiner Pseudosuchier und ein Saurischier aus 



den ostafrikanischen .Mandaschichten. Neues Jahrb. Min. Geol. Pal., 
Abt. B, 81: 61-69. 
1940. Eine Reptilfauna aus der altesten Trias Nordruss- 



lands. Neues Jahrb. Min. Geol. Pal., Abt. B, 84: 1-23. 
1942. Die Fauna der Panchet-Schichten in Bengalen. 



Zentralbl. Min. Geol. Pal., Abt. B, 1942: 354-360. 

_. 1944. Die fossilen Reptilien des Siidamerikanischen 



Gondwanalandes. Ergebnisse der Sauriergrabungen in Siidbrasilien, 
1928/1929. Munchen: C. H. Beck'sche. 332 pp. 
1956. Palaontologie und Phylogenie der niederen Tetra- 



poden. Jena: G. Fischer. 716 pp. 
1958. Ein Pseudosuchier aus dem Buntsandstein des 



Schwarzwalds. Jahresh. Ver. Vaterland. Naturk. Wiirttemberg, 113: 
111-112. 
1959. Saurians in China and their relations. Vert. Pal- 



asiatica, 3 : 119-123. 
1960. Ein grosser Pseudosuchier aus der Orenburger 



Trias. Palaeontographica, Abt. A, 114: 105-111. 
1962. Die Pseudosuchier als Wurzelgruppe der meisten 



Landsaurier der Jura- und Kreidezeit. Neues Jahrb. Geol. Min. Pal., 

Monatsh., 1962: 1-6. 
Hughes, B. 1963. The earliest archosaurian reptiles. S. Afr. Jour. Sci., 

59: 221-241. 
J.\EKEL, O. 1910. Ueber einen neuen Belodonten aus dem Buntsandstein 

von Bernburg. Sitzungsber. Ges. Naturf. Freunde, Berlin, 19I(i: 197— 

229. 
Janeksch, W. 1949. Ein neues Reptil aus dem Keuper von Halberstadt. 

Neues Jahrb. Min. Geol. Pal., Monatsh., Abt. B, 1949: 225-242. 
Jepsen', G. L. 194S. A Triassic armored reptile from New Jersey. New 

Jersey Dept. Conserv. Misc. Geol. Pap., 1948: 5-20. 

KiTCHING, J. W., J. \V. COLLINSON, D. H. El.LIOT, .A\D E. H. CoLBERT. 

1972. Lystrosauriij Zone (Triassic) fauna from Antarctica. Science, 
175: 524-527. 



22 BREVIORA No. 395 

Krebs, B. 1965. Ticinosuc/ius fcrox nov. gen. nov. sp. Ein neuer Pseudo- 
suchier aus der Trias des Monte San Giorgio. Schweiz. Palaont. Abh., 
81: 1-140. 

1966. Ziir Deutung der C/iirot/itrium-FahTten. Natur und 

Museum, 96: 389-396. 

. 1969. Ctrnosaurisnis kornetil (v. Huene), die Pscudosuchia 



und die Buntsandstein-Reptilien. Eclog. Geol. Helvet., 62: 697-714. 
McGregor, J. H. 1906. The Phytosauria, with especial reference to Mys- 

triosuchus and Rhytidodon. Mem. Amer. Mus. Nat. Hist., 9: 29-101. 
Nash, D. 1968. A crocodile from the Upper Triassic of Lesotho. J. 

Zool. London, 156: 163-179. 
Newton, E. T. 1894. Reptiles from the Elgin sandstone.-Description 

of two new genera. Phil. Trans. Roy. Soc. London, B 185: 573-607. 
Ortl.am, D. 1967. Fossile Boden als Leithorizonte fiir die Gliederung 

des Hoheren Buntsandsteins im nordlichen Schwarzwald und sud- 

lichen Odenwald. Geol. Jb., 84: 485-589. 
Otschev, V. G. 1958. (New data on Russian pseudosuchians.) Doklady 

Akad. Nauk USSR, 123: 749-751. 
Peabody, F. E. 1952. Prtrolacosaurus kanscnsis Lane, a Pennsylvanian 

reptile from Kansas: Vertebrata, Article 1. Univ. Kansas Pal. Con- 

trib., 10: 1-41. 
Price, L. L 1946. Sobre um novo pseudosuquio do Triassico superior 

do Rio Grande do. Sul. Bol. Serv. Geol. Min. Brasil, 120: 7-38. 
Reig, O. a. 1959. Primeros datos descriptivos sobre nuevos reptiles 

arcosaurios del Triasico de Ischigualasto (San Juan, Argentina). 

Rev. Asoc. Geol. Argentina, 13: 257-270. 
. 1961. Acerca de la posicion sistematica de la familia 

Rauisuchidae y del genero Saurosuchus (Reptilia, Thecodontia). Publ. 

Mus. Munic. Cien. Nat. Trad. Mar del Plata, 1: 73-114. 
1963. La presencia de dinosaurios saurisquios en los "es- 



tratos de Ischigualasto" (Mesotriasico superior) de las provincias de 

San Juan y La Rioja (Republica Argentina). Ameghiniana, 3: 3-20. 

1967. Archosaurian reptiles: a new hypothesis on their 



origins. Science, 157: 565—568. 
1970. The Proterosuchia and the early evolution of the 



archosaurs; an essay about the origin of a major taxon. Bull. Mus. 

Comp. Zool., 139(5) : 229-292. 
RoMER, A. S. 1971a. The Chanares (Argentina) Triassic reptile fauna. 

VIIL A fragmentary skull of a large thecodont, Luperosuclius jractus. 

Breviora, Mus. Comp. Zool., No. 373: 1-8. 
1971b. The Chanares (Argentina) Triassic reptile fauna. 

X. Two new but incompletely known long-limbed pseudosuchians. 

Breviora, Mus. Comp. Zool., No. 378: 1-10. 

1971c. The Chanares (Argentina) Triassic reptile fauna. 



XL Two new long-snouted thecodonts, Chanaresuchus and Gualo- 
siichiis. Breviora, Mus. Comp. Zool., No. 379: 1-22. 



1972 THECODONT CLASSIFICATION 23 

_. 1971d. Unorthodoxies in reptilian phylogeny. Evolution, 



25: 103-112. 

_. 1972a. The Chanares (Argentina) Triassic reptile fauna 



XII. The postcranial skeleton of the thecodont CJiaiKursiK hits. liieviora, 
Mus. Comp. Zool., No. 385: 1-21. 

.. 1972b. The Chanares (Argentina) Triassic reptile fauna. 



XIII. An early ornithosuchid pseudosuchian, Gracilisuchus stipanici- 
corum gen. et sp. nov. Breviora, Mus. Comp. Zool., No. 3S9: 1—24. 
1972c. The Chafiares (Argentina) Triassic reptile fauna 



in 



XIV. Lcivisuc/ius admixtus, gen. et sp. nov., a further thecodont fro 

the Chanares beds. Breviora, Mus. Comp. Zool., No. 390: 1-13. 
ROZHDESTVENSKII, A. K. 1964. ( Thecodontia ) . In Osnovy Paleontologii, 

A. Orlov, ed. [Volume on Amphibians, Reptiles and Birds], pp. 497- 

506. 
RUSCONI, C. 1947a. Reptil Triasico de Uspallata. Bol. Paleont. Buenos 

Aires, No. 22: 1-2. 
1947b. Primeros hallazgos de coprolitos de reptil en el 

Triasico de El Challao, Mendoza. Cien. Invest., 1947, No. Id: 521- 

523. 

1951. Laberintodontes Triasicos y Permicos de Mendoza. 



Rev. Mus. Hist. Nat. Mendoza, 5: 33-158. 

Sawin, H. J. 1947. The pseudosuchian reptile Typothnrax mradri. jour. 
Paleont., 21: 201-238. 

Sharov, A. G. 1970. [A strange reptile from the Lower Triassic of 
Fergania.] Paleont. Zhurnal, Akad. Nauk, 1970(1): 127-13S. See 
also Nauka i Zhizn, 7: 28-32, 1971. 

1971. [New flying reptiles from the Mesozoic of Kazakh- 
stan and Kirghizia.] Trudy Paleont. Inst., Akad. Nauk, 130: 104— 
112. 

Sill, W. C. 1967. Proterochampsa barrionuevoi and the early evolu- 
tion of the Crocodilia. Bull. Mus. Comp. Zool., 135(8): 415-446. 

Simmons, D. J. 1965. The non-therapsid reptiles of the Lufeng basin, 
Yunnan, China. Fieldiana: Geol., 15: 1-93. 

SoERCEL, W. 1925. Die Fahrten der Chirotheria. Fine palaobiologische 
Studie. Jena: G. Fischer, vii + 92 pp. 

Tatarinov, L. p. 1960. [Discovery of pseudosuchians in the l^pper Per- 
mian of the USSR]. Pal. Zhurnal, 1960(4) : 74-80. 

1961. [Materials on the pseudosuchians of the USSR]. 

Pal. Zhurnal, 1961: 117-132. 

Van Hoepen, E. C. N. 1915. Contributions to the knowledge of the rep- 
tiles of the Karroo formation. 4. A new pseudosuchian from the 
Orange Free State. Ann. Transvaal Mus., 5: 83-87. 

Walker, A. D. 1961. Triassic reptiles from the Elgin area: Stui/nnolrpis, 
Dasygnathus and their allies. Phil. Trans. Roy. Soc. London, B 244: 
103-204. 

1964. Triassic reptiles from the Elgin area: Ornillio- 



24 BREVIORA No. 395 

sticliui and the origin of carnosaurs. Phil. Trans. Roy. Soc. London, 
B 248: 53-134. 
. 1966. Elachistosuchus, a Triassic rhynchocephaiian from 



Germany. Nature (London), 211: 583-585. 
1968. Protosuclius, Protrrocliampsa, and the origin of 



phytosaurs and crocodiles. Geol. Mag., 105: 1-14. 
. 1970. A revision of the Jurassic reptile Hallopus 'Vic- 



tor (Marsh), with remarks on the classification of crocodiles. Phil. 

Trans. Roy. Soc. London. B 257: 323-372. 
Welles, S. P. 1947. Vertebrates from the Upper Moenkopi foriratlon of 

northern Arizona. Univ. Calif. Publ., Bull. Dept. Geol. Sci., 27(7): 

241-294. 
Westphal, F. 1963. Phytosaurier-Gattungen und -Arten aus dem siid- 

westdeutschen Keuper (Reptilia, Thecodontia). Neues Jahrb. Geol. 

Palaont. Abh., 118: 159-176. 
1970. Phytosaurier-Hautplatten aus der Trias von Mad- 

agaskar — ein Beitrag zur Gondwana-Palaogeographie. Neues Jahrb. 

Geol. Palaont. Mh., 1970, H. 10: 632-638. 
Woodward, A. S. 1907. On a new dinosaurian reptile (SclcromocIAus 

taylori gen. et sp. nov.) from the Trias of Lossiemouth, Elgin. Quart. 

Jour. Geol. Soc. London, 63: 140-144. 
Young, C. C. 1936. On a new Chasuiatosauitis from Sinkiang. Bull. 

Geol. Soc. China, 15: 291-311. 
1944. On a supposed new pseudosuchian from Upper 

Triassic saurischian bearing red beds of Lufeng, Yunnan, China. 

Amer. Mus. Novitates, No. 1264: 1-4. 
1951. The Lufeng saurischian fauna in China. Palaeont. 



Sinica, New Ser. C, No. 13: 19-96. 
1958. On the occurrence of Chasmatosaurus from Wuh- 



siang, Shansi. Vert. Palasiatica, 2: 259-262. 
1963. Additional remains of Chasmatosaurus yuani 



Young from Sinkiang, China. Vert. Palasiatica, 7: 220-222. 
1964. The pseudosuchians in China. Palaecnt. Sinica, 



New Ser. C, No. 19: 106-205. 



MAf? 1 8 1985 

'HARVARD 



B R E V I 0"W''% 



^loseom of Conipsiratave Zoology 

Cambridge, Mass. November 6, 1972 Number 396 

THE GHANARES (ARGENTINA) 

TRIASSIG REPTILE FAUNA. 

XVII. THE GHANARES GOMPHODONTS 

Alfred Sherwood Romer 

Abstract. Much of the gomphodont material in the Chanares collec- 
tions appears to pertain to two closely related but distinct species, I\Ias- 
setOfftiai/ziis pascual'i and Massctognathus tcruggii. An especially large 
skull is described as Massctognathus major, sp. nov. Two skulls represent 
a distinct form, Mcgagompliodon oligodcns, gen. et sp. nov., differing mainly 
in its smaller but more numerous cheek teeth. 

As noted previously, a large proportion of the reptilian remain.s 
from the Chanares beds pertain to the peculiar gomphodont 
side branch of the Cynodontia, abundant in all Middle Trias.sic 
terrestrial faunas. /Although much of the 1964-65 collection 
from these beds was, because of political complications, long 
delayed in shipment, the one box that reached the laboratory 
promptly contained a number of gomphodont skulls, on the 
basis of which I described two species as Massetognathus pas- 
cuali and Massetognathus teruggii (Romer, 1967). The skull 
materials then available sorted out clearly into two size groups, 
differing in skull length by about 40 percent. This figure was 
too great to be due to .sex differences, and since there were in 
this sample of the collection no intermediates in size between 
the two groups, erection of two species seemed fully justified. 

Today, with the full collection axailable, a much larger array 
of gomphodont skull material lies before me, and a re-study of 
the situation is called for. In many instances in vertebrate pale- 
ontology, specific diagnoses have been made on .supposed size 
differences when only a few specimens were known, only to be 
pro^•ed invalid when more abundant material became avail- 
able. Might that not be the case here? Two large specimens 
pro\e to be of a distinct type (descril^ed below) but mo.st ap- 
pear to pertain to Massetognathus and show a wide \'ariation in 



2 BREVIORA No. 396 

size. They are here tabulated according to skull length meas- 
ured to the condyles. Owing to imperfections in the material or 
incomplete preparation, this measurement was axailable onlv in 
a fraction of the cranial specimens, but in a number of other 
cases this basal length could be calculated fairly accurately 
from other measurements. 

Table I. Massetognathus cranial specimens grouped by 

BASAL length, IN MM. 

61-70 mm 1 

71-80 mm 11 

81- 90 mm 16 

91-100 mm 13 

101-110 mm 8 

111-120 mm 6 

121-130 mm 4 

131-140 mm 5 

141-150 mm 2 

151-160 mm 1 

161-170 mm 4 

200-210 mm 1 

In any population of living reptiles or any adequate sample 
of a fossil form, the size distribution is a characteristic one; the 
great proportion of the specimens represent young adults, but 
in addition there are present a few forms of somewhat greater 
size, presumably older individuals in which further growth had 
occurred. To some degree our distribution is of the type that 
suggests a single species, with the greater part of the specimens 
concentrated in the size range attributed in my earlier paper to 
M. pascuali, the holotype skull of which measured 87 mm in 
length. But in two regards the collection does not agree with 
the assumption that we are dealing with a single species. ( 1 ) In 
a typical one-species population only a very few "elderly" in- 
dividuals are present far above the "young adult" size; here 
nearly half of the specimens extend onward toward sizes far 
exceeding that of the young adults. (2) Even excluding the 
single extra-large skull tabulated, these larger specimens run 
upward to a skull size about double that of "young adults" of 
M. pascuali — a situation quite out of the range of possibility 
of size increase in any known reptile population. It seems cer- 
tain that in these larger specimens we are dealing with repre- 
sentatives of a second species, M. teruggii, less abundant than 



1972 CHANARES GOMPHODONTS 3 

M. pascuali, but of larger size — the holotype skull of M. teruggii 
measurino^ 125 mm in basal length. 

Confirmation of this conclusion is given b\ a consideration of 
the "molar" dentition. In many cases the lower jaws arc firmly 
occluded with the upper, so that (in default of difficult prep- 
aration) the tooth surfaces are not clearly seen. Howc\cr, the 
"molars" are \isible in surface view in a number of specimens 
of both the smaller, M. pascuali, and of the larger, M. teruggii, 
types. In the M. pascuali specimens the t\pical "molars" have 
an anteroposterior width which a\"erages close to 3 mm; those 
of M. teruggii average 3.75 mm. Further, Mr. John Hillman, 
who has studied gomphodont dentitions intensi\-ely, pointed out 
to me that in the teruggii specimens the "molars" are also 
proportionately broader mediolaterally than in the smaller M. 
pascuali specimens. 

It is highly improbable that this increase in indi\-idual molar 
size could occur during the lifetime of an indixidual. There is 
almost no exidence of any \ertical replacement of these very 
deep-rooted teeth in the adult. There is e\-idence here, as in 
other gomphodonts, of a trend for suppression of one or two of 
the smaller anterior "molars" during the hfetime of an individ- 
ual, and for the addition of one or more teeth at the posterior 
end of the series. But there is no indication of de\elopment of 
the complex type of tooth replacement found in the manatee, 
by which a whole series of "molars" might be replaced by larger 
successors pushing forward from the back end of the series. If 
such replacement were to occur, we would expect the new teeth 
added at the back of the series to increase in size. For the most 
part the "molars" in the back part of the series are larger than 
those anteriorly placed. But Mr. Hillman, who has made careful 
measurements of the Massetognathus dentitions, tells me that 
in a number of instances the last one or two teeth in the series 
are smaller, rather than larger, than those anterior to them. 

Apart from the two species of Massetognathus discussed above, 
the Chafiares gomphodont series certainly includes other vari- 
ants. As discussed below, two specimens appear to represent a 
form generically distinct from Massetognathus. And in addition, 
the specimens assigned to that genus are quite surely not all as- 
signable to M. pascuali and M. teruggii. As mentioned above, 
the M. teruggii type has a skull length of 125 mm. \\'e would 
reasonably expect a number of older indi\iduals to exceed this 
figure to a modest degree, up to about 160 mm or so — -that is, 



BREVIORA 



No. 396 




Figure 1. The holotype skull of Massrtoc/iiat/ius major in dorsal view. 
X 2/5. 



to a skull length of as much as 30 percent in excess of a "young 
adult." But the finding of four specimens in the 160-170 mm 
bracket is disturbing, and a skull of over 200 mm in length — 
two-thirds again as large as the type- — ^ gives us an impossible 
situation. We have here, quite certainly, a third, large species. 

Massetognathus major sp. no v. 

Holotype: La Plata Museum, No. 64-XI-14-15, (field no. 
55 ) . From the Chafiares formation, about 4 km southeast of 
the mouth of the Rio Chaiiares, La Rioja Province, Argentina. 

Diagyiosis. Generally comparable to other species of Mas- 
setognathus, but orbits extend relatively far forward, antorbital 
region narrower than in other species; cheek tooth rows less 
divergent posteriorly; size large, the holotype with a basilar 
skull length of 205 mm. 

The species is based primarily on a single skull (Figs. 1, 2) 
far larger than any other assigned to this genus. The specimen 



1972 



CHANARES GOMPHODONTS 









Figure 2. The holotype skull of Massrtngnaihus major in ventral view. 
X 2/5. 



was found exposed ^\•ith the palatal surfaces upward, and in a 
somewhat weathered condition, so that the cheek teeth do not 
show the crown pattern well, and the posterior part of the skull 
is imperfectly preserved. In most regards the skull agrees well 
with the previously described species of Alassetognathus. Dis- 
tinctive, howe\er, is the relati\'e narrowness of the snout and 
a consequent!)- lesser de\elopment of the broad shelf which, in 
ventral view, extends far out on either side of the cheek tooth 
series. The series of cheek teeth are but little cur^•ed, and diverge 
but little posteriorly. The orbits, instead of being essentially sub- 
circular in outline, extend forward in triangular fashion, with 
the apex of the triangle lying at the entrance to the lacrimal 
duct. In relation, presumably, to large size, the sagittal crest is 
well developed, the ridges bounding the temporal openings fus- 
ing medially a short distance back of the postorbital bar, with 
complete obliteration of the parietal foramen. Thirteen cheek 
teeth are present on either maxilla; the most anterior are rela- 



BREVIORA 



No. 396 




Figure 3. The skull of Megagomphodon oligodcns in dorsal view. This 
figure and figs. 4—6 are composites, based on the holotype and MCZ 4138. 
X 2/5. 



lively smaller in size, as compared with those farther back, than 
is the case in either M. pascuali or M. teruggii. 

Megagomphodon oligodens gen. et sp, nov. 

Holotype: La Plata Museum No. 64-XI-14-16, (field no. 
65 ) . Chanares formation. La Rioja Province, about 6 km ENE 
of the mouth of the Rio Chaiiares. 

Diagnosis. A relatively large traversodontid gomphodont, 
with a basal skull length on the order of 180 mm. Cheek teeth 
relatively small, especially anteriorly, and about 17-18 in num- 
ber. Skull relatively slender, the width across the orbital region 
being but about two thirds the total skull length. 

In the collection two skulls, the holotype and MCZ 4138, rep- 
resent a gomphodont type clearly distinct from Massetognathus. 
Incom.plete jaws, but no postcranial materials, are associated with 
both. Neither is too well preserved; the holotype has fairly well 
preserved cheek teeth, but is imperfect posteriorly; the posterior 
part of the braincase is preserv^ed in MCZ 4138, but teeth are 
represented only by their roots. My description is based on a 
combination of features present in one skull or the other, and 
my illustrations (Figs. 3-5) are hkewise composite. On neither 
skull are the sutures well shown, and I have in consequence 
omitted most of them in mv figures. 



1972 



CHANARES GOMPHODONTS 




Figure 4. The skull of Megagnmphodon oligoJitis in lateral view. 
X 2/5. 




Figure 5. The skull of Megagomphodon oligodcns in ventral view. 
X 2/5. 




Figure 6. The dentary of 'Megagomphodon oligodcns in lateral view. 
X 2/5. 



8 



BREVIORA 



No. 396 




c^^^ 




Figure 7. Left upper and lower cheek teeth of Mcgagomphodon ol'igo- 
dens. X 3/2. 



The skull is somewht more slender than in Aiassetognathus. 
The "muzzle" is notably narrow, with a constriction back of the 
canine region, and expansion in width does not take place until 
well back toward the subcircular orbits. In correlation with 
large size, the sagittal crest is well dexeloped, with the two crests 
becoming closely apposed not far back of the le\el of the post- 
orbital bar. 

Dentaries are preserved in both specimens (Fig. 6). The 
"angular" region is well developed, and in MCZ 4138 has a 
backwardly pointed tip. As in all traversodonts the ascending 
ramus is highly developed, and extends far back dorsally. In 
most specimens of Chaiiares gomphodonts in which the bone is 
well preserved, the posterior end of the ramus is rounded; in 
both specimens of the present form this process is sharp-tipped 
posteriorly. On the inner surface of the dentary there is a longi- 
tudinal recess, typical of advanced cynodonts, for the reception 
of the supporting bar formed by surangular + angular + pre- 
articular. 

The most distinctive feature of this genus is the nature of the 
cheek teeth (Figs. 5, 7). These are numerous, and in the holo- 
type, where the dentition is nearly completely preser\ed, there 
appear to be 17 or 18 "molars" in both upper and lower jaws. 
This is, of course, a definitely higher count than in Masseto- 
gnathus. On the other hand, the individual teeth are definitely 



1972 CHANARES GOMPHODOXTS 9 

smaller than in that genus. The length of the entire row of 
upper cheek teeth is about a third of the skull length in this 
form and in Massetognathus as well; but since the number of 
teeth in Megagoynphodon is greater, the anteroposterior dimen- 
sions of indi\idual teeth is relati\ely small; the a\erage antero- 
posterior length of an indi\idual tooth in Massetognathus is 
about 3 percent of the skull length, in Megagoynphodon only 
about 2 percent. The Megagoynphodon teeth are also relatively 
small in transverse measurement; the broadest teeth in this genus 
measure only about 4 percent of skull length, whereas this width 
in Massetognathus is approximately 5 percent. 

Except for the reduced tooth size, Megagoyyiphodon is ob- 
\-iouslv not distantly related to Massetognathus and this genus 
mav perhaps ha\"e been derived from such a form as M. yyiajor. 

Collection and preparation of this material was made possible 
bv successive grants from the National Science Foundation. 

REFERENCE CITED 

RoMER, A. S. 1967. The Chanares (Argentina) Triassic reptile fauna. 
III. Two new gomphodonts, Massetognathus pascuali and .1/. tcruggVi. 
Breviora, Mus. Comp. Zool., Xo. 264: 1—25. 



ML'S. CCMP. ZOOL 
LIRRARY 

MAR 1 8 1965 

B R E V I Q.Ja,A 

l^Ioseiim of Comparative Zooilogy 



Cambridge, Mass. November 6, 1972 Number 397 

XYLOREDO, A NEW TEREDINID-LIKE 

ABYSSAL WOOD-BORER 

(MOLLUSCA, PHOLADIDAE, XYLOPHAGAINAE) 

Ruth D. Turner 

Abstract. Xylorrdo, a new genus of Xylophagainae characterized by 
having a long burrow with a calcareous lining, is described. Three new 
species belonging to this genus are also described, Xylorrdo nooi and X. 
ingolfia from the Atlantic and A', naccli from the Pacific. The systematic 
position of the genus, its superficial resemblance to the Teredinidae, and 
the factors controlling the distribution of the species are discussed. 

INTRODUCTION 

The Xylophagainae are benthic, mainly deep-sea wood-borers. 
Their distribution is dependent on the presence of woody plant 
material and their dispersal on the free-swimming lar\al stage. 
From available e\idence it appears that wood in the deep sea is 
scarce and its distribution patchy. Bruun (1959) reported on 
the remnants of plant material collected by the R/\' GALA- 
THEA during a circumna\igation of the globe. J. Knudsen 
(1961) reported on the Xylophaga collected on that cruise and 
also discussed the presence of plant debris in the deep sea, sum- 
marizing the results of earlier expeditions. An analysis of the 
data in these reports shows that 56 of the 180 successful bottom 
stations made by the GALATHEA in depths between 400 and 
7290 meters yielded plant material. Only 22 of these hauls had 
woody material of sufficient size (e.g., over 10 grams) to sup- 
port borers, but species of Xylophaga were taken at only 1 1 
stations. 

On the basis of material now available, the ranges of many 
of the species in this subfamily appear to be limited. The known 
ranges may be a reflection of the limited number of dredging 
stations but may also be because 1 ) the dispersal efficiency of 
the larvae is poor J. Knudsen, 1961) and 2) wood in the deep 
sea is not only scarce but is found in "islands" of plant debris, 



2 BREVIORA No. 397 

usually in deep trenches near land, particularly in the regions of 
large rivers (Bruun, 1959). In order to better understand the 
distribution of these borers, we need to know more about their 
life history, particularly the length of time the larv^ae spend in 
the plankton and their behavior at the time of settlement. 

Most Xylophaga are obtained by dredging, though a few 
species occurring in high latitudes can be found in shallow water, 
some even just subtidallv in old \vrecks or piling (Tomlin, 1920; 
Dons, 1929a, 1929b, 1940). In 1961 H. Turner attached a 
series of wood panels to the mooring line of a buoy set out 
by the Woods Hole Oceanographic Institution at 39°30'N; 
69°40'W (on the Gay Head, Martha\s Vineyard-Bermuda 
transect) in 3000 meters. When the panels were retrieved in 
October 1961, the bottom one was found to have been heavily 
attacked by Xylophaga (approximately 100 per square centi- 
meter). This was the first time that "new" wood had been 
submerged to such a depth sufficiently rapidly so that there was 
no possibility of the larvae settling on the way down. Unfor- 
tunately, the specimens were too young for positive identification 
and no additional panels have been exposed at this site. About 
the same time the United States Navy began testing the per- 
formance of various materials, including wood, in the deep sea 
(DePalma, 1969; Muraoka, 1964-1967). These tests at depths 
of 1000 to 2000 meters have added greatly to our knowledge 
of the Xylophagainae and resulted in the discovery of a new 
genus, represented by three new species. Species in this new 
genus {Xyloredo, p. 3) are remarkably close to the teredinids 
in superficial appearance, as they produce long, often tortuous 
tunnels which have a calcareous lining; the valves, as in all 
Xylophagainae, are teredinid-like. The lack of pallets for clos- 
ing the burrow and of apophyses for the attachment of the foot 
muscle, and the presence of a mesoplax, definitely place them in 
the Pholadidae, subfamily Xylophagainae. 

Though mentioned in the hterature (R. Turner, 1966a, 
1969), this new genus was not named or described. Specimens 
of Xyloredo nooi (p. 5), the first new species to be received, 
were taken from tests put out by the U. S. Na\'al Oceanographic 
Office in the Tongue of the Ocean, Bahama Islands, in 1737 
meters (DePalma, 1969). At this site, panels set in contact 
with the bottom were completely riddled, while those 15 meters 
ofT the bottom were entirely free of borers. Wood panels sub- 
merged by the U. S. Naval Civil Engineering Laboratory south 
of San Miguel Island off Port Hueneme, California, in 2072.6 



1972 XYLOREDO 3 

meters were the source of the second new species, Xyloredo 
naceli (p. 9). At this site, panels 15 centimeters off the bot- 
tom were much more se\erely attacked b\' two species of xylo- 
phagaids than those only one meter up on the rack (Muraoka, 
1966b). Tipper (1968), working with Xylophaga washingtona 
Bartsch off the Oregon coast, found that the number of borer 
penetrations dropped most markedly within the first 6 centi- 
meters abo\e the bottom. These findings suggest that the larxae 
probably remain close to the bottom throughout their free- 
swimming stage and that, except in areas where there are strong 
bottom currents, thev probablv are not widelv dispersed (R. 
Turner, 1966b, 1968). 

Xyloredo ifigolfia (p. 7), the third new species belonging 
to this genus, came from a piece of wood dredged in 1896 by 
the INGOLF Expedition south of Eyrabakki, Iceland, from 
1 783 meters. This species appears to be more closely related to 
the California species than to the one from the Tongue of the 
Ocean. 

As with many species of molluscan wood-borers, it is diffi- 
cult to determine which characters of the species of Xyloredo 
will pro\'e most stable and useful for taxonomic purposes until 
large series from several localities, representing all growth stages, 
are a\ailable for study. As intimated above, wood-borers from 
great depths are difficult to obtain, and such series may not be 
a\'ailable for many years. In fact, the publication of the taxa 
included in this report has been delayed for five years while ef- 
forts were made to obtain additional material. Hopefully, 
knowledge of the existence of these remarkable Xylophagainae 
will stimulate personnel on dredging cruises and deep submers- 
ible dives to collect and preserve woody plant material. In 
addition to borers, dredged wood is a rich source of many spe- 
cies of benthic invertebrates. 

Xyloredo^, new genus 

Type species. Xyloredo nooi Turner, new species. 

Description. X^ahes t\pical of subfamih Xylophagainae and 
virtually indistinguishable from valves of Xylophaga; apophyses 
lacking; chondrophore, internal ligament, umbonal-ventral sul- 
cus and ridge, and ventral condyle well developed. Mesoplax 

'A C(;mpouiid word, combining Xylu- from XylophcKja and -redo from 
Teredo, indicating the teredinid-like appearance of these Xylophagainae. 



4 BREVIORA No. 397 

small, flat, triangular and only lightly calcified. Burrow, de- 
pending on size and age of specimen, 5 to 30 times length of 
valves. Posterior two-thirds of burrow with calcareous lining 
marked with distinct rings and covered by thin, outer periostra- 
cal layer which extends anteriorly as border (Plate 1, fig. 2). 
Periostracal covering of vaKes extending posteriorly as sheath 
that is continuous with heavy periostracal band on anterior end 
of tube. Mantle fold attached to anterior end of tube, point of 
addition of both periostracum and calcium. Burrow lining in 
very young specimens either entirely periostracal or lightly 
calcified. 

Combined incurrent and excurrent canals extend length of 
tube and attach to calcareous lining at anterior end of siphons. 
Siphons short, separate and apparently margined with few 
cirri. 

Visceral mass similar to Xylophaga (Purchon, 1941; Turner 
and Johnson, 1971), contained completely between valves. 
Ctenidium of sinsle demibranch nearly coNcring sides of vis- 
ceral mass. Mouth broad and slitlike; labial palps inconspic- 
uous. Esophagus short, wide, and flattened; stomach large, 
subglobular, with large crystalline style sac opening into it at 
posterior end. Style sac bent forward so that lower end pro- 
trudes slightly through \isceral mass postero\entral to foot. 
Caecum large, U-shaped, and located mainly on the right side. 

Remarks. The general description of the anatomy of Xylo- 
redo given above was derived from a study of the remains of 
specimens of all three species. Unfortunately none was suffi- 
ciently well preserxed for detailed dis.section or sectioning. 

Xyloredo differs from Xylophaga by having : 1 ) a long, tere- 
dinid-like burrow with a calcareous lining, 2) a thin periostracal 
sheath extending from the valves to the heavier periostracal 
border of the tube, and 3) extended incurrent and excurrent 
canals. The siphons of Xyloredo are short and separate, and 
when the animal is remo\'ed from its burrow it resembles a 
shipworm without its pallets. 

Unlike the Teredinidae, Xyloredo has short, broad ctenidia, 
and all of the visceral mass is contained between the valves; a 
mesoplax is present but pallets and apophyses are lacking. In 
addition, the calcareous tube of Xyloredo is thin, regularly 
ringed, and covered externally with a shiny periostracum. 

The superficial resemblance of Xyloredo to the Teredinidae 
is a result of convergence. The Xylophagainae are probably 
most closely related to the Jouannetinae, a subfamily of rock- 



1972 XYLOREDO 5 

boring pholads, which also lack apophyses and have the pedal 
retractor muscle inserted in the typical bi\'al\'e position anterior 
to the posterior adductor muscle. 

The disco\"ery of this teredinid-like genus of Pholadidae neces- 
sitates a reexamination of fossil tcredinids, especially those with 
ringed tubes. If \al\es are not present, or if the inner surface 
of the valves cannot be examined to ascertain whether apophy- 
ses are present, it may be impossible to distinguish fossil Xyloredo 
from the Teredinidae. It is, of course, possible that all fossil 
tubes with regular rings are Xyloredo or some closely related 
fossil genus. Many teredinid tubes are marked with rings, but 
not in the regular, ex'enly spaced fashion found in Xyloredo. 

Fossil deposits produced in the deep sea are \irtually un- 
known, so perhaps there is no need to be concerned about fossil 
Xyloredo. Howexer, like Keopilina, it might well be that at one 
time their ancestors li\ed in shallower waters. It is possible that, 
as a result of competition with the more efficient Teredinidae, 
only the deep-water species sur\'ive. 

Range. Species of Xyloredo are known in the Atlantic 
Ocean from off Iceland and from the Tongue of the Ocean, 
Bahama Islands. In the eastern Pacific they are known only 
from ofT San Miguel Island, near Port Hueneme, California. 
Xyloredo may well be world-wide in distribution in depths of 
over 1500 meters, but further dredging and testing are needed 
to proxe this. 

Xyloredo nooi^, new species 
Plates 1 and 2 

Type locality. From test panels submerged in the Tongue 
of the Ocean ('25°54'N; 77°49'W), off the north end of Andros 
Island, Bahama Islands, in 1737 meters, from 4 April 1962 to 
17 February 1965 i DePalma, 1969). 

Types. Holotype, Museum of Comparative Zoology 279631. 
Paratypes from the same and other panels exposed at the same 
locality. Museum of Comparati\e Zoology 279632, 279633, 
279634, 279635, and the Universitetets Zoologiske Museum, 
Copenhagen. 

Description. Shell globose, valves reaching 10 mm in length 
and 10.5 mm in height, thin, fragile: umbos prominent. Perio- 

'An acronym based on the initials of the Navy Oceanographic Office, 
Washington, D.C., which was responsible for the tests in the Tongue of 
the Ocean, Bahama Islands, from which the specimens were obtained. 



6 BREVIORA No. 397 

stracum relati\ely thick, golden brown, glistening and covering 
entire valve. Beaked portion of anterior slope sculptured with 
numerous, finely denticulated ridges (24 ridges on holotype). 
Posterior portion of anterior slope narrow; ridges closely spaced, 
coarsely denticulated, usually thickened posteriorly and extend- 
ing only about one-half distance to shallow umbonal-ventral 
sulcus. Disc sculptured with fine growth lines only. Posterior 
slope high, reflected near dorsal margin and sculptured with fine 
growth lines. Umbonal reflection thin, wide and adhering closely 
to surface of valves posteriorly, free anteriorly. 

Inner surface of valves smooth and glistening. Umbonal-ven- 
tral ridge narrow-, high, slightly to strongh segmented and not 
greatly enlarged at ventral condyle. Chondrophore and inter- 
nal ligament prominent (Plate 2, fig. 5). Disc separated from 
posterior slope by pronounced groove extending from umbo to 
posterior ventral margin. Posterior adductor muscle scar large, 
elliptical, and divided; upper part (on reflected portion of pos- 
terior slope) with irregular impressions, lower part with chevron- 
shaped impressions. Anterior adductor muscle scar covering 
umbonal reflection. Siphonal retractor muscle scars not im- 
pressed. Pedal retractor muscle scar elongate, located about 
midway on anterior margin of posterior adductor muscle scar. 
Mesoplax small, consisting of two flat, broad, subtriangular 
plates lying on dorsal surface of anterior adductor muscle and 
composed largeh- of periostracum. 

Burrow long, teredo-like, with thin, calcareous lining for 
about three-fourths its length. Tubular lining marked with dis- 
tinct rings and covered with thin, yellow-brown periostracum 
which extends as border anteriorly ( Plate 1 , fig. 2 ) . Portion of 
animal between valves and tube covered by smooth, golden- 
brown, periostracal sheath, continuous anteriorly with perio- 
stracum of \-al\es and posteriorly with periostracal border of 
tube. Tubes of young specimens (up to 10 mm long) may be 
entirely periostracal. Burrow opening on surface of wood with 
white, slighth raised, often divided cone about 1 mm in diam- 
eter. Siphons short and separate. Protoconch unknown. 

Valve measureynents. 



Length 


Height 




9.5 mm 


10.0 mm 


Holotype 


9.5 


9.8 


Paratype 


7.2 


7.8 


// 


6.7 


6.9 


// 


5.0 


5.1 


// 



1972 XYLOREDO 7 

Remarks. This species is closely related to both Xyloredo 
naceli and A', ingolfia (see also Remarks under ingolfia). It 
differs from them in ha\-ing- a much thinner burrow linins:, a 
high, reflected posterior slope on the \al\es, and a proportion- 
ately smaller, more highlv placed and di\'ided posterior adduc- 
tor muscle scar. In addition, the periostracal sheath extending 
between the \al\cs and the calcareous tube is smooth. On the 
basis of the material at hand, nooi appears to be a much larger 
species. Size in borers, howe\er, is often not a reliable taxo- 
nomic character, and an understanding of the size range re- 
quires the examination of large series from \ aricd substrata. 
Stenomorphic adults often result from o\er-crowding or from 
penetration of an unusually hard substance. 

The panels from Avhich specimens of nooi were remo\ed had 
been in the w^ater for o\-er 2/2 years and were so badly rid- 
dled that the white, calcareous tubes showed through the thin 
surface. The general appearance of the wood was the same 
as that of a panel riddled with teredinids. It was not until the 
borers were exposed and examined closely that their pholad af- 
finities were recognized. Many of the specimens were dead and 
their burrows filled with mud, with only the calcareous tube 
and \'ah'es remaining. 

Unfortunately, the panels were not preser\ed immediately 
upon remo\al from the water, so the borers were in poor con- 
dition. Therefore, little anatomical work could be done, though 
it was possible to determine that the visceral mass was contained 
between the \ahes and that there was a large, U-shaped, wood- 
storing caecum. 

The bottom at the test site was a grav mud; the temperature 
was 4°C and the salinity 35 7oo (DePalma, 1969). 

Range and specimens exaynined. Known onh' from the t)pe 
locality. 

Xyloredo ingolfia\ new species 
Plates 3-5 

Type locality. From wood dredged by the INGOLF Expe- 
dition at station 67, south of Evrabakki, Iceland (61°30'N; 
22°30'W), in 975 fathoms [1783 meters]. 

Types. Holotype, Museum of Comparati\-e Zoology 279636. 
Paratypes from the same locality. Museum of Comparati\e 

^Named for the Danish IN'GOLF Expedition, which collected the wood 
from which the specimens were obtained. 



8 BREVIORA No. 397 

Zoology 279637, and the Universitetets Zoologiske Museum, 
Copenhagen. 

Description. Shell globose, val\-es reaching 2.5 mm in length 
and 2.0 mm in height, thin, fragile, with prominent umbos; 
thin, glistening, almost colorless periostracum covering disc and 
posterior slope. Beaked portion of anterior slope wide, extend- 
ing more than one-half distance to ventral margin; sculptured 
with close-set and very finely denticulated ridges. Posterior por- 
tion of anterior slope about two-thirds width of beak, sculptured 
with close-set ridges which extend to very slightly impressed 
umbonal-ventral sulcus. Disc sculptured with well-marked 
growth lines. Posterior slope small, low and not clearly demar- 
cated on outer surface of valve. Umbonal reflection thick, nar- 
row, short and free except at posterior end. 

Inner surface of N'alves smooth, slightly shiny to chalky (per- 
haps owing to long preservation ) . Umbonal-xentral ridge wide, 
flattened, often varying in width, irregularly segmented and not 
enlarged at \entral condyle. Chondrophore and internal liga- 
ment large. Disc not clearly separated from posterior slope. 
Posterior adductor muscle scar large, slightly raised, elliptical, 
extending nearly to ventral margin, with irregular, transxerse 
impressions. Anterior adductor muscle scar covering umbonal 
reflection. Siphonal retractor muscle scars not impressed. Pedal 
retractor muscle scar small, elongate to oval and located just an- 
terior to posterior adductor muscle scar. Mesoplax of two very 
small, narrow, subrectangular, flat, calcified plates lying on 
dorsal surface of anterior adductor muscle. 

Burrow 10 to 15 times length of valves; calcareous tubular 
lining three-fourths length of burrow. Tube relatively heavy, 
marked with uniform, close-set, raised rings, and covered with 
light tan periostracum which extends anteriorly as border. Por- 
tion of animal between vahes and tube co\'ered by thin, irregu- 
larly ridged periostracal sheath. Burrow opening often with 
small, white cone about 0.5 mm in diameter. Siphons short; 
incurrent siphon slightly longer than excurrent. Protoconch large, 
medium golden-brown and sculptured with fine, concentric 



ridges. 












Valve yneasurcments. 


Len 


gth 


Hei 


ght 






2.5 


mm 


2.3 


mm 


Holotype 




2.5 




2.0 




Paratype 




2.1 




2.0 




// 




2.0 




1.9 




ft 




1.5 




1.4 




It 




1.5 




1.2 




" 



1972 XYLOREDO 9 

Remarks. This species is most closely related to Xyloredo 
nacdi from the eastern Pacific. It difTers in havins^ a less well- 
de\'eloped posterior slope, a shallow, indistinct umbonal-\'entral 
groo\e, a flattened umbonal-\entraI ridge, and in having the 
valves longer than high. In addition, the periostracal sheath 
posterior to the valves is not pustulose as in naceli, and the cal- 
careous tube is much heavier and has raised rings. It also differs 
from naceli in having a much larger protoconch, which is sculp- 
tured with fine, concentric ridges. 

It differs from A', nooi in ha\'ing the valves longer than high, 
in haxing a low, rounded posterior slope, in lacking the distinct 
groove on the inner surface separating the disc from the poste- 
rior slope, and in ha\ing the plates of the mesoplax subrectangu- 
lar and well calcified. 

The wood dredged b) the INGOLF Expedition had been 
hea\'ily attacked by large teredinids before sinking to 1783 
meters. These were all dead, only the empty tubes remaining as 
evidence of their presence. Filling every available space be- 
tween the teredinid burrows were pipe-organ-like tubes of Xylo- 
redo going both across and with the grain of the wood. Many 
of these tubes were 40 mm long and 1.2 mm in diameter. Jen- 
sen (1912: 56), in discussing the distribution of Idas [= 
Idasola] argenteus Jeffreys, states: "The specimens of the IN- 
GOLF Expedition were taken south of Iceland (61°30'N.L.; 
22°30'W.L.) at 975 fathoms, under similar conditions to the 
last; the trawl brought up two large pieces of pine-wood, which 
had been pierced through and through by Teredo; in some of 
the Teredo tunnels were in addition mud and worm-tubes, fur- 
ther worms and small bivalves, namely Idas argenteus.'^ This is 
the same piece of wood, remnants of which were preserved in 
the Copenhagen Museum, from which specimens of Xyloredo 
ingolfia were taken. The small, straight tubes were probably 
thoue;ht to be vouns; teredinids. 

The bottom at station 67 was a brown-gray transition clay 
of a fine, clayey-sandy consistency (Boeggild, 1900); the tem- 
perature was 3°C and the salinitv 35.18 "/oo (M. Knudscn, 
1899). 

Range and specimens examined. Known only from the type 
locality. 

Xyloredo naceli\ new species 
Plate 6 



^^n acronym based on the initials of the Naval Civil Engineering Lab- 
oratory, Port Hueneme, California, which was responsible for the tests 
from which the specimens were obtained. 



10 BREVIORA No. 397 

Type locality. From the U. S. Naval Civil Engineerina; Lab- 
oratory Submersible Test Unit at test site I [STU 1-4], about 
30 miles south of San Miguel Island, off Port Hueneme, Cali- 
fornia (33°46'N; 120°46'W), in 6800 feet [2072.6 meters], sub- 
merged from June 1964 to July 1965. 

Types. Holotype, Museum of Comparati\e Zoology 279638. 
Paratvpes from the same localit\', Museum of Comparative 
Zoology 279639. 

Description. Shell globose, valves reaching 1.5 mm in length 
and 1.5 mm in height, thin, fragile; umbos prominent. Perio- 
stracum thin, pale yellow, covering disc and posterior slope. 
Beaked portion of anterior slope with 8-12 pronounced, widely 
spaced, finely denticulated ridges. Posterior portion of anterior 
slope narrow; ridges extending to uml:)onal-\entral sulcus and 
more coarsely denticulated. Umbonal-\'entral sulcus narrow, 
slightly depressed. Disc and posterior slope sculptured with fine 
growth lines only. Umbonal reflection narrow, free except at 
posterior end. 

Inner surface of valves smooth and glistening. Umbonal- 
ventral ridge narrow and indistinctly segmented. Chondrophore 
and internal ligament well developed. Disc separated from 
posterior slope by low ridge. Posterior adductor muscle scar 
elliptical, extending from dorsal margin nearly to ventral mar- 
gin and with indistinct, irregular, transx-erse markings. Pedal 
and siphonal retractor muscle scars impressed. Mesoplax of two 
small, thin, triangular plates composed almost entirely of perio- 
stracum. 

Burrow six times length of valves; calcareous tubular lining 
two-thirds length of Ijurrow, thin, marked with distinct rings, 
and covered with glistening, nearly colorless periostracum, which 
extends anteriorly as border. Portion of animal between \alves 
and anterior end of calcareous tube with finely pustulose, light 
tan periostracal sheath. Siphons short, of about equal length 
and apparently lacking cirri. Protoconch \'erv small, smooth 
(Plate 6, fig. 5). 

Valve measurements. Lenejth 

1 . 1 mm 
1.5 
1.2 
1.0 

Remarks. This species, of which only eight small specimens 
were found, appears to be most closely related to Xyloredo in- 



Height 




1.2 mm 


Holotype 


1.5 


Paratype 


1.3 


n 


1.0 


n 



1972 XYI.OREDO 1 1 

golfia from off Iceland. It diffeis by ha\"ing a imu h smaller 
protoconch (hence the Ian,ae in the pedixeliger stai^je must be 
about t\s()-thirds the size of those of ingolfia) . It also differs 
in that the umbonal-Nentral sulcus is impressed and the height 
of the \al\es is the same as or greater than the length. 

The l)ottom at the test site \vas a green mud; the tempera- 
ture was 2.1 °C. the salinity 34.52 °/o„, and the dissohed 
oxygen 1.26 ml L Muraoka, 1966b). This species has not 
been taken at test site II of the Naval Cixil Engineering Labora- 
tory, located north of San Miguel Island (34°06'N; 12b°42'W), 
where the depth is 2340 feet [713.23 meters^, the bottom tem- 
perature 7.2^C, the salinity 34.37 '^/n.i, and the dissohed 
oxygen 0.42 ml/L .Muraoka, 1965). Specimens of Xyiophaga 
washingtona Bartsch were found at both sites. It would appear 
that temperature and /or dissoh'ed oxygen may be factor (s) 
controlling the distribution of Xyloredo naceli. How'e\-er, so 
little is known about the biology of the Xylophagainae that no 
definite statements can be made at this time. 

Range and specimens examined. Known onl\ from the t\ pe 
locality. 

ACKXO\\LEDGMENTS 

The author is grateful to John DePalma, U. S. Navy Ocean- 
ographic Office, Washington, D.C.; James Muraoka, U. S. 
Na\al Clixil Engineering Laboratory, Port Hueneme, California: 
and Dr. Jorgen Knudsen, L'ni\-ersitetets Zoologiske Museum, 
Copenhagen, Denmark, for the test panels and dredged wood 
from which the specimens were obtained. Thanks are also 
extended to Prof. Kenneth J. Boss of the Mollusk Department. 
Museum of Comparati\e Zoology, and Nancy Kno\\lton. my 
assistant, who critical!) read the manuscript. The aid granted by 
the Office of Naval Research through Contract no. N00014-67- 
A-0298-0027 with Har\ard L^niversity is particularly appre- 
ciated. 

REFERENCES 

BoEGGiLD, O. B. 1900. The deposits of the sea-bottom. The Danish IX- 
GOLF-Expedition, Copenhagen, Vol. 1(2): 1-S9, plates 1-7, 5 text- 
fi.gs. 

Bruun', a. F. 1959. Cjeneral introduction to the reports and list of deep- 
sea stations. GALATHEA Report, Copenhagen, \'oi. 1 : 7-48, plates 
1 — (■, text-figs. 2—11 [section 5, Remnants of plants found in the deep- 
sea, pp. 1 5—17]. 



12 BREVIORA No. 397 

DePalma, J. R. 1969. A study of deep ocean fouling, Straits of Florida 
and Tongue of the Ocean, 1961 to 1968. Informal Report IR No. 69- 
22, U. S. Naval Oceanographic Office, Washington, D.C., pp. 1-26, 
text-figs. 1—6. 

Dons. C. 1929a. Zoologiske notiser. IV. Xylophaga pracstans, ny for 
Norges fauna. Norske Vid. Selskab Forhandlinger, 1(57): 169-172, 
text-figs. 1-7. 

1929b. Zoologiske notiser. V. Xylophafia dorsalls i Norge. 

Norske Vid. Selskab Forhandlinger, 1(65): 196-199, text-figs. 1-6. 

1940. Marine boreorganismer. III. Vekst og voksemate hos 



Xylopluuja dursalis. Norske Vid. Selskab Forhandlinger, 13(18): 76-78, 

text-figs. 1-2. 
Jensen, A. S. 1912. Lamellibranchiata (Part I). The Danish INGOLF- 

Expedition, Vol. 2(5) : 1-119, plates 1-4, text-figs. 1-5. 
Knudsen, J. 1961. The bathyal and abyssal Xylophaga (Pholadidae, 

Bivalvia). GALATHEA Report, Vol. 5: 163-209, text-figs. 1-41. 
Knudsen, M. 1899. -Hydrography. The Danish INGOLF-Expedition, Vol. 

1(1): 23-161, plates 2-35, 2 text figs. 
MURAOKA, J. S. 1964. Deep-ocean biodeterioration of materials — Part I. 

Four months at 5,640 feet. Technical Report R 329, U. S. Naval Civil 

Engineering Laboratory, Port Hueneme, California, pp 1-35, text- 
figs. 1-24. 
1965. Deep-ocean biodeterioration of materials — Part II. 

Six months at 2,340 feet. Technical Report R 393, U. S. Naval Civil 

Engineering Laboratory, pp. 1-42, text-figs. 1-5. 

1966a. Deep-ocean biodeterioration of materials — Part 



III. Three years at 5,300 feet. Technical Report R 428, U. S. Naval 
Civil Engineering Laboratory, pp. 1—47, text-figs. 1—36. 

1966b. Deep-ocean biodeterioration of materials — Part 



IV. One year at 6,800 feet. Technical Report R 456, U. S. Naval 
Civil Engineering Laboratory, pp. 1-45, text-figs. 1—31. 

1966c. Deep-ocean biodeterioration of materials — Part 



V. Two years at 5,640 feet. Technical Report R 495, U. S. Naval 
Civil Engineering Laboratory, pp. 1—46, text-figs. 1—36. 

1967. Deep-ocean biodeterioration of materials — Part 



VI. One year at 2,370 feet. Technical Report R 525, U. S. Naval 
Civ'il Engineering Laboratory, pp. 1—57, text-figs. 1—43. 

PuRCHON, R. D. 1941. On the biology and relationships of the lamelli- 
branch Xylopluuja dorsalis (Turton). Journal of the Marine Biologi- 
cal Association of the United Kingdom, 25: 1-39, text-figs. 1-16. 

Tipper, R. 1968. Ecological aspects of two wood-boring molluscs from 
the continental terrace off Oregon. Department of Oceanography, 
School of Science, Oregon State University, pp. 1-137, text-figs. 1-50. 

ToMLiN, J. R. L. B. 1920. Notes on Xylophaga pracstans, Smith. Pro- 
ceedings of the Malacological Society of London, 14(2-3): 73. 

Turner, H. J. 1961. Deep teredo [Xylophaga]. Oceanus, 8(2): 11. 



1972 XYLOREUO 13 

Turner, R. O. 1966a. Implications of recent researcli in the Teredinidae. 
In Hoiz und Organisnien (Cj. Beci<er and \\'. Lie«e, eds.) Berlin: 
Duncker and Humbolt, pp. 437—446, text-figs. 1-4. 

1966b. Some results of deep water testing. Annual Re- 
port for 1965, American Malacological Union, pp. 9-11. 

1968. The Xylophagainae and Teredinidae — A study in 



contrasts. Annual Report for 1967, American Malacological Union, 
pp. 46-48. 
1969. The superfamily Pholadacea. In Treatife on In- 



vertebrate Paleontology, Part N, Vol. 2, Mollusca 6, Bivalvia, pp. 702- 
742, text-figs. 162-214. 
, AND A. C. Johnson. 1971. Biology of marine wood-boring 



molluscs. /// Marine borers, fungi and fouling organisms of wood 
(E. B. G. Jones and S. K. Eltringham, eds.) Paris: Organization for 
Economic Co-operation and Development, pp. 259—301, text-figs. 1—14. 



14 



BREVIORA 



No. 397 




Plate 1 
Xylorrdo noo'i Turner 
Figure 1. Section of panel submerged in the Tongue of the Ocean, 
Bahama Islands, from April 4, 1962 to February 17, 1965 in 1737 meters 
(about 1/4 ^" ) . Figure 2. Enlarged section cf panel showing the calcare- 
ous lining of the burrow with anterior pericstracal border, and dorsal view 
of apposed valves (about 4 X ). 

Plate 2 
Xylurrdo nooi Turner 
From panels submerged in the Tongue of the Ocean, off Andros Island, 
Bahama Islands. Figures 1-2. Holotype. Fig. 1. Outer view of left valve 
showing high, Haring posterior dorsal margin. Fig. 2. Inner view of right 
valve showing di\ide(i muscle scar, the deep groove separating the disc 
from the posterior r.lope, and the umbonal reflection. Figures 3—4. Outer 
and inner views of left valves of smaller specimens. Figure 5. Anterior 
view of apposed valves showing the chondrophore and internal ligament. 
Figure 6. Dorsal view of apposed valves showing mesoplax. 



1972 



XYI.OREDO 



15 




16 



BREVIORA 



No. 397 










I. JJ -^ -** * 






Plate 3 



Xylorc-io ijujolfia Turner 
Figure 1. Section through wood dredged at INGOLF-Expedition sta- 
tion 67, south of Eyrahakki, Iceland (6l"30'\: 22°30'W) in 17S3 meters, 
showing narrow, parallel tubes (5 >'). Figure 2. Enlarged section of 
burrow showing calcareous lining with anterior periostracal border (10 X ). 
Figure 3. Surface of wood showing entrances of burrows (5 X ). 



1972 



XYLOREDO 



17 




1 mm 



Plate 4 



XylorrJo Ingnlfid Turner 

Fiom INGOLF-Expeditlon, staticn 67, south of Eyrabakki, Iceland. 

Figures 1—2. (Holotype.) Lateral and dorsal views of entire animal 
showing the periostracal sheath, the extended anal and siphonal canals, and 
the short siphons. Figure 3. Lateral view of very young specimen show- 
ing large prodisoconch. Figure 4. Enlarged dorsal view of anterior end 
of entire specimen showing umbonal reflection, anterior adductor muscle, 
and small plates of the mesoplax in place. 



18 



BREVIORA 



No. 397 




Plate 5 



Xyloredo ingolfia Turner 

From INGOLF-Expedition, station 67, south of Eyrabakki, Iceland. 

Figure 1. Outer view of left valve showing w'ide anterior slope and 
low, rounded posterior slope. Figure 2. Inner view of left valve showing 
large prodisoconch, strong umbonal-ventral ridge, reduced posterior slope, 
and lightly impressed muscle scars. Figure 3. Inner view of right valve. 
Figure 4. Inner view of left valve. Figure 5. Outer view of left valve 
with periostracal sheath attached. 



1972 



XYLOREDO 



19 




Plate 6 



Xylorrdo naceli Turner 

From U. S. Naval Civil Engineering Laboratory test site I, south of 
San Miguel Island, Santa Barbara Islands, California. 

Figure 1. Lateral view of specimen partially dissected from the wood 
showing the calcareous tube \vith the anterior periostracal margin and 
the papillose periostracal sheath covering the animal between the tube 
and the valves. Figure 2. Anterolateral view of entire specimen showing 
foot, mesoplax, and siphons. Figure 3. Lateral view of holotype. Figure 
4. Outer view of left valve. Figure 5. Inner view of left valve showing 
muscle scars, chondrophore, umbonal-ventral ridge, and small prodiso- 
conch. 



MAR i 8 7985 

B R E V I O t-* 

IMuseinn of Comparative Zoology 

Cambridge, Mass. March 6, 1973 Number 398 



PSEUDOBEACONIA, 

A PERLEIDIFORM FISH FROM 

THE TRIASSIC SANTA CLARA FORMATION, 

ARGENTINA 

Peter Hutchinson^ 

AbstR-ACT. Two species of the genus Psevdobeaconia, P. braeac- 
dnii and P. elegans, fx-om the Santa Clara Formation, Argentina, are 
described. The genus is an advanced member of tlie family Colobo- 
dontidae (subclass Chondrostei, order Perleidiformes), and is closely- 
related to Meyidocinia from the same locality. It is suggested that 
the success of the Perleidiformes is due, in part, to the evolution of 
a crushing dentition. The detailed structure of the unpaired fin 
rays indicates that control of the fin web area was possible in Pseudo- 
beaconia, and probably in all Perleidiformes. The identification of 
Psendobcacoiua does not help the determination of the age of the 
Santa Clara Formation. 

Abstracto. Se describen dos especies del genero Pseudobeaconia, 
P. bracacclnii y /'. eleyans, de la Formacion de Santa Clara, Argen- 
tina. El genero es un miembro avanzado de la familia Colobodontidae 
(subclase Chondrostei, orden Perleidiformes), y esta intimamente 
relacionado con Mendocbiia de la misma localidad. Se sugiere que 
el exito de los Perleidiformes se debe, en parte, a la evolucion de una 
denticion molar. La estructura detallada de los rayos de las aletas 
impares indica que el control del area de la membrana de las aletas 
era posible en P.'^rudnbcnconia, y probablemente en todos los Perleidi- 
formes. La identificacion de Pseudobeaconia no facilita la deteniii- 
nacion de la edad de la Formacion de Santa Clara. 

INTRODUCTION 

In 1944 three genera of fossil fish were described from the 
Santa Clara Formation, a series of shales, mudstones, and sand- 
stones that outcrop between the pro\inces of San Juan and 
Mendoza, western Argentina ( Bordas, 1 944 ) . Two of these 
genera were new, Mendocinia and Pseudobeaconia, while the 

'Gothic House, Drayton, Berks, England. 



2 BREVIORA No. 398 

third was doubtfully identified as Cleithrolepis. More recently 
Schaeffer (1955) has redescribed Mendocinia and has shown 
that it is a member of the Perleidiformes. Pseudobeaconia re- 
mained relatively unknown, although it appeared in most class- 
ifications as a member of the Redfieldiiformes {e.g., Gardiner, 
1967a) because Bordas had described similarities between it 
and a redfieldiiform genus called Beaconia ( = Brookvalia, 
Hutchinson, 1972) from the Triassic Hawkesbury Series of New 
South Wales ( Wade, 1 935 ) . 

The material described here was collected by an expedition 
from the Museo de La Plata and the Museum of Comparative 
Zoology, Harvard University, in 1964, and is derived from two 
localities in the Santa Clara Formation. The first is near the 
north end of the formation, near the Puesto of Santa Clara, at 
the east margin of the Precordillera in southern San Juan Prov- 
ince. This is the same locality that produced the fish described 
by Bordas in 1 944. The second locality is north of the Quebrada 
de la Montana, a short distance east of the boundary' between 
the Santa Clara beds and the Villavicencio group (Romer, 
1966). 

There is little doubt that the specimens described here rep- 
resent the two species of Pseudobeaconia described by Bordas 
in 1944. The reasons for this identification are given in the dis- 
cussion below. P. bracaccinii occurs in both the locaUties men- 
tioned above, while P. elegans occurs only in the former. 

Pseudobeaconia is preserved in a matrix of dark mudstone or, 
more commonly, shale. The shales from the second locality 
usually contain bands of a light calcareous deposit. There may 
be up to ten such bands in a sample 10 mm thick. As well as 
fish, fragments of plant material are preserved in the shales. 
Preservation is extremely good and the fish have either been 
prepared with a fine needle or sometimes, when they occur in 
calcareous bands, dissolved in acetic acid to reveal natural 
moulds. All the specimens had been crushed to a thickness of, 
at most, 2 mm. 

The Santa Clara Formation is difficult to date because it is 
limited to a block isolated by faulting during the Tertiary. Bor- 
das (1944) is of the opinion that the Santa Clara Formation is 
contemporaneous with the Hawkesbury Sandstone of Brook vale. 
New South Wales. This conclusion is based on the supposed 
close relationship between Pseudobeaconia and Beaconia, a con- 
clusion that is unfounded according to evidence presented here. 
Harrington (unpublished MS) notes that, near the locality de- 



1972 TRIASSIC FISH FROM ARGENTINA 3 

scribed by Bordas, there are plant remains that include Glossop- 
teris, Walkonia. Pecopteris, Gondwanidium, Cordaites, and Di- 
cranophyllu7n, which are Permian in age. But, as noted above, 
the Santa Clara Formation is isolated from other sediments in 
the area, and there is no evidence that the flora listed by Har- 
rington occurred within its boundaries. Schaeffer ( 1 955 : 3 ) 
has estimated the age of the beds of the Cacheuta basin (and 
by inference the Santa Clara Formation) to be Norian. This 
estimate is based on a similarity between floras of the Cacheuta 
and Upper Triassic sediments of the Northern Hemisphere. 
Lack of similar comparati\e data from the Middle Triassic 
means that this estimate can only be regarded as an indication 
of the youngest possible age of the Cacheuta beds. 

A possible estimate of the age of the Santa Clara Formation 
depends on correlations made recently bv Bonaparte and runs 
as follows. The Santa Clara Formation forms one of several 
outcrops that lie within the Cacheuta basin of deposition 
(Stipanicic, 1967: 3). Another of these outcrops contains the 
Potrcrillos Formation from ^vhich a cynodont lower jaw, Col- 
bertosaurus, has been recovered (Minoprio, 1954). Colbertosau- 
rus is similar to Pascualgnathus from the Puesto Viejo Forma- 
tion (Bonaparte, 1966a: 4-5). Thus the Potrerillos and Puesto 
Viejo formations can be considered to be very similar in age. 
The age of the Puesto Viejo formation is considered to be Scy- 
thian because it also contains a dicynodont Kannemeyeria (Bon- 
aparte, 1966b: 266) and a cynodont Cynognathus (Bonaparte, 
1969), both typical genera of the Lower Triassic South- African 
fauna. 

It should be emphasized that this estimate of Lower Triassic 
age for the Santa Clara Formation is tentatiNC and open to two 
objections. Firstly, it is not certain that all the outcrops within 
the Cacheuta basin were once continuous. Secondly, the re- 
semblance between Colbertosaurus and Pascualgnathus is not 
completely convincing, and has limited significance, according 
to Bonaparte (1966a: 5). 

The identification of Alendocinia and Pseudobeaconia from 
Santa Clara adds nothing of significance to the debate on the 
possible age of the formation. Both genera are more advanced 
than Perleidus from the Lower Triassic of Madagascar, Green- 
land, and Spitsbergen but, because they are not directly de- 
scended from Perleidus, this does not necessarily indicate that 
the Santa Clara Formation is younger than Lower Triassic. An- 
other member of the same perleidid radiation, Meidiichthys, 



4 BREVioRA No. 398 

from the Cynognathus zone of South Africa, demonstrates that 
advanced features such as an anteriorly inclined suspensorium 
evolved extremely rapidly, and the presence of such characters 
in forms from Santa Clara does not exclude the possibility that 
the Santa Clara Formation is Lower Triassic in age. 

DESCRIPTION 

Order Perleidiformes 

Family COLOBODONTIDAE Stensio 1916 

Pseiidobeaconia Bordas 1944 

Pseudoheaconia Bordas, 1944: 454. 

Type species. Pseudobeaconia hracaccinii Bordas 1944 

Diagnosis (emended). Skull short anteroposteriorly. Skull 
roof evenly curved to snout tip, upper jaw steeply inclined. Skull 
roof composed of two extrascapulars, a parietal, and a frontal. 
Lateral walls of skull roof composed of a dermopterotic and an 
accessory dermopterotic. Ventral end of the suspensorium in- 
clined anteriorly so that the jaw articulation lies vertically be- 
low the posterior orbital edge. Opercular series composed of 
opercular, subopercular, and three or four branchiostegal rays. 
Opercular anterior in position so that its anterior edge lies in 
advance of the level of the posterior edge of the parietal. Sub- 
opercular larger than opercular. Triangular dermohyal and sub- 
orbital elements. Dermohyal small, extending ventrally along 
half the anterior edge of the opercular. Dermosphenotic short 
anteroposteriorly, in contact dorsally with both the posterior 
supraorbital and the dermopterotic. Dorsal orbital edge bor- 
dered by two supraorbital elements. Maxilla not greatly ex- 
panded posteriorly. Snout composed of paired nasals, antor- 
bitals and (probably paired) toothed premaxillae, and a medial 
rostral and postrostral. Nasal bordering anterior orbital edge. 
Rostral large, bordering the anterior edge of the nostril. Nos- 
tril elliptical. Marginal teeth on maxilla and dentary conical 
and pointed, palatal teeth low and rounded. 

Body fusiform, attaining a total length of up to 120 mm. 
Dorsoventral length of scales in flank region up to three times 
their anteroposterior length. Scales in the transverse row im- 
mediately behind the pectoral girdle twice as long anteropos- 
teriorly as succeeding body scales. Scale ornament of between 
two and six ridges running concentrically along posterior and 
ventral borders. Scales in lateral line series with an indentation 
in their posterior borders. No hinge line at base of caudal lobe 



1972 ■ TRIASSIC FISH FROM ARGENTINA 5 

of body. Ridge scales present anterior to unpaired fins and on 
the dorsal side of the body midway between the skull and dor- 
sal fin. Dorsal and anal fins posterior in position. Caudal fin 
hemiheterocercal. All fin rays jointed and usually bifurcated. 
Basal fulcra present anterior to all fins. Paired fin rays appar- 
ently lacking fringing fulcra. Proximal lepidotrichia of all fins 
at least three times as long as distal lepidotrichia. Proximal 
lepidotrichia of unpaired fins equal in number to endoskeletal 
supports, and expanded proximally, with insertions for depres- 
sor and erector muscles. 

Pseudoheaconia bracaccinii Bordas 1944 

Figures 1-5 ■ 

Pseudoheaconia bracaccinii Bordas, 1944 : 454-455, pi. 1. 

Diagnosis (emended). Parietal just under half the length of 
the frontal. Opercular plate roughly rectangular, its anteropos- 
terior length being under half its dorsoventral length. Ventral 
edge of opercular deeply convex. Posterior edge of maxilla lying 
anterior to the level of the posterior orbital edge, x\nterior supra- 
orbital extending anteriorly as far as the anterior end of frontal. 
Suprascapulars separated at midline. Dorsoventral length of 
supracleithrum twice that of cleithrum. 

Body fusiform. Body length, from snout tip to the posterior 
tip of the caudal lobe of the body, three times the greatest depth 
of the body. Scale formula : 

25 

13 21 35 

Length of the longest pectoral fin rays 1.5 times the length 
of the longest pelvic fin rays. Some paired fin rays bifurcated 
distally. Dorsal fin rays not bifurcated. Dorsal fin slightly in 
ad\'ance of anal fin. Dorsal lobe of caudal fin sHghtly longer 
than ventral lobe. 

Ilolotypc. Museo Argentino de Ciencias Naturales at Ber- 
nardino Rivadavia, Catalogue of Paleontology (Vertebrates) 
No. 14.868 (Bordas, 1944: pi. 1). 

Additional material. Thirty-two specimens in the collection 
of the Museum of Comparative Zoology, Cambridge, Massa- 
chusetts. 

Skull. The dermal bones of the skull are restored in Figure 
1. The skull roof slopes evenly to the snout, and is composed 
of extrascapulars (which are divided [MCZ 12891]), parietals. 



BREVIORA 



No. 398 



ex 



pmx 




clav 



Figure 1. Pseudobeaconia bracaccinii Bordas. Restoration of 
skull and pectoral girdle, ornament omitted. X 5 approx. 



na 



pmx 



pspi 




Fig-ure 2. Pseudobeaconia bracaccinii Bordas. MCZ 12883. Left 
side of skull and pectoral girdle. 



1972 TRIASSIC FISH FROM ARGENTINA 7 

and frontals. The parietal is square; its anteroposterior length 
is just under iialf the length of the frontal. The frontal is deep 
and excavated above the orbit so that its widest part lies above 
the posterior edge of the orbit. 

The lateral part of the skull roof is composed of two ele- 
ments, a dermopterotic and an accessory dermopterotic, which 
are separated from one another by a vertical suture that runs 
from a point midway along the \entral margin of the parietal 
(Fig. 2). In MCZ 12883, the dermopterotic appears to be 
shallow, but this is because the preopercular has been displaced 
and overlaps its ventral part. The full depth of the dermoptero- 
tic is seen in MCZ 12894. The dermopterotic meets the pos- 
terodorsal border of the dermosphenotic, and may also have 
point contact with the posterior supraorbital element. A post- 
spiracular lies posterior to the accessory dermopterotic and meets 
the \entral border of the extrascapular series. In lateral view 
the anterior part of the postspiracular is overlapped by the 
opercular, but in MCZ 12893 it is seen to extend anteriorly 
along the entire dorsal border of the opercular. 

The opercular is circular, and the subopercular roughly ob- 
long but with a curved posteroventral border. In most speci- 
ments the ventral part of the subopercular is crushed against 
the pectoral girdle, but its outline is clearly seen in MCZ 12898, 
where it has been slightly displaced. Its dorsoventral length is 
slightly longer than that of the opercular. There are only three 
or four rather broad branchiostegal rays. The preopercular is 
high, and meets the \entral edge of the dermopterotic. Its 
ventral part is inclined forwards so that the lower jaw articula- 
tion lies almost directly belovv the posterior edge of the orbit. 
The preopercular has a short infraorbital process. In MCZ 
12883 (Fig. 2) an element lies between the posterior edge of 
the maxilla and the anteroventral corner of the subopercular. 
This is not a fragment of the preopercular, for in MCZ 12900 
there is a notch on the median surface of the \entral end of the 
preopercular which indicates the position of a separate element 
(Fig. 1). It is likely, therefore, that a quadratojugal is present. 
There is no sign of any sensory canal on the quadratojugal. The 
dorsal end of the preopercular is constricted between triangular 
dermohyal (MCZ 12893, 12891) and suborbital elements 
(MCZ 12883). In MCZ 12893, which includes a natural 
mould of the median surface of the cheek, there is a small 
triangular accessory opercular at the antcrodorsal corner of the 
subopercular. This element is not visible in lateral view, as it 
is completely overlain by the subopercular. 



8 



BREVIORA 



No. 398 



The dorsal orbital edge is bordered by two supraorbitals. 
There is some variation in the relative lengths of these elements; 
in MCZ 12898 and probably also in MCZ 12883 (Fig. 2) the 
posterior element is slightly longer than the anterior element, 
whereas in MCZ 12891 (Fig. 3A), the posterior element is 
slightly shorter than the anterior one. The posterior orbital edge 
is bordered by a narrow dermosphenotic, which has an ex- 
panded dorsal end, and the ventral orbital edge by two in- 
fraorbitals. The anterior infraorbital is expanded anteriorly, 
while the posterior infraorbital is expanded midway along its 
length. 

In MCZ 12883 (Fig. 2), the maxilla has been displaced so 
that its posterior part is overlapped by the preopercular. The 
preopercular is so thin in this region, however, that the outline 
of the maxilla is clearly seen as a ridge. The upper jaw mar- 
gin is steeply inclined, but lies at right angles to the anterior edge 
of the subopercular. The maxillary teeth are pointed (MCZ 
12894, 12895, 12898, and 12901),' and appear to be present 
onh' along the anterior half of the maxillar)- border. There are 
about 20 teeth in the upper jaw row. 

The snout is best preserved in MCZ 12893, 12891 (Fig. 
3A), and 12894 (Fig. 3B). The nasal borders the anterior or- 
bital edge and is in contact dorsallv with both the frontal and 



rr(l) 




2nnm 



A B 

Figure 3. Pseudobeaconia bracaccinii Bordas. A. MCZ 12891» 
Snout regon, right side unless otherwise indicated. B. MCZ 12894. 
Snout region, left side. 



1972 TRIAS SIC FISH FROM ARGENTINA 9 

the anterior supraorbital. The \entral part of the snout is corri- 
posed of paired antorbitals, which border the antero\entral 
corner of the orbit, and premaxillaries, which bear up to six 
teeth (MCZ 12894, Fig. 3B). It is not known for certain 
whether the premaxillaries are paired or fused. The anterior 
part of the snout is composed of medial rostral and postrostral 
elements. The rostral bears the ethmoid commissure and ex- 
tends dorsally so that it borders the anterior margin of the nos- 
tril (Fig. 3), while the postrostral is a small bone which just 
borders the anterodorsal comer of the nostril. The suture be- 
tween the rostral and postrostral elements is indistinct in MCZ 
12891, but clearly seen in MCZ 12894 (Fig. SB). 
The lower jaw is best preserxed in MCZ 12893. 
Most of the dermal bones of the skull are smooth. Tubercles 
are present, however, on the anterior ends of the frontal and 
anterior supraorbital, on the nasal, and on the infraorbital part 
of the maxilla. Low ridges are seen on the opercular and sub- 
opercular elements and near the edges of the branchiostegal 
ravs ( Fig. 4 ) . 

The sensory canals of the head are enclosed in massive tubes 
which open to the surface through well-defined pores (MCZ 
12883, Fig. 2). In adchtion, there is a vertical pit line on the 
ventral part of the preopercular, while short, curved, middle 
and posterior pit lines are present on the parietal. 

All the skulls of P. bracaccinii are crushed and few details of 
the internal anatomy are displayed. In MCZ 12897, however, 
part of the palate is visible and about ten low and rounded 
palatal tusks can be seen. In MCZ 12897 two branchial arch 
elements are seen, lying against the anterior edge of the cleith- 
rum; they are flattened and their posterior edges bear numerous 
toothlike projections that are less than 0.1 mm long. 

Pectoral girdle. The suprascapulars are D-shaped and do 
not meet at the midline as they are separated by a pair of scales 
that meet the posterior edges of the dorsal extrascapular ele- 
ments ^MCZ 12883, Fig. 2; MCZ 12891). The infraorbital 
sensory canal crosses the antcro\'entral corner of the suprascapu- 
lar (MCZ 12893). The supracleithrum, which extends ven- 
trally to a point close to the ventral edge of the opercular, has 
an excavated dorsal end. This excavation does not appear to 
fit directly onto the ventral edge of the suprascapular, for in 
several specimens it terminates at a level nearer the midline 
than does the suprascapular edge (MCZ 12893, 12891, and 
12894). It is possible, therefore, that the dorsal end of the 
pectoral girdle is buttressed, in part at least, against the pos- 
terolateral wall of the braincase. The donsoventral length of the 



10 



BREVIORA 



No. 398 




73 
0) 



to 

V 

u 
_o 

a; 

c 
o 



c 
£ 



c 



en 



to 

o 

m 

i 

e 
c 
e 



5£ 
o 

o 

Oh . 
X 

o 

Si n 

s 
btl "= 

X 



1972 TRIASSIC FISH FROM ARGENTINA 11 

cleithrum is twice as long as that of the supracleithrum. The 
cleithrum has a broad \cntral plate, and its anterior border 
bears an ornament of sinuous rugae. Small cla\icles are pres- 
ent (MCZ 12893, 12899, 12901).' 

Paired fins. The paired fins are restored in Figure 4. In 
both, the rays are jointed so that the proximal lepidotrichia are 
up to six times as long as the distal ones, while the longest 
rays are bifurcated at their extreme distal ends ( MCZ 1 2893, 
12906). Fringing fulcra were not observed and are probal)ly 
absent, but in both fins a stout basal fulcral scale participates in 
the leading edge. The pectoral fin is composed of 1 3 rays, which 
have between 4 and 12 joints. In well-preserved specimens (for 
example MCZ 12883, Fig. 2), the bases of the anterior rays are 
inserted in a more dorsal position than are those of the poste- 
rior rays. The pcKic fin lies two-thirds of the way between the 
pectoral and anal fins. It is composed of seven rays, which ha\e 
between four and eight joints. 

Unpaired fins. The unpaired fins are restored in Figure 4. 
In all, the rays are jointed, so that the proximal lepidotrichia 
are from three to eight times as long as the distal ones. The 
rays of the anal and caudal fins are bifurcated, and in all the 
unpaired fins the leading rays bear fringing fulcra. The dorsal 
fin is composed of 16 rays, which have between three and nine 
joints, and is preceded by three basal fulcra. The anal fin is 
composed of 14 rays, which have between 9 and 22 joints, and 
is preceded by a single basal fulcral scale. The dorsal and anal 
fins are similar in shape, but the former is larger and is slightly 
more anterior in position than the anal fin. The caudal fin is 
hemiheterocercal and the rays of its dorsal lobe are slightly 
longer than those of its ventral lobe. It is composed of 21 rays, 
which have between 15 and 24 joints. The four rays, which 
have their origins on the dorsal side of the caudal lobe of the 
body, do not bifurcate. There are about six basal fulcral scales 
on the dorsal side of the caudal lobe of the body; their exact 
number is difficult to determine because anteriorly they become 
very similar in form to the ridge scales, which extend posteriorly 
from the posterior edge of the dorsal fin. 

In MCZ 12892 and 12905 some details of the articulation of 
the anal fin rays are seen, and in MCZ 12912 (Fig. 5) similar 
details are seen in the dorsal fin. In both fins the number of 
radials equals the number of rays. In the dorsal fin the prox- 
imal lepidotrichia have expanded proximal ends but, whereas 
the posterior lepidotrichia are iilmost straight, the anterior ele- 



12 



BREVIORA 



No. 398 




<zr 



1 mnri 



Figui-e 5. Pseudobeaconia bracaccinii Bordas. MCZ 12912. Prox- 
imal lepidotrichia and distal radials of dorsal fin rays 7-10. Right 
side. 

ments become progressively more curved. Two grooves extend 
along the entire length of each fin ray, running close to the an- 
terior and posterior edges. At their proximal ends, these grooves 
open into broad depressions so that there is an anterior and a 
posterior depression on each side of the base of every fin ray. 
As one passes anteriorly along the dorsal fin ray series, the an- 
terior depressions become progressively larger, while the poste- 
rior depressions become progressively smaller ( Fig. 5 ) . The 
probable significance of these features is discussed below. 

Squamation. The scales are restored in Figure 4. There are 
35 transverse scale rows between the anterior transverse scale 
row and the row in front of the anterior caudal fin ray. The 
positions of the pelvic, dorsal, and anal fins relati\'e to the scale 
rows is given in the formula : 

25 

13 21 35 

(See WestoU, 1944: 20. Note that, because there is no clear 
hinge line in Pseudobeaconia, the last figure gi\en in Westoll's 
formula is not used here.) The scales in the trans\erse row im- 
mediately behind the pectoral girdle are considerably longer an- 
teroposteriorly than are the scales in any other row. Each scale 
in the lateral line series has a slight indentation in its posterior 
border, coinciding with the lateral line canal. The outlines of 
all the other scales are smooth. There is no hinge line between 
the scales of the body and those of the caudal lobe. Ornament 



1972 



TRIASSIC FISH FROM ARGENTINA 



13 



is limited to between 2 and 6 ridges that run concentrically 
along- the posterior and ventral borders of each scale. 

Ridge scales occur in front of the unpaired fins and also on 
the dorsal side of the body between the skull and the dorsal fin. 



Pseudoheaconia elegans Bordas, 1944 

Figures 6-8 

Pseudoheaconia elegans Bordas, 1944: 456, pi. 1. 

Diagnosis (eynended). Parietal just over half the length of 
the frontal. Opercular plate D-shaped, its anteroposterior length 
being o\er half its dorsoventral length. \'entral edge of the 
opercular slightly convex. Posterior edge of the maxilla lying 
posterior to the le\el of the posterior orbital edge. Anterior 
supraorbital extending anteriorly beyond the anterior end of 
the frontal. Suprascapulars meeting in the midline. Dorsoven- 
tral length of supracleithrum equal to that of the cleithrum. 

Body fusiform. Body length, from snout tip to the posterior 
tip of the caudal lobe of body, 3.5 times the greatest depth of 
the body. Scale formula : 

23 



11 



20 



34 



ro 



pm 




clav 



Figure 6. Pseudobeaconia elegans Bordas. Restoration of skull 
and pectoral girdle, ornament omitted. X 5 approx. 



14 



BREVIORA 



No. 398 



ex. 



pro 




pmx 
(I). , 



clav(l) clav 



2mr 



Figure 7. Pseudobeaconia elegants Bordas. MCZ 12892. 
and pectoral girdle, right side unless otherwise indicated. 



Skull 



Length of the longest pectoral fin rays about 2.5 times the 
length of the longest pelvic fin rays. Paired fin rays probably 
not bifurcated. Dorsal fin rays bifurcated. Dorsal fin almost 
entirely in advance of anal fin. Caudal fin equilobate. 

Holotype. Museo Argentino de Ciencias Naturales at Ber- 
nardino Rivadavia, Catalogue of Paleontology (Vertebrates) 
No. 14.689 (Bordas, 1944 : pi. 1 ). 

Additional material. Four specimens in the collection of the 
Museum of Comparative Zoology, Cambridge, Massachusetts. 

A full description of P. elegans is not necessary, as many of 
its features ha\e already been observed in the type species P. 
bracaccinii. 

Skull. The restoration of the dermal bones of the skull 
(Fig. 6) is based mainly on MCZ 12892 (Fig. 7). The an- 
teroposterior length of the parietal is just over half the length 
of the frontal. The dermopterotic appears to be deeper than 
in P. bracaccinii, and the anterior supraorbital extends ante- 



1972 TRIASSIC FISH FROM ARGENTINA 15 

riorly beyond the anterior tip of the frontal. The opercular is 
less rounded than in P. brncaccinii, and has a straighter \cntral 
edge, and the subopcrcular is relati\ely longer anteroposteriorly 
(MCZ 12893). As a result of these differences the opercular 
plate is D-shaped rather than rectangular, as in P. bracaccinii. 
The maxilla, although similar in shape to that of P. bracaccinii, 
extends posteriori) beyond the Ie\el of the posterior edge of the 
orbit. 

Pectoral girdle. The suprascapulars are D-shaped and ap- 
pear to meet in the midline. The supraclcithrum extends ven- 
trally beyond the \entral edge of the opercular. The dorsoven- 
tral length of the supraclcithrum is the same as that of the 
cleithrum. 

Paired fins. The paired fins are restored in Figure 8. Both 
are incompletely preserved. The rays are jointed so that the 
proximal lepidotrichia are up to eight times as long as the distal 
ones. Fringing fulcra are not preser\ ed, and the rays appear not 
to be bifurcated. The pectoral fin is composed of at least seven 
rays that are extremely long, so that the fin is a fifth of the length 
of the entire fish ( MCZ 1 2893 ) . The pehic fin is composed of 
at least seven ra)s that ha\e up to four joints, and lies two-thirds 
of the way between the pectoral and anal fins. 

Unpaired fins. The unpaired fins are restored in I'igure 8. 
In all, the rays are jointed so that the proximal lepidotrichia 
are from five to eight times as long as the distal ones. The an- 
terior ravs of each fin bear fringing fulcra, and all the others are 
bifurcated. The dorsal fin is composed of 13 long ra\s that ha\e 
between 11 and 14 joints, and is preceded by three basal fulcra. 
The anal fin is composed of 15 rays that have between 10 and 
18 joints, and is preceded by two basal fulcra. The base of the 
dorsal fin lies almost entirely in ad\ance of the level of the an- 
terior end of the anal fin. The caudal fin is hemiheterocercal 
and equilobate. It is composed of 20 rays that ha\e between 18 
and 25 joints. There are about seven basal fulcral scales on the 
dorsal side of the caudal lobe of the body. 

Squamation. The scales are restored in figure H. There are 
34 transverse scale rows between the anterior transverse scale 
row and the row in front of the anterior caudal fm ra\ . The 
scale formula is: 

23 

11 20 34 

The transverse scale rows are less steeply inclined than in P. 



16 



BREVIORA 



No. 398 




■1^ 



S 
o 

CO 



<!) 
■*-> 

w 

o 

o 

c 

s 

c3 



_o 

c3 

o 
■*-> 

CO 






o 

K 






o 
CO 



00 



o 

X 



1972 TRIASSIC FISH FROM ARGENTINA 17 

bracaccinii, a feature to be expected in the squamation of a 
more fusiform body. 

DISCUSSION 

The Systematic Position of the Genus Pseudobeaconia 

Before discussing the systematic position of Pseudobeaconia, 
it is necessary to state why the material described in this paper 
is considered to represent the genus described by Bordas in 1944. 
Pseudobeaconia is usually assigned to the order Redfieldiiformes 
(Gardiner, 1967b), whereas the specimens described here are 
clearly members of the Perleidiformes, as will be sho\vn below. 

The original diagnosis of Pseudobeaconia (Bordas, 1944: 
454) is based on poorly preser\ed material and is therefore 
brief. The most important diagnostic features of the genus noted 
by him are: the absence of branchiostegal rays; the presence of 
a reduced tabular and postparietal ; a dorsal fin divided into two 
parts, an anterior part near the head and a posterior part lying 
in the posterior part of the body; and finally, scales longer dor- 
soxentralh- than anteroposteriorly. These features are seen in 
the material described here, or can reasonably be considered as 
misinterpretations based on poorer material of the same taxa 
investigated in the present paper. The tabular and postparietal 
of Bordas correspond to the two extrascapular elements, and 
the anterior dorsal fin to the row of ridge scales that lie between 
the head and the dorsal fin. The supposed absence of branch- 
iostegal rays is probably due to the poor preser\ation of Bordas' 
material. The similarities discussed so far would not allow cer- 
tain identification of the present material with Pseudobeaconia. 
However, Bordas recognized two species of the genus: P. 
bracaccinii, \vhich was characterized by a quadrangular opercu- 
lar, a subopercular slightly longer dorso\entrally than antero- 
posteriorly, and a dorsal fin that was slightly anterior to the 
level of the anal fin; and P. elegans, which was characterized 
by a more delicate body shape, an opercular that was less quad- 
rangular than in P. bracaccinii, and a dorsal fin that was more 
clearly anterior to the le\el of the anal fin (Bordas, 1944: 455, 
456). These difTerences correspond exactly to the differences 
between the two species recognized in the present material {cf. 
Figs. 4 and 8 ) , strongly suggesting that we are dealing with the 
same taxa. The material described here can therefore be as- 
cribed to the genus Pseudobeaconia with a high degree of con- 
fidence. This conclusion is strongly supported b)- the fact that 



18 BREVIORA No. 398 

Bordas* material and most of that described here have been 
collected from the same locality. 

Returning to the systematic position of Pseudobeaconia, it is 
clear that the genus is not a redfieldiiform. Members of the 
Redfieldiiformes are characterized by a number of features, such 
as a dermosphenotic that borders the orbital edge, an R-shaped 
preopercular, an antorbital that borders the nostril edge, and an 
excess of rays over endoskeletal supports in the unpaired fins 
(Hutchinson, 1972), none of which are seen in Pseudobeaconia. 
On the contrary, consideration of the morphology of Pseudo- 
beaconia leaves no doubt that it is a member of the Colobodon- 
tidae ( = Perleididae ) , a perleidif orm family well represented 
in both marine and freshwater Triassic deposits. W^ith t\vo ex- 
ceptions, Pseudobeaconia has all the diagnostic characters of 
this family {cj. SchaefTer, 1955: 19). The exceptions are the 
presence of a suspensorium that slopes forwards, and a rostral 
that extends dorsally to take part in the nostril border. The 
first of these characters represents an advanced state of a trend 
well known in the family, while the second must be regarded 
as a specialization of the genus. 

When compared to other members of the Colobodontidae, 
Pseudobeaconia appears to be similar to Mendocinia (Fig. 9B), 
which occurs in the same locality (Bordas, 1944; SchaefTer, 
1 955 ) , but differs from that genus in having the following 
characteristics: subdixided extrascapulars; lateral part of skull 
roof composed of dermopterotic and accessory dermopterotic 
elements; ventral end of suspensorium inclined anteriorly; op- 
ercular anterior in position relatixe to skull roof elements; der- 
mohyal small; two supraorbital elements; maxilla not greatly 
expanded posteriorly; rostral large and bordering anterior edge 
of nostril; scales with ornament of ridges; scales of lateral line 
series indented; unpaired fins with fringing fulcra. 

Pseudobeaconia is also extremely similar to Praesemionoius 
from the Lower Triassic of Germany (Jorg, 1969). There are 
howexer, two characters mentioned in the diagnosis of Prae- 
semionoius that do not appear in Pseudobeaconia. They are 
an interopercular and a narrow preopercular that separates the 
opercular series from the suborbital bones (Jorg, 1969: 89). 
The presence of tl;iese characters led to the classification of 
Praesemionoius in the order Parasemionotiformes. Jorg has 
recently revised this opinion (personal communication), and 
now bclie\es that there is no interopercular, and that the preop- 
ercular extends over that part of the cheek prexiously described 



1972 TRIASSIC FISH FROM ARGENTINA 19 

as being composed of both preopercular and suI)orbital bones. 
There Ts Uttle doubt that Prncsemionolus is a perlcidiform, and 
that it should be referred to the famih' Colobodontidae. 

In conclusion, therefore, Pseudobeaconia is in no way re- 
lated to the redfieldiiforni Beaconia. It represents the termina- 
tion (as far as is known at present) of a line of colobodontid 
perleidiforms. Although its immediate ancestors are not known, 
they are likely to have been those that also ga\e rise to Men- 
docinia and Praesemionotus. 

The Evolution of the Skull in the Perleidiformes 

The evolution of the holostean feeding mechanism from the 
palaeoniscoid type is well known. The functional adxantages 
of the former have been discussed by Schaeffer and Rosen 
(1961) and by Gardiner (1967b), and reviewed recently by 
Olson (1971).^ As Schaeffer and Rosen have noted (1961: 
203), the modifications that occurred during the palaeoniscoid- 
holostean transition "permitted a considerable adaptive radiation 
in the feeding mechanism at the holostean level." However, 
during the Triassic period, two chondrostean orders, the Red- 
fieldiiformes and the Perleidiformes, underwent a considerable 
adaptive radiation of their own, yet neither group achiexed the 
holostean condition. The Redfieldiiformes owed their success 
in part to the evolution of a suctorial habit of feeding (Schaef- 
fer, 1967; Hutchinson, 1972), and it remains to examine the 
Perleidiformes to see if there are equally obvious reasons for 
their success. 

The skulls of four perlcidiform genera are compared in Fig- 
ure 9. They are not directly related to one another, but they 
are members of a radiation that must ha\e had a beginning 
sometime during Upper Permian times. In an earlier radiation, 
haplolepid genera show parallel evolution of an almost ver- 
tical suspensorium from an oblique one (VVestoll, 1944) ; in the 
perleidiforms the skull structure shows even more variation. 
Three main trends in their evolution can be traced : 1 . The 
e\olution of an antero\entrally inclined suspensorium from one 
that is posteroventrally inclined. 2. The shortening of the lower 
jaw from a length equal to 68 percent of the anteroposterior 
length of the skull (Perleidus madagascariensis) to one only 
55 percent of the length of the skull (Pseudobeaconia bracac- 
cinii). 3. The ele\ation of the anterior end of the upper jaw. 

Associated with these changes are alterations in the shape of 



20 



BREVIORA 



No. 398 



the maxillary-preopercular complex, and a shift of the level of 
the jaw articulation from a point well behind the orbit to a 
point below the orbit. 

The advantage of a suspensorium that is anteroventrally in- 
clined has been demonstrated by Schaeffer and Rosen; it al- 
lows greater posterior and lateral expansion of the orobranchial 
chamber (1961: 191). However, evolution of such a suspen- 
sorium does not necessarily involve significant shortening of the 
lower jaw (See, for example, the haplolepids [Westoll, 1944: 
fig. 44] ) , and the latter phenomenon in perleidids cannot there- 
fore be regarded simply as a corollary of the former. 

The advantage of a short lower jaw over a longer one is 







Figaire 9. Restorations of the skulls and pectoral girdles of rep- 
resentatives of four perleidifoi-m genera to show variation in shape 
of the preopercular, maxilla, and dentary. Ornament and sensory 
canals omitted. Not to scale. 

A. Perleidus madagascariensis. Lower Triassic, Madagascar. 
From Lehman, 1952, fig. 85. 

Mendocinia brevis. Lower Triassic, Argentina. From Schaef- 
fer, 1955, fig. 4. 

Cleithrolejndina exto'ni. Lower Triassic, Oi-ange Free State. 
From Hutchinson, 1972, fig. 41. 
D. Psenxiobeaconia bracaccinii. Low^er Triassic, Argentina. 



B. 



C. 



1972 TRIASSIC FISH FROM ARGENTINA 21 

that it can exert a greater force when adducted, assuming other 
factors such as the mass of the adductor muscle are constant. It 
may be concluded, therefore, that changes in the perlcidid skull 
architecture are related, not only to the e\'olution of more effi- 
cient abduction of the orobranchial chamber, but also to the 
need to exert a powerful force on adduction of the lower jaw. 
It is not difficult to discover why this last requirement was of 
great adaptive significance to the perleidids. The group is char- 
acterized by the presence of low rounded teeth on the palate 
(Perleidus, ?Colobodus, and Pseudobeaconia) and on the lower 
jaw (Perleidus, Colobodus, and Aieridensa). These teeth sug- 
gest that an important element of the perleidid diet consisted 
of food that required crushing, such as small molluscs and crus- 
taceans. 

The third trend in perleidid evolution, the gradual elevation 
of the anterior end of the upper jaw, is probably related to the 
evolution of shorter jaws. The disad\antage of short jaws is 
that the gape is reduced, but this appears to have been com- 
pensated for by elevation of the anterior end of the maxilla. 

This brief analysis suggests that an important reason for 
perleidid success is that the evolution of short jaws and crushing 
teeth enabled them to feed, in part, on prey not accessible to 
other forms. This interpretation is not complete, however, for 
it does not explain the success of the Cleithrolepididae, which 
have rather slender lower jaws and few teeth ( Fig 9C ) . Mem- 
bers of this family, although early offshoots from perleidid stock 
(Hutchinson, 1972), have an advanced type of suspensorium 
and laterally flattened bodies, and must have occupied a some- 
what different niche to that of other perleidids. 

The Structure of 
THE Unpaired Fins in the Perleidiformes 

The detailed structure of the unpaired fin rays in Pseudobea- 
conia shows that control of the fin web area was possible. 

The unpaired fins of modern selachians are similar to those 
of chondrosteans in that each radial is associated with a lars^e 
number of ceratotrichia (lepidotrichia in chondrosteans). In 
selachians, lateral movement of the fin web is effected by con- 
traction of radial muscles that originate on the radials and that 
have broad insertions on either side of the proximal ends of 
the ceratotrichia. The area of the fin web cannot, however, be 
altered., and the fins cause considerable drag even when not 



22 BREvioRA No. 398 

being used to aid stability. The unpaired fins of Acipenser have 
essentially the same structure and restricted range of mo\ements 
as have those of selachians (Alexander, 1967: 38), and the same 
is almost certainly true of the fossil chondrosteans in which the 
number of rays exceeds the number of radials. 

In the unpaired fins of teleosts a discrete musculature, which 
originates on the radials, enables erection, depression, and lat- 
eral movement of each ray. Such a system is only possible be- 
cause the numbers of rays and radials are equal. 

Westoll (1944: 83) has suggested that control of the fin 
web area was possible in haplolepids because in the Haplolepi- 
didae the unpaired fin rays are small in number, stout, and bev- 
eled along their anterior edges to facilitate overlapping when the 
web is collapsed. In Pseudobeaconia there is even more evi- 
dence which suggests that in that genus, and possibly in all 'sub- 
holosteans' in which the unpaired fin rays and radials are equal 
in number, the fins were capable of almost as much mobility as 
is seen in teleosts. 

The dorsal fin of P. bracaccinii has been described above 
( Fig. 5 ) . The rays and radials are equal in number, a condi- 
tion which suggests that a discrete musculature between the two 
was present. This is confirmed by the presence of an anterior 
and a posterior depression at the proximal end of each ray. 
The anterior depression is here interpreted as the point of in- 
sertion of an erector muscle, and the posterior depression as that 
of a depressor muscle. Anteriorly, the erector muscle insertions 
become progressively larger and the depressor insertions pro- 
gressively smaller. Because of their position and greater length, 
the anterior rays are more affected by water forces that tend 
to collapse the fin. As a result, more effort is required to erect, 
and less to depress, these anterior rays, and these requirements 
are reflected by the variation in size of the erector and depres- 
sor muscle insertions. There is no sign of any depression \vhich 
would indicate the presence of an inclinator muscle, but inclina- 
tion of the fin is theoretically possible if the erector and de- 
pressor muscles of one side are contracted simultaneously. Such 
a mechanism does in fact occur in modern Siluriformes (Alex- 
ander, 1967: 40), but cannot be confirmed in Pseudobeaconia 
because the state of preservation of the joint between each lepi- 
dotrichium and radial does not allow an accurate assessment of 
the degrees of freedom of the ray. 

The grooves that run along the anterior and posterior edges 
of the rays in P. bracaccinii (Fig. 5) probably helped to sup- 
port the soft fin web tissues. 



1972 TRIASSIC FISH FROM ARGENTINA 23 

ACKNOWLEDGMENTS 

I am grateful to Dr. C. Barry Cox of the Department of 
Zoology, University of London, King's College, for the loan of 
equipment which made this work possible and for reading and 
criticizing this paper; also to Maite Lores, who translated many 
of the references and typed the manuscript. My best thanks, 
howe\er, are due to Professor Alfred S. Romer, who kindly 
made available the material described in this paper. 

REFERENCES 

Alexander, R. McN. 1967. Functional Design In Fishes. London: 
Hutchinson University Library. 160 pp. 

BONAP.^RTE, J. F. 1966a. Chronolog-ical survey of the tetrapod- 
bearing Triassic of Argentina. Breviora, No. 251 : 1-13. 

. 1966b. Una nueva "fauna" Triasica de Argen- 
tina (Therapsida: Cynodontia: Dicynodontia). Consideraciones 
filogeneticas y paleobiogeograficas. Ameghiniana, 4: 243-296. 

1969. Cynognathus minor n. sp. (Therapsida: 



Cynodontia). Nueva evidencia de la vinculacion faunistica Afro- 

Sudamericana a principios del Triasico. In Gondwanaland 

Stratigraphy. LU.G.S. Symposium, Buenos Aires, 1967. Paris, 

UNESCO, pp. 273-281. 
BORDAS, A. F. 1944. Feces triasicos de la quebrada de Santa Clara 

(Mendoza y San Juan). Physis, 19: 454-460. 
Gardiner, B. G. 1967a. Further notes on palaeoniscoid fishes with 

a classification of the Chondrostei. Bull. Brit. Mus. (Nat. Hist.), 

Geology, 14: 143-206. 
1967b. The significance of the preoperculum in 

actinopterygian evolution. J. Linn. Soc. (Zoology), 47: 197-207. 
Harrington. H. J. Descripcion geologica de la Hoja 22c — Ram- 

blon Provincias de Mendoza y San Juan. Unpublished MS: 

1-62. 
Hutchinson, P. 1972. A revision of the redfieldiiform and per- 

leidiform fishes from the Triassic of Bekker's Kraal (South 

Africa) and Broolcvale (New South Wales). Bull. Brit, Mus, 

(Nat. Hist.), Geology, 22: 197-318. 
JoRG, E. 1969. Eine Fischfauna aus dem Oberen Buntsandstein 

(Unter-Trias) von Karlsruhe-Durlach (Nordbaden). Beitr. 

Naturk. Forsch. SudwDtl., 28: 87-102. 
Lehman, J. -P, 1952. fitude complementaire des poissons de I'Eo- 

trias de Madagascar. K. Svenska Vetensk. Akad. Handl., 2: 

1-201, 
MiNOPRio, J. L. 1954. Theriodonte en el Triasico de Mendoza. An. 

Soc. Cient, Argentina, 157: 31-37. 
Olson, E. C. 1971. Vertebrate Palaeozoology. New York: Wiley- 

Interscience, 899 pp. 



24 



BREVIORA 



No. 398 



ROMER, A. S. 1966. The Chanares (Argentina) Triassic reptile 

fauna. I. Introduction. Breviora, No. 247: 1-14. 
SCHAEFFER, B. 1955. Mendochiia, a subholostean fish from the 

Triassic of Argentina. Am. Mus. Novitates, No. 1737: 1-23. 
. 1967. Late Triassic fishes frojn the Western 

United States. Bull. Am. Mus. Nat. Hist, 135: 285-342. 
. , AND Rosen, D. E. 1961. Major adaptive levels in 



the evolution of the actinopterygian feeding mechanism. Am, 

Zool., 1: 187-204. 
Stipanicic, p. N. 1967. Triasico. 1st Internat, Symposium Gond- 

wana Stratigraphy Palaeontology. Mar del Plata, 1967. 9 pp. 
Wade, R. T. 1935. The Triassic Fishes of Brookvale, New South 

Wales. London: British Museum (Nat. Hist.). 110 pp. 
Westoll, T. S. 1944. The Haplolepidae, a new family of late 

Carboniferous bony fishes. A study in taxonomy and evolution. 

Bull. Am. Mus. Nat. Hist., 83: 1-122. 

ABREVIATIONS 



aop 


accessory opercular 


adp 


accessory dermopterotic 


an 


angular 


ant 


antorbital 


b 


branchial arch element 


br 


branchiostegal ray 


el 


cleithrum 


clav 


clavicle 


d 


dermohyal 


de 


insertion of depressor muscle 


den 


dentary 


dp 


dermopterotic 


ds 


dennosphenotic 


ecomm 


ethmoid commissure 


er 


insertion of erector muscle 


ex 


extrascapular 


f 


fulcral scale 


fr 


frontal 


g 


groove 


inf 


infraorbital 


(1) 


left side 


mx 


maxilla 


na 


nasal 


no 


nostril border 


op 


opercular 


pa 


parietal 


par 


parasphenoid 


pmx 


premaxilla 


po 


postorbital 


pop 


preopercular 


pro 


postrostral 


pspi 


postspiracular 


q.j 


quadratojugal 


r 


radial 


ro 


rostral 


sb 


suborbital 


so 


suprascapular 


scl 


supracleithrum 


so 


supraorbital 


sop 


subopercular 









B R E V I O R A/ 

Museom of Comparative Zoology 



Cambridge. Mass 



March 6, 1973 



Number 399 



THE E\ OLUTION OF MAMMALIAN 
FROM REPTILIAN DENTITIONS 

J. ^V. OSBORN^ AND A. W. CrOMPTON^ 

Abstract: A motlel to explain the evolution of a typical 
mammalian tooth replacement pattern is proposed. Osborn (1971) 
has described the early development of the teeth and replacement 
pattern in a typical modem reptile (Lacerta vivipara). The re- 
placement of the postcanine teeth in an early Triassic mammal-like 
reptile, a cynodont, Thrinaxodo'n liorhinus, is basically the same as 
that of Lacerta vivipara except that tJie number of replacements 
closely approaches the mammalian condition. The rate of replace- 
ment has slowed down considerably and towards the back of the jaw 
only three teeth develop at each locus. Three different types of 
postcanine teeth can be recognized, an A tjV^ with a simple crown 
pattern towards the front of the row, an M type in the midsection 
with a more complex pattern, and a P tyi^e with the most complex 
pattern towards the back of the row. A P type is replaced by an 
M type and the latter by an A tjTpe. The pro^essive reduction in 
the complexity of the teeth produced at each locus in Thrinaxodon 
and in mammals (deciduous molar as compared with the premolar 
replacing it) is ascribed to the aging of the dental lamina. The 
typical mammalian pattern can be derived from that of Thrinaxo- 
don by further reducing the rate of replacement. 

INTRODUCTION 

Two theories ha\'e been proposed to explain the e\okition 
of mammalian dentitions, in which teeth are only replaced once 
(diphyodonty), from reptilian dentitions in which teeth are 
replaced throughout life (polyphyodonty), Bolk (1922) con- 
cluded that reptilian dentitions consist of two sets of alternat- 



'Anatomy Department, Guy's Hospital Medical School, London 
SEl, 9RT, England 

'Museum of Comparative Zoology, Harvard University, Cambridge, 
Massachusetts 



BREVIORA 



No. 399 



Ai 



eje|e|ejejej.eieAeAe|e 
• •••••••oo 

ooooooooooo 
oooooooooo 



All 



eeeeeeeeeee 



Bi 




o 






o 



e 



°-°»:» 



Bi 



eeeeeeeeeee 



Ci 




Cii 



'/ '^ 'j C Dm/ Dm^ Dnrij Dm^ M/ M^ M^ 



h h h c 



Pm^ Pnrij Pm^ 



FigTire 1. Ai represents Bolk's concept of the sets of alternat- 
ing teeth present in a reptile. All members of a set develop at the 
same time. Aii shows what he considered to be the homologies be- 
tween polyphyodont reptilian and diphyodont mammalian dentitions. 
Bi (reptile) and Bii (mammal) illustrate Edmund's (1960, 1962) 
concepts. He suggested that teeth develop in the sequence showTi 
by the arrows. (Time is represented by a vertical axis.) Ci shows 
the actual sequence of tooth development in a primitive reptilian 
dentition (note alternation of teeth). Cii (arrows) shows the actual 
sequence in mammalian postcanine dentitions (Osborn, 1970). In 
all diagrams anterior is to the left and the mouth above. 



1972 MAMMALIAN DENTITIONS 3 

ing teeth, each set being replaced alternately throughout life 
(Fig. lAi). He suggested that parts of a mammalian dentition 
is equi\'alent to one set that has become pushed together and 
that the replacing teeth are equivalent to the other alternat- 
ing set (Fig. lAii). Edmund (1960, 1962) concluded that 
reptilian teeth are initiated in sequence from the front to the 
back of the jaw in rows which he called Zahnreihen (Fig. IBi). 
He suggested that the two mammalian dentitions are equiva- 
lent to two reptiHan Zahnreihen (Fig. IBii) . 

It has recently been demonstrated (Osborn, 1971) that in 
an embryo reptile teeth are initiated in alternating sequences 
from the back to the front of the jaw (Fig. ICi), while in the 
postcanine dentitions of embryo mammals (Osborn, 1970), the 
first (most anterior) deciduous molar is always the last of the 
deciduous molars to develop and only the permanent molars 
develop in sequence from the front to the back of the jaw (Fig. 
ICii). We conclude that neither Bolk's nor Edmund's theory 
is acceptable because neither takes into account the sequences 
in which teeth actually de\elop. 

One other feature clearly distinguishes between the dentitions 
of mammals and primitive reptiles: in mammahan dentitions, 
the postcanine teeth generally have difTerent shapes (hetero- 
donty) whose complexity reaches a peak somewhere within the 
permanent molar series; primitive reptilian teeth are generally 
all peg-shaped and alike (homodonty). 

We are concerned in this paper with the phylogenetic process 
by which the heterodont, diphyodont dentitions of mammals 
e\olved from those of the homodont, polyphyodont dentitions 
of primitive reptiles. In the absence of any recent animal which 
has a dentition e\en remotely intermediate between these two 
extremes, a solution to these problems must either be entirelv 
speculative (Bolk, 1922; Edmund, 1960, 1962) or based on 
e\'idence from the fossil record. The gap between mammals 
and reptiles was bridged by the mammal-like reptiles of the 
Triassic. For our study we have chosen Thrinaxodon, whose 
tooth morphology and pattern of tooth replacement have al- 
ready been investigated (Crompton, 1963; Parrington, 1936), 
but not in sufficient detail to answer the problems which we have 
posed. 

POST-CANINE TEETH OF THRINAXODON 

Preliminary Arrangement of Material 

For the present in\estigation ten extremely well-preser\ed, 
as yet undescribed, Thrinaxodon jaws have been studied, to- 



4 BREVIORA No. 399 

gether with four of those described by Crompton ( 1 963 ) , two 
of which were available to Parrington ( 1936) . These have been 
labelled A to N. The investigation was confined to the lower 
postcanine teeth. 

In order to arrange these specimens in a growth series, photo- 
graphs, reproduced at the same magnification, were taken of 
the lingual sides of the jaws. Tracings were made of the jaw 
outlines of all the available material. These tracings were com- 
pared by superimposing one on another. It was necessary to 
speculate on the growth processes of the mandible to make ac- 
curate comparisons between the positions of the postcanine den- 
titions in jaws of different sizes. These speculations will be 
described later. On the basis of this study we arranged the 
specimens in a tentative growth sequence. 

We speculated that the majority of growth in length of the 
jaw took place at the posterior end of the dentary where room 
was being made for the addition of postcanine teeth (Crompton, 
1963) and that there was only a hmited subperiosteal deposi- 
tion of bone at the anterior end of the jaw (Fig. 2, cf. Fig. 7). 
From this we now concluded that successive replacement incisors 
remained in approximately the same anteroposterior position in 
relation to a hypothetical fixed marker, or that they erupted in 
only a slightly more anterior position than their predecessors. 
Successive canines might have become more posteriorly located 



Figure 2. A diagTam in which growth processes of the lower 
jaw of Thrinaxodon are postulated. Uninternapted line = younger 
animal; interrupted line = older animal. The jaw has increased in 
length by the deposition of bone at the borders of the dentary. The 
anterior border of the coronoid process has been resorbed, making 
room for the addition of anotlier postcanine in the older animal. 
There has been very little deposition of bone at the anterior end of 
the jaw. The most anterior of the post^-anines in the younger ani- 
mal has been lost in the older ani^nal. 



1972 



MAMMALIAN DENTITIONS 



because we obserxed that the tip of a replacing canine was gen- 
erally located posterior to the functional canine. However, we 
finally decided that as a canine erupted it moved into a more 
anterior position than would ha\ e been predicted from the posi- 
tion of its erupting tip ( Fig. 7 ) . 

The tentative growth sequence established in this way was 
modified following a study of the shapes of the teeth and the 
positions of replacing teeth. For this purpose the sections of the 
prints containing the postcanine teeth were cut from full prints 
of the jaws. These small prints were rearranged until the best 
correspondence between adjacent dentitions was achieved, bear- 
ing in mind the original tentative arrangement based on jaw 
sizes. 

Tooth Shape in Thrinaxodon 

It appeared that the preliminary arrangement was in many 
cases inaccurate because frequently there was poor correspond- 
ence between what initially appeared to have been equivalent 
teeth in jaws which had been judged as being of comparable age 
( based on the similarity of their sizes ) . By comparing tooth 
shapes and by making estimates of the ages of teeth based on 
Crompton's (1963) observations, a definitive growth sequence 
was established. Before presenting this we will briefly describe 
the shapes of the teeth. 

We finally established that Thrinaxodon possessed only three 
distinctive postcanine tooth types. It was because variations of 
each tooth type were present that Crompton (1963) did not 
recognise the division into three tooth types. W^e will refer to 
these as A = anterior, M = middle, and P =r. posterior. Be- 
tween two and four A types (Figs. 3, 4) are always present at 



LeftE Right B LeftE LeftE LeftE LeftB Right B 

position4 positions positions position? positions position? positions 




M types 



P types 



A types 

Figure 3, The three tooth types defined in this study. P = pos- 
terior, M = middle, A =r anterior. More detailed descriptions of 
the postcanines in Thrinaxodon are given by Crompton (1963). 



6 BREvioRA No. 399 

the front of the postcanine row. The crown is dominated by 
the main cusp and flanked by very small anterior and posterior 
accessory cusps. Lingual cingulum cusps are poorly developed 
or absent. Frequently the main cusp is markedly tilted poste- 
riori)-. The most anterior A type is often extremely small (Fig. 
4). Either two or three M types (Figs. 3, 4) are always present 
in the middle of the postcanine row. The main cusp dominates 
the tooth but accessory cusps are usually situated much higher 
up its flanks than in the A type, reducing its apparent height 
when compared with the equivalent cusp on the A types. Cingu- 
lum cusps are generally well developed, producing a frill around 
the lingual surface of the tooth. The postcanine dentition ter- 
minates with either two or three P types (Figs. 3, 4) (dentition 
E contained only one ) . Included with these in the older animals 
may be found a "hybrid" MP type situated at the back of the 
dentition {e.g., H, I, J, M in Fig. 4). These MP types will be 
considered with the P types because they are present at the back 
of the jaw. Their significance will be discussed later. In the 
P types the main cusp is very much reduced in height by large 
anterior and posterior accessory cusps situated high up on its 
flanks. Well-developed anterior and posterior cingulum cusps 
are usually placed at the ends of the tooth in such a way as 
considerably to increase its anteroposterior proportions when 
compared with A and M types. A well-de\eloped array of 
lingual cingulum cusps gives the tooth what approximates to 
an occlusal surface although it did not occlude with the upper 
teeth. The occasional MP type has, as might be expected from 
the terminology, a shape intermediate between M and P types. 
From the above definitions it can be visualised that the sizes 
of the anterior and posterior accessory cusps 1 ) increase from 
the front to the back of the dentition and 2) are directly re- 
lated to the prominence of the Ungual frill of cusps. In all the 
present material and in all well-preserved material which we have 
subsequently examined the postcanine dentition begins with be- 
tween two and four A types which can be clearly distinguished 
from the succeeding two or three M types. These M types can 
be clearly distinguished from the P types which terminate the 
dentition. 

Figure 4. Fifteen postcanine quadrants described in this study 
arrang-ed in the suggested age sequence. The prints of the left jaws 
were reversed to simplify comparisons between dentitions. A and P 
types are shaded. Developing teeth are shown in the appropriate 
positions. 



1972 



MAMMALIAN DENTITIONS 



1 i2 3 1 4 5 I 6 I 7 I 8 I 9 




« 



BREVIORA 



No. 399 



The Definitive Arrangement of Specimens 

We finally decided on the sequence of dentitions shown in 
Figure 4. This can be compared with Figure 5 in which each 
postcanine has been allotted a tooth form A, M, or P and Fig- 
ure 6, which is a simplified version of Figure 5. Mandibular 
sizes are shown in Figure 7. 

1 2 3 4 5 6 7 8 9 10 11 12 



A 
B 
C 
D 
E 
F 
G 
H 

I (Right) 

J 

K 

L (Left) 

M 

N 



a a [a] rri m P. -^ 



? ?A ? a rriA m PA p ^ 
-^ -^^ a 7 7a ? 7. 7 A 



a aA 
^ a 



X 
X 

X 

X 

X 

X 
7 



a^ a m^ m p^ p 



a^ a^m 



m 



M 



m. m 



.fiA 



m 



Pa 



E 



a 
a. 



rriA m 



P P 



a -^m. rn p. p 



x^ a 
7 _ 



a a^ ^ 

X a aA a mA 
■? 7 ? 7 m? 



X X 

m^ m p^ p 
m^ m^p 

A YA ~ 



m^p 



m 



a nrij^ m 



m 



X B o JDAJILm n. _B_ 



aA 
aA a 



mA 

XA 



Pa P 
m p^ 



m 



■^p • 



1972 MAMMALIAN DENTITIONS 9 

In support of this sequence the tooth types present in each 
dentition in positions 4 to 9 should be studied from Figures 4 
and 5, 

Position 4 : The M type of dentition A is replaced in denti- 
tion B b\ an A type. This A type is replaced twice b>- further 
A types. Only a root is present in dentitions M and N. 

Position 5 : The M type of dentitions A and B is replaced 
in C by an A type, which is replaced once by another A type. 

Position 6: The P type of dentition A is replaced in B by 
an M type. This tooth can be traced through to G being re- 
placed in I by an A type, which can be recognised in the remain- 
ing dentitions. 

Position 7: The P type of dentitions A and B is replaced 
in C by an M type, which can be recognised through to denti- 
tion L. On one side of dentition M and probably of N it has 
been replaced by an A t)pe tooth. 

Position 8: The P type in dentitions A to D is replaced in 
E by an M t\pe. This M t\pe can be recognised through to 
the last dentition. The small size of the P type in dentitions A, 
B, and C will be referred to later. 

Position 9: In C a P type has erupted in the equivalent 
alveolus present in B. This P type can be recognised up to den- 
tition K. In dentitions L, M, and N it has been replaced by 
an M type tooth. 

Figure 5. The dentitions of all the specimens studied are rep- 
resented in the form of an ag-e sequence. Specimen identifications 
are to the left and tooth positions above. Where the teeth on left 
and right sides of a dentition differed the right side is represented 
above a line and the left side below the line. X represents a badly 
mutilated or lost tooth in an intact region of the jaw. The query 
marks represent teeth that were probably present in regions from 
which the jaw was missing. The triangles represent developing 
t«eth, the ovals are the crypts which contained them. The origin of the 
material is as follows: A =r TM 80; B = TM ?; C = TM SOB; 
D = BMNH R3731 = Crompton's specimen C := Pai-ring-ton's spec- 
imen E; E = NM C354; F = SAM; G = NM C318; H = AMNH; 
I = SAM K377 = Crompton's specimen D; J = TM 180; K = 
BMNH R511 =: Crompton's specimeii F; L = TM 81; M = SAM 
?; N = BMNH R511A = Crompton's specimen H = Parrington's 
specimen I. TM = Transvaal Museum; BMNH = British Museum 
(Natural History); SAM = South African Museum; NM ^ Na- 
tional Museum (Bloemfontein) ; AMNH = American Museum of 
Natural History. 



10 BREVioRA No. 399 

In each of the above sequences it is important to recognise 
that correct aging of teeth has been observed. Consider position 
7. The young P type in dentition A is old in B. The replacing 
M type in C is young (there is no replacing crypt). It is not 
until dentition H that a crypt is observed beneath this now 
aging M type tooth. On one side of dentition M (Fig. 5) and 
probably of N (Fig. 4) it is replaced by an erupting A type. 

Tooth Replacement in Thrinaxodon 

Before discussing the present data we must point out that it 
is by no means certain that all the animals studied belonged to 
a homogeneous population. The discrepancies between denti- 
tions, particularly the smallness of the P types in position 8 of 
dentition A, position 9 of C, D, and E, position 10 of G, and 
position 1 1 of I and J could represent terminal stages in the 
growth series of animals of different sex or species. Nevertheless 
we will treat the data as if the animals belonged to a homoge- 
neous population. 

If the growth series presented here is correct then the pattern 
of tooth replacement in the heterodont postcanine dentition of 
Thrinaxodon conforms precisely with that described by Ed- 
mund (1960) for the majority of the homodont dentitions of 
nonmammalian vertebrates. This is shown in Figure 8B in 
which the numbering of the tooth positions corresponds to that 
in Figures 4, 5 and 6. By adding two further positions on the 
left (Po and P-i) we calculate that it is possible to fit Cromp- 
ton's (1963) very small specimen A, in which only the maxil- 
lary teeth are visible, into the series. 

In most reptiles the actual number of tooth replacements at 
each tooth position is determined first by the longevity of the 
animal and second by the rate at which replacement teeth are 
initiated. Compared with other reptiles the rate of replacement 
of postcanine teeth in Thrinaxodon appears to have been so re- 
duced in proportion to its longevity that posterior to position 6 
at most only three teeth occupy each position. The number of 
replacement teeth in the anterior postcanine positions are limited 
in the following way. The replacing canines developed very 
close to the roots of the most anterior A types (Hopson, 1964). 
Hopson (ibid.) suggested that the presence of the developing 
canine prevented the development of replacements for these A 
types. The result was that even in the anterior postcanine posi- 
tions, at most fi\'e teeth (position 4) erupted in any tooth posi- 



1972 



MAMMALIAN DENTITIONS 11 

1 2 3 4 5 6 7 8 9 10 11 12 



A Right a a^ ^ ^a '^ Pa P fP 
B Left 



C Right *1 

D Right 

E Left 

F Left*2 

G Right 

H Left 

I Right 

J Left*3 

K Right 

L Left 

M Left*4 

N Right *5 



a a, 



m^ m p^ p • 
a a^ a a m^ m p^ p 
X a a^ a m^ m p^ p 



X aj^ a m^ m [mj Pa ^ 
X a a m^ m [m] p^ ^ 
X a amj^m[m]p p 



X a^ a [aj m^ m p^ mp 

m^?m Pa P [p 



X x^ a 



a a^ a^ [aj ^A f^ Pa P EE 
X a aA a rriA m Pa p ^ 

Pa P 



a a a n\ m 



m 



X a^ a 



X a. 



m^ m p^ mp 



aj Xa m Pa P 



*1 2-4 left 

*2 Position 10 right 

*3 Position 9 right 

*4 Position 11 right 

*5 Positbn 10 left 

Figure 6. The information in Fig. 4 is simplified in this dia- 
gram. Where a tooth is missing or unrecognisable on one side of the 
jaw the corresponding tooth on the intact side of the dentition is rep- 
resented. In the few cases where the left and right dentitions of an 
animal are different the tooth form that more closely fits the predicted 
sequence is shown. A box indicates that the tooth was not yet fully 
erupted. A triangle represents an erupting tooth, an oval represents 
an alveolus. 



12 



BREVIORA 



No. 399 




Figure 7. Tracings of the jaws and postcanine dentitions from 
one side of all specimens except I, M, and N for which suitable 
photogi-aphs were not available. The tracings have been arranged 
in accordance with the growth processes suggested in Fig. 1. 



tion. Thus, although Thrinaxodon was strictly polyphyodont 
(because the dental lamina was active in budding new teeth 
through the life of the animal) the number of replacement teeth 
was severely limited when compared with most modern reptiles 
which have been studied. 

The following argument suggests that Thrinaxodon may have 
been evolving towards a condition in which the activity of the 
dental lamina terminated during the life of the animal. This is 
the characteristic which distinguishes polyphyodont from other 
dentitions. In the anterior postcanine tooth positions the last 
replacement teeth were so small (position 3 in dentitions D to 
K; position 4 in dentitions L to N) that they can hardly have 
had any functional value. Thus, it can be argued that, despite 
its continued presence, the dental lamina in these positions no 



1972 MAMMALIAN DENTITIONS 13 

longer contributed significantly towards the maintenance of an 
efficient dentition. This is likely to be the first stage in the evo- 
lution of non-polyphyodont {e.g., diphyodont) from polyphyo- 
dont dentitions. 

The same type of process appears to have taken place at the 
back of the dentition. Instead of bulky P type teeth, xery re- 
duced hybrid MP types were erupting at the back of dentitions 
H, I, and J: once more, the dental lamina seems to be showing 
a reduced activity in the adult animal. In conclusion, although 
Thrinaxodon was undoubtedly polyphyodont, the adult ani- 
mal shows a tendency toward producing small teeth of lim- 
ited value at the front and back of the dentition. It is as if the 
dental lamina was becoming senescent during vigorous adult 
life, a feature which is not typical of a polyphyodont dentition. 
As described in the next section, this possibility may have impor- 
tant implications for determining the shapes of successional 
teeth. 

Tooth Morphology in Thrinaxodon 

It will be observed that in each tooth family there is a pro- 
gressive simplification of cusp detail starting with the complex 
P type and ending with the very simple A types. This was ac- 
counted for by Crompton (1963) in the following way. Basing 
his argument on the concept of embryological fields ( Butler, 
1939), he proposed that several morphogenetic fields moved 
anteroposteriorly along the jaws of Thrinaxodon. In terms of 
the present observations he suggested that initially a "P type" 
field traxelled slowly backwards through the jaw. Under its 
influence the first tooth in each tooth position developed into a 
P type. Following closely behind this was an "M type" field 
which induced the second tooth in each tooth position to de- 
velop into a M type. In turn this was followed by two or three 
A type fields. However, there is a more simple and biologically 
probable explanation for the present observations. We speculate 
that in each tooth position there was a gradual decrease in the 
morphogenetic capacity of the dental lamina and that this was 
an aging phenomenon. A newly differentiated "segment" of 
the dental lamina was capable of inducing the development of 
a complex P type tooth. This same "segment" of dental lamina 
now aged in such a way that the second tooth it produced was 
the less complex M type tooth. If time allowed, further aging 
resulted in its producing successively less complex A types. This 



14 



BREVIORA 



No. 399 



A 



I 3 5 
2 4 6 

I 3 5 
2 4 6 

I 3 5 
2 4 6 



7 9 \ 
8 10 \ 

7 9 II ^ 
8 10 12 

7 9 II 13 
8 10 12 14 



7 8 9 10 



B 



c 




Hatchling 



Theoretical 



Dentition A 



a./ 0/ Aj M5 Py 

Op A 4 M 5 Pg 

Qj A5 M^ Pg 

O4 ^6 ^8 P/0 



M. 



" Dentition M 




Figure 8. "A" shows the sequence in which teeth develop in a 
primitive reptile (Osbom, 1971). Teeth equivalent to those at the 
right of the interrupted line may have evolved into postcanines in 
ThHnaxodon (B). A mammalian dentition may be represented by 
tlie teeth enclosed within the box in B. Equivalent teeth are given 
mamjnalian nomenclatures in C. 



1972 MAMMALIAN DENTITIONS 15 

same sequence of aging was reproduced sequentially in each 
segment of dental lamina as it extended backwards through the 
jaw into each new tooth position. In other words, heterodonty 
in the postcanine dentition of Thrinaxodon was achieved by a 
genetic process linked to the phenomena of aging rather than to 
the diametrically opposed phenomena of embryology (typified 
amongst other things by morphogenetic fields). Such a hypo- 
thesis could have considerable e\olutionary significance, for it 
will be noted that if the postulated senescence of the dental 
lamina were to be accelerated, a stage would be reached at 
which the dental lamina was no longer capable of initiating a 
new tooth. Such a dentition ceases to be polyphyodont and 
suggests a possible origin for diphyodonty. 

It is now necessary to allot a functional significance to the 
waning of the potential of the dental lamina to induce the de- 
velopment of cusps in the replacement teeth of Thrinaxodon. 
Newly hatched reptiles require a functional dentition. Thus 
even in the youngest animals strongly cusped teeth (relative to 
the size of the animal) are required at the back of the jaw 
where maximum power is developed. The teeth in positions 3 
and 4 were probably at the back of the postcanine dentition in 
the new-born Thrinaxodon (Fig. 8B). Thus P types would 
initially have been present here and throughout life new P types 
would have erupted in sequence at the back of the jaw. But a 
neonatal Thrinaxodon jaw could only accommodate a relatively 
short and fragile P type tooth in position 4 in comparison to 
the size of the P type tooth which could be accommodated in 
an adult jaw. This same restricted space was a\"ailable for the 
teeth which subsequently occupied position 4 in older animals. 
It is not difficult to visualise that it was an advantage to fill this 
space with a proportionately shorter but more robust M type 
tooth than a proportionately longer and more fragile P type. 
This same space was sufficient to accommodate an e\en more 
robust A type tooth in the adult animal. Thus, by reference to 
Figure 4, it can be seen that in position 6 the A type of denti- 
tion L has roughly the same length as the M type of dentition 
E and the P type of dentition A. However, the final A types in 
positions 3, 4, and 5 were very diminuti\e teeth which had little 
functional significance. 

The most significant feature of the abo\e hypothesis in rela- 
tion to the ontogeny of postcanine teeth in Thrinaxodon is the 
suggestion that the successive tooth types in each tooth position 
were developed in relation to a form-determining stimulus which 



16 BREVIORA No. 399 

was repeated in every section of the dental lamina (or adjacent 
ectomesenchyme ) . No sequential stimulus travelling between 
adjacent segments of the dental lamina, such as that inherent in 
the embryological concept of a morphogenetic field, is required. 
The same difference in interpretation separates the hypotheses 
of Edmund (1960, 1962) and Osborn (1970, 1971) which have 
been put forward to explain the manner in which wave replace- 
ment of teeth develops in lower vertebrates. Edmund (1960, 
1962) suggests that the timing of tooth initiation from the free 
margin of the dental lamina is co-ordinated by an impulse that 
travels backwards through the length of the jaw {cf. a morpho- 
genetic gradient travelling backwards along the length of the 
jaw) . By modifying Edmund's hypothesis Osborn (1970, 1971 ) 
explains wave replacement in terms of a control which is localised 
to each tooth position and which is repeated in every tooth posi- 
tion ( cf. the P, M, and A types repeated in each tooth position ) . 
The absence of any evidence of a caudad impulse or a morpho- 
genetic gradient (in postcanine dentitions) suggest that, in terms 
of biological possibilities, the more simple explanations offered 
here are more probable. 

THE EVOLUTION OF MAMMALIAN DENTITIONS 

It is now possible to speculate on the evolution of mamma- 
lian postcanine dentitions from polyphyodont dentitions (Fig. 
8 ) . We speculate that dental ontogeny in a primiti\e early rep- 
tile was hke that described for a modern reptile (Osborn, 1971 ) 
which retains the archetypal pattern of tooth replacement ob- 
served in fossil reptiles (Fig. 8A). In this figure a canine might 
have evolved at position 6; therefore we are only concerned with 
positions 7 and beyond (the postcanine teeth). We speculate 
that in Thrinaxodon (Fig. 8B) postcanine development might 
have begun at about position 9. Anterior to this, just as in the 
primitive reptile, teeth could have developed alternately from 
back to front although there is no evidence for this. The tooth 
at position 8 developed later than that at positions 9 and 7, and 
owing to aging of the dental lamina, may therefore have been 
an M type rather than a P type. Behind position 9, teeth de- 
veloped in sequence from front to back, just as in the primitive 
reptile. The first teeth at each tooth position were the complex 
P types. It can be seen that the sequence of tooth development 
in Thrinaxodon may have been similar to that in a primitive 
reptile. 



1972 MAMMALIAN DENTITIONS 17 

We now speculate that a primithe mammalian postcanine 
dentition with four deciduous molars, four premolars, and 
three molars was equi\alent to the teeth within the box in Fig- 
ure 8B. Equi\alcnt teeth ha\"e been gixen mammalian nomen- 
clatures in Figure 8C. The sequence of de\elopment of decid- 
uous molars and premolars is from back to front. This sequence 
matches data which show that Dm 4 is usually the first and Dm 1 
is always the last to develop (Osborn, 1970). Furthermore these 
teeth decrease in complexity from back to front just as in the 
box shown in Figure 8B. The Dm 1 in mammals appears ne\er 
to be replaced. This also is a trend which can be seen in the 
equivalent tooth position in Thrinaxodon. The most molari- 
form teeth in mammals develop in sequence from front to back; 
again this agrees with the model in Figure SB. Finally, the pre- 
molars are generally less complex than the deciduous molars and 
this too matches our model ( Fig. SB) . 

SUMMARY 

In a primiti\e reptilian embryo the first tooth in each jaw 
quadrant develops at about the ninth tooth position from the 
front of the jaw. In front of this position teeth dexelop in the 
odd-numbered positions from the back to the front (9, 7, 5, 3, 
1 ) to be followed by a similar sequence in the even-numbered 
positions (8, 6, 4, 2). Behind this position teeth develop in 
sequence from front to back (9, 10, 11, 12, etc.). Teeth con- 
tinue to be replaced throughout life. 

We ha\e confirmed that the pattern of tooth replacement in 
the postcanine dentition of the mammal-like reptile, Thrinaxo- 
don, was similar to that seen in primiti\e reptiles. We specu- 
late that the sequence of tooth dexelopment in embryos was 
also similar to that in a primiti\e reptile. However, the rate 
of tooth replacement was so much slower that at most fi\e, and 
usually fewer, replacement teeth were developed at each tooth 
position. 

It has been shown (Osborn, 1971) that in primitive mam- 
mals either the fourth or third deciduous molar is the first to 
develop and the 1st deciduous molar is the last to develop (i.e., 
development is from back to front as in primitive reptiles and 
Thrinaxodon) . Behind the fourth deciduous molar, teeth de- 
velop in sequence from front to back (again as in the reptiles). 
Thus, by further slowing down the rate of tooth replacement in 
Thrinaxodon, we arrive at the peculiar sequence of tooth initi- 
ation which has been obser\ed in primitive mammals. 



18 BREVIORA No. 399 

In Thrinaxodon the first tooth to develop in each postcanine 
tooth position was the most complex, to be follwed by replacing 
teeth which were necessarily less complex. We speculate that 
this sequence was related to aging of the dental lamina, the 
epithelial ingrowth from which all teeth develop, and further- 
more, that the same is true for mammals. This theory explains 
why both the deciduous molars and premolars become progres- 
sively less complex from back to front; it is the sequence in 
which they develop. Premolars are less complex than the teeth 
they replace (the deciduous molars) because they develop later. 
Permanent molars are complex because they are the first teeth 
to develop at their respective tooth positions. 

REFERENCES 

BoLK, L. 1922. Odontological essays. 5. On the relation between 
reptilian and nianimalian dentition. J. Anat., 57: .55-75. 

Butler, P. M. 1939. Studies of the mammalian dentition. Dif- 
ferentiation of the i:)Ost-canine dentition. Proc. Zool. Soc. Lon- 
don (B), 109: 1-36. 

Crompton, a. W. 1963. Tooth replacement in the cynodont Thnn- 
axodon liorhinus Seeley. Ann, South African Mus., Cape To%\ai, 
46: 479-521. 

Edmund, A. G. 1960. Tooth replacement in the lower vertebrates. 
Roy. Ontario Mas., Life Sci. Contrib., 52: 1-190. 

1962. Sequence and rate of tooth replacement in 

the Crocodilia. Roy. Ontario Mus., Life Sci. Contrib., 56: 1-42. 

HOPSON, J. A. 1964. Tooth replacement in cynodonts, dicynodont 
and therocephalian reptiles. Proc. Zool. Soc. London, 142: 625- 
654. 

OsBORN, J. W. 1970. New approach to Zahnreihen. Nature (Lon- 
don), 225: 343-346. 

1971. The ontogeny of tooth succession in Lacerfn 

vivipara Jacquin (1787). Proc. R. Soc. London, B, 179: 261-289. 

Parrington, F, R. 1936. On tooth replacement in the theriodont 
reptiles. Phil. Trans. Roy. Soc. London (B), 226: 121-142. 



''"^'-'c:. CCMP. 200L 



BREVIORA 

IMoseom of Comparative Zoology 



Cambridge, Mass. March 6, 1973 Number 400 

THE CAROLINA SALT MARSH SNAKE: 

A DISTINCT FORM OF 

A^^ TRIX SIPEDON 

Roger Conant and James D. Lazell, Jr. 

Abstract. Hybridization between two species and intergradation 
between races of one of them has long obscured the nomenclatural 
status of the water snakes of the Natrix sipedon-fasciata group in 
eastern North Carolina. A study of recently acquired material and 
the recognition that hybridization is occurring in several localities 
where fresh- and brackish-water habitats meet make possible a clari- 
fication of the confusing engelsi situation. Because the name engelsi 
was inadvertently applied to a snake of the wrong species, a tax- 
onomic adjustment is made and the name Natrix sipedon william- 
engelsi is erected for the salt marsh snake that occurs on several of 
the islands of North Carolina's Outer Banks and along the mainland 
perimeter of Pamlico Sound and associated estuaries. 

INTRODUCTION 

The confusing status of the water snakes of the Natrix sipedon 
complex from the Outer Banks of North Carolina and the ad- 
jacent mainland has long intrigued both of us. Conant (1963: 
29-32) summarized much of the data available at that time 
under the heading of "brackish-water populations." Lazell, as 
part of studies on the biota of the Banks, has recently accumu- 
lated considerable fresh material that helps to throw light on the 
situation. We ha\'e deemed it advisable to pool our knowledge 
and resolve the "engelsi" problem, which we review in some de- 
tail. 

Three recognizable forms of the Natrix sipedon-fasciata group 
occur in eastern North Carolina. Natrix sipedon and Natrix 
fasciata are distinct species, based on their wide sympatry in the 
Mississippi Valley, along the Gulf Coast, and in northeastern 
North Carolina (Conant, ibid.). In the Carolinas fasciata is 
confined almost exclusi\'ely to freshwater habitats on the Coastal 



2 BREVIORA No. 400 

Plain ; sipedon occurs in the mountains and Piedmont, but it also 
invades the Coastal Plain north of Albemarle Sound, occurs on 
the peninsula between Albemarle and Pamlico Sounds, enters 
and is quite tolerant of brackish-water conditions bordering Pam- 
Uco and Core Sounds, and also occurs on the islands of the Outer 
Banks. In much of this general area it is differentiated at the 
subspecific level, thus providing the third form of the complex 
from eastern North Carolina. 

That pecuhar specimens of Matrix sipedon occur on the Outer 
Banks is a view scarcely novel with us. Barbour ( 1 943 ) described 
Matrix sipedon engelsi on the basis of a single individual taken 
from Mullet Pond on the Shackleford Banks. Mullet Pond, which 
was formerly a bay on the north shore of the island (Engels, 
1952: 704), now has no connection with salt water and is fresh; 
it supported a dense stand of cattails (Typha) and there was 
very little open water when a field party of the American So- 
ciety of Ichthyologists and Herpetologists visited it on 5 Septem- 
ber 1964. The snake Barbour used for his type (MCZ 46688) 
unfortunately is identifiable as Matrix jasciata. We have exam- 
ined this specimen both independently and together, and we 
agree that from all diagnostic morphological characters it is a 
jasciata. The name engelsi, therefore, should be relegated to the 
synonymy of jasciata, a view previously suggested by Conant 
(1961: 19). 

The history of the description of engelsi is of considerable in- 
terest, and we quote from a letter that Dr. William L. Engels 
sent to the senior author under the date of 27 February 1962: 
"Dr. Barbour spent a few days in Chapel Hill in May 1942 .... 
Dr. Coker brought him around to me. I was just finishing the 
manuscript for my report on the Ocracoke fauna, and I showed 
him the king snake I had from there which was certainly a pecu- 
har looking, chocolate-colored individual. I also showed him the 
only other hve snake I had, the rat snake from Shackleford 
Banks. He immediately got quite excited, came back the next 
day and stared at those snakes some more, and wanted me to 
sit down immediately to describe them as new. I was about to 
go into the Air Force, did not have any comparative material, 
and at any rate was totally lacking in experience in systematic 
herpetology. The end of it was that I sent him the snakes and 
went myself into service. He wrote me proposing that either my 
name should be given to one of them or that my name should 
be put on the descriptions as joint author. I agreed to the lat- 
ter, and that ended our collaboration. I dehberately did not 



1973 THE CAROLINA SALT MARSH SNAKE 3 

show him the water snakes — I thought that was a Httle project 
I could keep for myself, for after the war (they did strike me as 
being unusually dark). But Dr. Coker did not know this. He 
did know that Dr. Barbour had become quite excited about 'the 
new herpeto fauna' which I 'had discovered' — the quotes are 
Barbour's words. So, having come into possession of a Matrix 
from the Mullet Pond on Shackelford, while I was in Egypt, he 
sent it to Barbour. That became the type of .V. s. engelsi." 

The practice of describing new taxa from single specimens, 
although always potentially risky, was widespread 30 years ago. 
Consideration should also be given to the fact that Barbour was 
an alpha taxonomist; he seldom examined large series of speci- 
mens, and he was accustomed to describing new species from 
islands of the West Indies. It is understandable that he could 
engender strong enthusiasm about unusual insular specimens. His 
type of engelsi, a \ery dark snake in which black pigment ob- 
scures man\ of the pattern details, is now known to be a rep- 
resentati\e of a deme inhabiting Mullet Pond that shows strong 
exidence of hybridization between sipedon and jasciata. As is 
noted below, some of the snakes of this population are morpho- 
logically indistinguishable from sipedon, others could be identified 
as jasciata, and others combine characters of both. This phe- 
nomenon is not unique to the Shackleford Banks. It also occurs 
at other places where fresh- and saltwater habitats meet and in 
a few localities within the Fall Line zone of both the Carolinas 
and Georgia (Conant, 1963). 

In an effort to end the confusion that has thwarted a real 
understanding of the Matrix sipedon-jasciata complex of the re- 
gion for three decades, we take pleasure in describing: 

NATRIX SIPEDON WILLIAMENGELSI n. ssp. 

Holotype. MCZ 129298, a young adult male, collected by 
Paul Elias, 14 June 1971, along Island Greek on Ocracoke 
Island, Hyde Co., North Carolina. 

Paratypes. (All from North Carolina. ) CORE BANKS, Car- 
teret Co.: North Carolina State Museum (NCSM) 11 796, MCZ 
131993, Cape Lookout. HATTERAS ISLAND, Dare Co.: 
United States National Museum (USNM) 72573; MCZ 
130278-79, Hatteras \^illage. OCRACOKE ISLAND, Hyde 
Co.: MCZ 129290, Horse Pen Creek; MCZ 129299, 129302, 
Island Creek; MCZ 129242, between Island Creek and Pamlico 
Sound; MCZ 131990-91, Ocracoke ViUage; MCZ 131988, 



4 BREvioRA No. 400 

Quork Hammock; MCZ 129288, 129300, Quorks Point Creek; 
MCZ 129249, Old Hammock Creek; MCZ 129250, Pony Pas- 
ture Creek; NCSM 11798, 1 mi. W of The Knoll; NCSM 
11797, Try Yard Creek. PORTSMOUTH ISLAND, Carteret 
Co.: MCZ 129251-74, 129289, Portsmouth Village. MAIN- 
LAND: Carteret Co.: NCSM 8003, 1 1 mi. S of Merrimon; 
MCZ 131992, SW part of Cedar Island. Hyde Co.: American 
Museum of Natural Histor>' (AMNH) 88076, Rose Bay Creek. 
Pamlico Co.: AMNH 88075, Oyster Creek near Lowland. 

Diagnosis. A strongly melanistic race of Matrix sipedon in 
which: (1) the dorsum in adults is essentially black with the 
interspaces between the crossbands and blotches so dark that 
pattern details often can be discerned only if the specimen is 
immersed in hquid; (2) the venter posterior to midbody is pre- 
dominantly black; (3) there are no reddish- or brown-centered 
ventral crescents or half -moons posterior to the 50th ventral; 
and (4) the light scales between the dark crossbands average 
one and one-half (maximum three) on the neck at the level of 
the second dorsal scale row. 

Description of the type. MCZ 129298 is a young adult male 
725 mm in total length; the tail is complete and measures 168 
mm. There are 138 ventrals, 72 subcaudals, and a maximum of 
23 dorsal scale rows. The anal is divided and both hemipenes 
are everted. The general coloration is black with dull gray in- 
terspaces between the black bands and blotches. The interspaces 
vary, at the level of the second dorsal row, from one to two scales 
in width; most are about a scale and one-half. The dark dor- 
sal bands are continuous, from ventral edge to ventral edge across 
the anterior part of the body; they are replaced posteriorly by 
alternating lateral bars and middorsal blotches, as in the nomi- 
nate race. The entire snake is so dark that the pattern is most 
clearly observed under liquid. 

The ventrals are virtually uniformly black at the posterior end 
of the body, but are lightly spotted with white; the white spots 
are larger and much more prominent on the anterior third of the 
body. There are no reddish- or brown-centered ventral crescents 
posterior to ventral 48. 

Variation. We have examined 50 specimens of this race (26 
males and 24 females). Of these, the senior author examined 
three alive, the junior author more than 20 alive or freshly killed 
on the road. The very dark coloration of williamengelsi is little 
affected by preservation, and there is far less variation than in 
the nominate race. All specimens are black or dark brownish 



1973 THE CAROLINA SALT MARSH SNAKE 5 

black with \er\ narrow, lighter, grayer interspaces between the 
daric bands. The bands are so wide relati\c to the interspaces 
that most indi\iduals appear uniformly black in the field. One 
specimen ( MCZ 131990) from Ocracoke \'illage was aberrantly 
red. The areas normally white or gray, ventrally and laterally, 
were rose or rustv. There were, howe\ er, no red-centered cres- 
cents posterior to \entral 44. 

Many specimens show an indication of the pale vertical bar 
that crosses the angle of the jaws as illustrated in Conant ( 1963, 
fig. 6E). This is frequent in Natrix sipedon from southern 
coastal localities, and usually quite different from the pale longi- 
tudinal stripe of Natrix jasciata (Conant, 1963, fig. 6F) . 

Most specimens have red- or brown-centered ventral crescen- 
tic markings on the anterior \enter; the most posterior of these 
\ary in position from ventral six to \entral 48. Ven- young speci- 
mens, such as MCZ 129252-74, from Portsmouth, Portsmouth 
Island, have no red- or brown-centered \entral crescents at all. 

Scale characters are of little use in distinguishing Natrix jas- 
ciata from Natrix sipedon sipedon (Conant, 1963: 11-15), 
and they are of no value in separating williamengelsi from the 
nominate race of sipedon. Overall variation in williamengelsi 
is as follows: ventrals 129-143, mean 138 (females 133 to 143, 
mean 138; males 129 to 143, mean 137); subcaudals 57-84, 
mean 70 (females 57-70, mean 64; males 62 to 84, mean 75) ; 
maximum number of dorsal scale rows in both sexes 21-23 
(among the 50 specimens, only six vary from 23 — three have 
22 and three have 21). 

Size. Our largest specimen (MCZ 129290, a female from 
near Horse Pen Creek in the "Up Trent" region of Ocracoke 
Hammock, collected 16 June 1971, by Numi C. Spitzer) meas- 
ured 1224d= mm (48 in.) in life. This snake is illustrated in 
Figure 1. The smallest specimen examined is 139 mm in total 
length. This snake (a male) is the runt of a litter of 24 the rest 
of which vary from 193 mm to 216 mm (mean 204 mm). These 
are young of a female (MCZ 129251) from Portsmouth, Ports- 
mouth Bank, that measures 1040 mm, total length, and has a 
stump tail 1 25 mm long. 

Comparisons. Melanism occurs sporadically in several parts 
of the range of the nominate subspecies (sipedon). Occasional 
individuals of that widely distributed complex may be as dark 
dorsally as t)pical specimens of williamengelsi, and some large 
adults may be almost plain black or very dark brown. Ventrally, 
however., there is great variation in such dark snakes. In some. 



6 BREVIORA No. 400 

the under surface may be virtually plain black, but in others 
there may be bold black half-moons with or without deep red 
centers, a longitudinal midventral stripe of white, yellow, orange, 
or red may be present, or the entire belly may be dusted with 
gray or black, etc. Among the many thousands of specimens of 
A^. s. sipedon that we (collectively) have examined, we have 
never encountered a melanistic population that combines all four 
diagnostic characteristics of williamengelsi. Melanism in other 
populations is local or restricted to a few individuals, usually 
large adults. 

INTERGRADATION 

Intergradation between williamengelsi and nominate sipedon 
is extensive. The fresh- and brackish-water marshes of Pea Island, 
which is the northern end of Hatteras Island, and on Bodie 
Island, just north of Oregon Inlet, have yielded a considerable 
number of specimens from intergradient populations. Among 13 
from the Bodie Island marshes, five appear to be typical sipedon, 
four are indistinguishable from williamengelsi, and four are in- 
termediate. 

Among 27 specimens from Pea Island, eight show the pattern 
characteristics of sipedon, six those of williamengelsi, and the 
other 13 are intermediate. We have 11 additional specimens 
from the zone of intergradation between sipedon and william- 
engelsi. A specimen from east of Manteo, Roanoke Island, is 
virtually typical A^. s. sipedon; another from the same locality is 
close to williamengelsi. Of two from Waves, Hatteras Island, 
one is intermediate, the other virtually williamengelsi. A speci- 
men taken eight miles north of Avon, Hatteras Island, has all 
the characteristics of williamengelsi. Another, from 4.3 miles 
north of Buxton, Hatteras Island, is typical of A^. s. sipedon. 
From the Buxton area, the southernmost limit of the zone of in- 
tergradation, one specimen appears to be typical of sipedon, two 
are typical of williamengelsi, and two are intermediate. Two 
specimens from Hatteras Island, without precise data, but pre- 
sumably from north of Hatteras Village, are A^. s. sipedon in 
most characters. A specimen from Creed's Hill, Frisco, within 
the Buxton Woods region but well west of Buxton proper, is in- 
termediate. 

We consider the following preserved specimens ( all from Dare 
Co., North Carolina, except as indicated) to be intergrades be- 
tween Matrix sipedon sipedon and A^^. s. williamengelsi: BODIE 



' 1973 THE CAROLINA SALT MARSH SNAKE 7 

ISLAND: AMNH 108839, 108842-50; Duke University col- 
lection (DU) 288. HATTERAS ISLAND: MCZ 129294, 8 
mi. N of Avon; DU 227, 229, and MCZ 129296, Buxton; MCZ 
129293, 4.3 mi. N of Buxton; NCSM 11800, Cape Hatteras; 
DU 228, Creed's Hill; USNM 72571-72, NE of Hatteras VU- 
lage; DU 242, and MCZ 129301, Waves. PEA ISLAND: 
AMNH 108835, 108837-38, 108840-41, 108851-64 and MCZ 
129309-10; DU 231, 1/4 mi. S of Oregon Inlet; DU 226, 230, 
2 mi. S of Oregon Inlet; MCZ 129295, 4 mi. N of Rodanthe. 
ROANOKE ISLAND: Carnegie Museum (CM) 23074, E of 
Manteo; DU 120, 2 mi. SE of Manteo. MAINLAND: Hyde 
Co.: AMNH 89149, Engelhard. 

Another specimen of Natrix sipedon sipedon is now available 
from the peninsula between Pamlico and x\lbemiarle Sounds, an 
area in which members of this taxon apparently are quite un- 
common. This is NCSM 11475 from Slades Creek, near Slades- 
\iUe, Hyde County. The onl\- other specimen of the nominate 
race reported from the peninsula is AMNH 43427 from Lake 
Phelps (Conant, 1963: 26). 

HYBRIDIZATION 

Hybridization with Natrix fasciata is also a major feature of 
A^. s. willia7nengelsi biology. An apparently typical but excep- 
tionally dark female Natrix fasciata. from Ponzer, on the Hyde 
County mainland (AMNH 88077), produced a Litter (AMNH 
88078-101 ) that exhibits a variety of characteristics ranging from 
those of the mother f fasciata) to those of typical N. s. williain- 
engelsi. The pale marking passing through the angle of the jaws 
matches the \-ertical bar of Natrix sipedon in some, whereas in 
others it is comparable with the longitudinal stripe of Natrix 
fasciata. The dorsal markings \'ary from the posteriorly stag- 
gered middorsal and lateral bars of sipedon to the bands that 
are continuous throughout the length of the body in fasciata. 
All members of this litter are very dark. (See Conant, 1963 : 30, 
for further details.) The mother, as noted, is a ver\' dark A^. 
fasciata; we do not know the relevance of visual cues in mate 
selection in these forms of Natrix; perhaps they are important. 

A hybrid swarm occurs at Mullet Pond, on the Shackleford 
Banks, in a freshwater ecosystem of largely mainland ecological 
affinities. Among 22 specimens from Mullet Pond, four (in- 
cluding MCZ 46688, the type of "engelsi") are morphologically 
typical of Natrix fasciata, three are more or less typical of wil- 



8 BREVIORA No. 400 

liamengelsi, and the others, Including young from three litters and 
totalling 15 specimens, are intennediate. Snakes from Mullet 
Pond, Shackleford Banks, are: AMNH 75800-04 + 13 un- 
tagged; DU 10, 234; California State University, Long Beach 
(CSULB) 263-64. 

Apart from the Ponzer litter and Mullet Pond specimens, we 
have examined one other snake from the North CaroHna main- 
land which is apparently a hybrid. It is NCSM 11799 from 
Lennoxville Point, 3 miles east of Beaufort, Carteret County. 

We are of the opinion that much of the apparent introgres- 
sion manifest along the shores of Pamlico Sound and associated 
bodies of water is a direct consequence of the massive disturb- 
ance of habitats by the great cyclonic storms that sweep ashore 
rather frequently in eastern North Carolina. To visit the marshes 
after a hurricane, as Lazell did, following Ginger in early Octo- 
ber, 1971, is to see convincing evidence that salt water can be 
driven far inland by winds and tides. Conversely, as the result 
of torrential rains, vast quantities of fresh water move seaward 
after each tempest has passed, carrying many animals and plants 
of the freshwater biota with them on debris borne by the flood. 
Some of these organisms certainly must make landfalls in the 
salt marshes. Mayr, Linsley, and Usinger pointed out long ago 
(1953: 102) that hybridization between related species may oc- 
cur in localities where the natural ecological balance has been 
badly disturbed by human interference. Conant (1963: 16- 
23) marshalled evidence indicating introgression between sipe- 
don and fasciata along the Fall Line in the Carolinas where 
mankind has been active. Surely hurricanes are equally disturb- 
ing to the ecological balance, at least in some localities. 

The distribution of specimens examined is shown in Figure 2. 
We have made no attempt to show localities for Natrix fasciata. 
which probably occurs in \irtually every body of fresh water on 
the adjacent mainland. Some specimens on which we report pat- 
terns and scale counts are not included on our lists. A few snakes 
were discarded because of their poor condition, and one escaped 
near where it was collected. 

ECOLOGY 

Natrix sipedon ivilliamengehi is closely associated with marsh 
grasses of the genus Spartina and the rush, Juncus, wherexer 
these form salt and brackish marshes. On Ocracoke Island, wil- 
liajnengelsi is common in salt marshes adjacent to Pamlico Sound, 



1973 THE CAROLINA SALT MARSH SNAKE 9 

along the tidal creeks that carry brackish water among the dunes, 
and also in the fresh water of land-locked ecosystems such as 
that which occurs at Island Creek. The density of the water 
where some of the snakes were collected varied from that of the 
Sound proper, 1.018, to Island Creek at .999. The density at 
five collecting stations sampled in May and June, 1971, ranged 
from 1.007 to 1.015; the average density was 1.013. 

The relation of density to salinity is dependent on tempera- 
ture. The water samples taken at the times the snakes were col- 
lected varied from 25° to 29°C; the average was 27°C. Salin- 
ities of seven habitats sampled (including the Sound, Island 
Creek, and the above five) \aried, therefore, from 2.7 to 24 
grams per kilogram ; the average was 20.5 g/kg. 

In the laboratory, specimens of A^. s. williamengelsi fed vora- 
ciously on the minnow, Fundulus heteroclitus, the frogs, Hyla 
cinerea, Hyla squirella, and Rana pipiens, and the toad, Bufo 
fouieri. All of these are abundant on Ocracoke. We have the 
impression that these snakes will eat almost any fish or amphibian 
small enough to swallow, as is also the case with the nominate 
race. E\en after a full week of water star\'ation, our captives 
would not drink salt water, but they eagerly drank fresh water. 
Pettus ( 1 963 ) obtained similar results with Gulf Coast salt marsh 
snakes, Matrix fasciata darki. 

N . s. williamengelsi is a fa\'orite prey of marsh hawks, Circus 
cyanens, and is no doubt taken by other raptors. The numer- 
ous species of egrets and herons that frequent the Carolina salt 
marshes in all probability eat these snakes regularly, but we ha\e 
not observed this. Many specimens have mangled or abbre\iated 
tails that suggest predation by birds or the crabs that are com- 
mon in the habitat. 

The junior author, on field trips o\'er several years, has found 
A^ s. williarnenoehi abundant from May to September in the op- 
timal habitats south of Cape Point, Hatteras Island, on the Outer 
Banks. In the Buxton Woods area, north and west of Cape 
Point, Agkistrodon piscivorus is common, but it is not known 
from the Banks to the southward, at least to Cape Lookout. 
Natrix sipedon is scarce in the Buxton Woods area, where it may 
fall prey to the cottonmouth. 

The only other water snake known from the Outer Banks is 
Natrix taxispilota: it is confined to fresh water in the Buxton 
Woods area, and is scarce. 

It should be pointed out that a large part of the range of 
N. s. Willi ayn en gels i is within National Park borders. The blanket 



10 BREMORA No. 400 

protection provided for all animals includes this snake. Collect- 
ing is illegal within the Cape Hatteras National Seashore, ex- 
cept by specifically designated Park Service employees. 

ACKNOWLEDGMENTS 

We are indebted to staff members of the Cape Hatteras Na- 
tional Seashore Recreational Area for their cooperation with the 
junior author during his field work, especially to Clay Gifford, 
Park Naturalist; David Fletcher, Ranger; and John A. Musick, 
Collaborator. Specimens were lent from several study collec- 
tions by, respectively, Charles M. Bogert and Richard G. Zweifel 
of the American Museum of Natural History, Neil D. Richmond 
of the Carnegie Museum, Joseph R. Bailey of Duke University, 
Richard B. Loomis of California State University, Long Beach, 
Ernest E. Williams of the Museum of Comparative Zoology, and 
William M. Palmer of the North Carohna State Museum. Wil- 
liam L. Engels and John B. Funderburg loaned us material from 
their private collections and permitted us to deposit it in the 
North Carolina State Museum and the American Museum of 
Natural History, respectively. 

The senior author is grateful to Dr. Bailey for making it pos- 
sible to visit the Shackleford Banks. Harold A. Dundee fur- 
nished useful information, and Hobart M. Smith has read the 
manuscript and made a number of helpful suggestions. 

The junior author was actively aided in the field by James 
Woessner, John R. Alexander, Paul Elias, Edward James, Peter 
Rabinowitz, Numi C. Spitzer, and Dr. Musick. Meristics, meas- 
urements, and other data were assembled by Norma Rothman, 
Constance A. Rinaldo, and Miss Spitzer. Mrs. Rothman ob- 
tained many specimens for the senior author's studies on the 
genus Natrix in conjunction with her own field work in North 
Carolina. 



1973 THE CAROLINA SALT MARSH SNAKE 11 

LITERATURE CITED 

Barbour, T. 1943. A new water snake from North Carolina. Proc. 
New England Zool. Club, 22: 1-2, pi. 1. 

CONANT, R. 1961. A new water snake from Mexico, with notes on 
anal plates and apical pits in Natrix and ThamnopJns. Ameri- 
can Mus. Novitates, No. 2060: 1-22. 

1963. Evidence for the specific status of Natrix fas- 

ciata. American Mus. Novitates, No. 2122: 1-38. 

Engels, W. L. 1952. Vertebrate fauna of North Carolina coastal 
islands: Shacklefoi-d Banks. American Midland Nat., 47: 702- 
742. 

Ma^-r, E., E. G. Linsley, and R. L. Usinger. 1953. Methods and 
Principles of Systematic Zoology. New York, N.Y.: McGraw- 
Hill Book Co. 346 pp. 

Pettus, D. 1963. Salinity and subspeciation in Natrix sipedon. 
Capeia, 1963: 499-504. 



12 



BREVIORA 



No. 400 





N 


s 


r- 


4J 




o 

'So 


CD 


^H 




<D 


^ 






^ 




c 




^ 




"^ 










K 


-4-J 


a 


J^ 


r-^^ 


<—" 


-i^ 


J3 






o 


3 


g 


a 


r- 


St 


•i 


P 


"qj 










•"s 




o 


rs 


^ 


c; 




X 


<?^ 


r-" 


rt 


1— ( 




-l-» 


u 


>» 


?.* 
-§ 




(—• 




^ 


r^ 


4J 


o 


,^*^ 




o 


o 


V. 


?H 


:z; 


X 




o 




Ah 


•S 




•^ 






CM 


d 


QJ 


:?: 


OJ 


QJ 


> 




cc 


■r! 


, , 




r~i 


>> 


0$ 


QJ 

a; 


o 


w 


rH 


r-* 


r-I 
QJ 




QJ 
> 


Tfi 


?; 


hH 


CD 




1— < 

p 


QJ 


% 


g 




o 


yA 




o; 


o 


, 


-u 




rt 


> 




c^ 


?H 


r^ 


rt 
M 


QJ 


o 
O 


QJ 




«*-! 




1 


c3 




^ 


rt 


_S 


9 




m 


^^ 


o 


O 


p 




?: 


C 




0) 

CiC 




u 






M 


QJ 


tH 






^ 


a; 


< 


OJ 


P 


fn 




o 




3 


^ 


o 


r- 


fc/j 


^ — ^ 


rt 




E 


o 

C3 


O 


i 




(M 


-t 




CI 




(M 




C<1 


«(H 


Cvl 




1973 



THE CAROLINA SALT ^L^RSH SNAKE 



13 



LEGEND 

• N. s_ williomen q eisi 



o 


N. s sipedon and 




intergrades with 




williamenqelsi 


X 


N. s. williamenqelsi 




hybrids with 




N. f. fasciata 



Roanoke\^ \ 
" and 15 





Ocracoke 
Portsmouth 



Shacklefo 
Bank 



Cape Lookout 



Bodie 
Island 




Pea 
Island 



Waves 



Avon 

'Buxton 
Hatteras 



N 



Figoire 2. Map showing distribution of Natrix sipedon wilUa'nv- 
engeh'i and intergrades between it and A'. ,s. sipedon and hybrids be- 
tween wUliamengelsi and A^. fasciata fasciata. 



^- -CMP. 200, 

MAR 1 Q fn.— 

B R E V I O R A 

Rloseiim of Coniparatiye Zoology 

CIambridge, Mass. March 6, 1973 Number 401 



THE CHANARES (ARGENTINA) TRIASSIC 

REPTILE FAUNA. XVIII. PROBELESODOX MLXOR, 

A NEW SPECIES OF CARNIVOROUS CYNODONT; 

FAMILY PROBAINOGNATHIDAE NOW 

Alfred Sherwood Romer 

Abstract. A small species of Probelesodon, similar to P. lewisi 
but of smaller size, is described as Probelesodon. miyior sp. nov. 

Because of its advanced nature, particularly in the development 
of a squamosal socket for lower jaw articulation, it seems advisable 
to remove Pi-obainognatJius from the Chiniquodontidae and to erect 
for it the monotj^iic new family Probaino.crnathidae. 

Probelesodon minor sp. nov. 

Holotype: La Plata Museum 64-XI-14-18 (field no. 138 
pt. ). A skull and jaws (Figs. 1, 2). From the Chaiiares Forma- 
tion, about 4 km north of the mouth of the Rio Chaiiares, La 
Rioja Pro\ince, Argentina. 

Diag7iosis. A small species of Probelesodon, al;)out half the 
size of P. lewisi, known skulls having a basal length of about 
70 mm. 

In earlier papers on the carnivorous cynodonts from the 
Chaiiares (Romer, 1969, 1970), two forms were clearly dis- 
tinguishable, a large form described as Probelesodon lewisi, and 
a small, more advanced form described as Probainognalhus 
jenseni. At the time of publication, preparation of this carni\o- 
rous series had not been completed and it was assumed that 
all the small skulls pertained to Probainognalhus. \Vith further 
preparation by Head Preparator Arnold D. Lewis, it became 
obvious that this is not the case. Se\"eral small skulls and jaws 
are clearly of the Probelesodon type. All are apparently mature, 
and do not represent growth stages; all are about half the size 
of the "typical'' specimens of Probelesodon leivisi. and show a 
size difference too great to be sexual dimorphism. It is obvious 



2 



BREVIORA 



No. 401 




Figure 1. Probelesodon minor, holotype skull in dorsal view, X 3/2. 



that we have a second, small, species of Probelesodon which is 
herewith described. 

In addition to the holotype the new species is represented by: 
MCZ 4100, a skull and jaws; MCZ 4099, partial skull and jaws; 
MCZ 3777, jaws and some postcranial scraps; MCZ 4102, in- 
complete jaws; MCZ 4163, postcranial materials. 

The holotype skull is incomplete in the premaxillary area, but 
total length was approximately 78 mm from snout to the level 
of the posterior end of the squamosal arch, 68 mm to the 
condyle. The lower jaw length is 64 mm. In MCZ 4100 the 
comparable figures are 73, 70, and probably 62 mm (the jaw 
is incomplete posteriorly). The materials of MCZ 4099, MCZ 
3777, and MCZ 4102 are comparable in size. The postcranial 
material of MCZ 3777 (to be described in the next paper in 
this series) is comparable in nature to equivalent elements in 
P. lewis i except for smaller size, and the postcranial elements 
of MCZ 4163 are similar. 

In all major features the skull and jaws are comparable to 
those in P. lewisi; skull proportions are closely comparable and 
the sutural pattern is similar, except that in P. ?ninor the lacrimal 
and prefrontal are somewhat more expanded dorsally at the 



1973 



CHANARES GO^rP^ODONT.S 




Figure 2. Prohelesodon minor, holotype skull in lateral view, X 3/2. 

expense of the nasal. The secondary palate, as in P. lewisi, 
extends remarkably far back, exceeding in de\elopment e\en 
the contemporary Probainognathiis. As in Probainognathus 
jenseni the postero\entral flanges of the pterygoid are highly 
developed; also as in that species the posterior end of the lower 
jaw lies very close to the inner surface of the squamosal, and 
there was presumably a ligamentous connection, although no 
development of a '"glenoid" articular surface. 

As in P. lewisi there were four premaxillary "incisors," and 
there are, in the type and MCZ 4100, eight postcanine cheek 
teeth. As in P. lewisi these teeth are somewhat "hooked" — 
cur\ed backwards at their tips. Three "incisors" are present in 
the lower jaw, and seven "cheek" teeth behind the prominent 
canine. 

I ha\-e earlier suggested that Prohelesodon lewisi was ancestral 
to the larger Belesodon of the Brazilian Santa Maria beds. 
Closely related to the latter w'as the smaller, contemporary, 
Chiniquodon. Not improbably P. minor may ha\"e been an- 
cestral to Chiniquodon, a form of similar structure but of 
considerably larger size. 



Probaixogxathidae Fam. Nov. 

The carnivorous cynodonts from the Chanares Formation, 
Probainognathus and Prohelesodon. are both ob\iously ad\'anced 
forms, with an elongate secondary palate and other progressi\-e 
features. Similarly advanced are the Santa Maria forms, Chini- 
quodon and Belesodon, for which Huene (1944) erected the 
family Chiniquodontidae. In the past, I ha\e included both 



4 BREVIORA No. 401 

Chafiares genera in that family. TJiat this assignment is correct 
for Probelesodon seems certain, for that genus is obviously closely 
related to the Santa Maria forms. The position of Probain- 
ognathus, however, calls for further consideration. Like the 
proper chiniquodonts, Probainognathus is quite surely de- 
scended from the primiti\e galesaurid cynodonts of the early 
Triassic. But the progress has been in a somewhat different 
direction. The chiniquodonts proper are even more advanced 
in palatal construction than Probainognathus, but are apparently 
somewhat aberrant in dentition and in certain postcranial fea- 
tures to be described in a future number of this series of 
publications. Probainognathus, on the contrary, appears to be 
takinc; a direct course toward a mammahan condition and 
shows a very marked progressive situation in the de\elopment 
of a "glenoid" socket in the squamosal for lower jaw articulation. 
On the whole, it is perhaps best to separate Probaiiiognathus 
as the type of a new family Probainognathidae, characterized 
by progressive cynodont features, most important of which is the 
articulation of the lower jaw with the squamosal in mammalian 
fashion. 

Collection and preparation of Probelesodon and Probaino- 
gnathus were made possible by grants from the National Science 
Foundation. 

REFERENCES CITED 

HUENE, F.v. 1944. Die fossilen Reptilien des siidamerikanischen 
Gondwanalandes. Munich: C. H. Beck'sche Verlag-s. 332 pp. 

ROMER, A. S. 1969. The Chafiares (Argentina) Triassic reptile 
fauna V. A new chiniquodontid cynodont, Probelesodon letvisi — 
cynodont ancestry. Breviora, No. 333 : 1-24. 

. 1970. The Chaiiares (Argentina) Triassic reptile 

fauna VI. A chiniquodont cynodont with an incipient squamosal- 
dentary jaw articulation. Breviora, No. 344: 1-18. 



MAR 1 8 1985 

B R E V I O R. A 

Museum of Comparative Zoology 

Cambridge, Mass. June 1973 Number 402 

SPEGIATION IN THE GENUS OCHTHOECA 
(AVES: TYRANNIDAE)^ 

John W. Fitzpatrick 

Abstract. The nine Chat-Tyrants are small tyrannid flycatchers 
occurring in the Andes from Colombia and western Venezuela south 
to Bolivia and extreme northern Chile. Most of the species are di- 
vided into numerous isolated races. There are several distinct species- 
groups, and the species are highly synipatric. The complicated ranges 
appear to have arisen during the final three Pleistocene glacial 
periods in the northern Andes. During glacial peaks, the then-ex- 
isting species dispersed over wide ranges. They were divided during 
the subsequent interglacials. The three periods of dispersal and sub- 
sequent splitting first isolated the species-group precui-sors, then the 
species themselves, and finally the races of each species. The many 
isolated races of each species today are separated by geographic and 
vegetational barriers which probably affect a number of other mon- 
tane species as well. 

Introduction 

Several recent studies (B. Vuilleumier, 1971; F. Vuilleumier, 
1969, 1970) have revealed consistencies in the ranges and specia- 
tion patterns of the fauna of the South American Andes. The 
patterns indicate that a large number of Andean species orig- 
inated and diverged in conjunction with the periodic glaciations 
of the Quaternary. Alternating glacial advances and retreats 
resulted in the complex sympatry between related species oc- 
curring there today. 

The avian genus Ochthoeca reflects this type of complex sym- 
patr)'. Eight of its nine species have widely separated races, and 
although several species occur throughout the central and north- 
em Andes, others are confined to smaller areas. Sympatry among 
species in the genus is considerable. 

Tublished by a grant from the Wetmore Colles Fund. 



BREVIORA 



No. 402 




O. CINNAMOMEIVENTRIS 




O. RUFIPECTORAUS 




O. FRONTALIS 




O. FUMCOLOR 



Figure 1. Four members of the genus Ochthoeca. Species-groups 
diadema and oenayithoides are represented by O. frontalis and 0. 
fumicolor, respectively. About two-thirds life size. 



1973 SPECIATION IN OCHTHOECA 6 

Known collectively as the Chat-Tyrants, all species are tree- 
or shrub-inhabiting flycatchers of the middle- and high-altitude 
Andes. They occur in the upper subtropical, temperate, and 
paramo zones from the mountains of western Venezuela and 
Colombia south to western Bolivia and extreme northern Chile. 
Members of the genus have been collected at altitudes from 500 
to 4314 meters, and the a\erage for the group is slightly over 
2700 meters. They are small, generally chunky and large-headed 
flycatchers, and ail have a conspicuous supraloral stripe, or "eye- 
brow," varying in length and color (see Fig. 1). Their chief 
mode of prey-catching is gleaning insects from the vegetation, 
though several species also sit on conspicuous perches and hawk 
insects from the air (F. \'uilleumier, 1971). All species are 
nonmigratory and, with one apparent exception, they are com- 
mon within their respective ranges. 

The ranges of each species, plotted from their collection lo- 
calities, are shown in Figure 2. Based on these present ranges 
and on the Andean glacial exents during the Quaternary, I shall 
propose here a sequence of speciation for the group. 

The Genus 

The crenus Ochthoeca, as recognized b\- de Schauensee 
(1966), contains the nine species listed below. Two species are 
quite distinct from the others and ha\e widespread ranges. The 
remaining seven ha\e been placed in two species-groups by F. 
Wiilleumier ( 1971 ) . 

O. cinnamotyieiventris 

O. rufipcctoralis 
diadema species-group 

O. diadema 

O. frontalis 

O. pulchella 
oenanthoides species-group 

O. oenanthoides 

O. fumicolor 

O. leucophrys superspecies 

1 . leucophrys 

2. piurae 

F. Vuilleumier considered O. pulchella and O. frontalis as 
members of a superspecies within the diadema species-group. 
Although they are very similar, differing chiefly in the amount 
of yellow in the eyebrow, both species have been collected at 



BREVIORA 



No. 402 




cirmamomeiventris 





pulchella 




furaicolor 



oenanthoides 111' 



leucophrys 



piurae 



Figure 2. Ranges of species in Ochthoeca. The diadema and 
oenanthoides species-groups are shown on the lower left and lower 
right, respectively. 



1973 SPECIATION IN OCHTHOECA 5 

each of four localities, and their ranges show considerable over- 
lap south of the Maranon River. I shall, therefore, treat them 
as sibling species within the group. 

One species, cinnamomeiventris, occurs in the moist forests 
of the upper subtropical and temperate zones. Its dark colora- 
tion is adapted to these dense, wooded habitats. Its bill is the 
broadest and shortest of the genus, reflecting its diet of aerial 
insects captured by hawking. O. rufipectoralis and the diadema 
species-group inhabit open forests of the temperate zones. Their 
bills are also short, but all are much thinner than that of cin- 
namomeiventris. O. diadema is the only species whose plumage 
is largely green. The oenanthoides species-group contains forms 
frequendy found over 4000 meters. O. fumicolor lives in moist 
paramo habitats, while oenanthoides and the leucophrys super- 
species occur in the dry puna scrub and grasslands (F. Vuil- 
leumier, 1971). Members of this group have long, thin bills 
adapted to their diet of insects gleaned from the vegetation. All 
ha\'e significantly larger bodies than the other Ochthoeca species, 
with relatively longer wings and tails. O. leucophrys is predomi- 
nantly grey, while the other two are buffy brown. 

Pleistocene Climatic Influences 

The zoogeographic history of these species seems closely re- 
lated to the periodic altitudinal lowering of their habitats dur- 
ing the Quaternary glaciations. Glacial climates affected the 
forest- and grass-life zones in two ways. First, the limits of each 
were lowered, creating connections or closer proximity between 
previously isolated habitats. During maximum glaciation, the 
life zones were lowered sufficiently to become nearly continuous 
along the entire mountain chains (B. Vuilleumier, 1971). Sec- 
ond, the valleys and lower mountain areas became more humid 
as the ice and glacial lakes surrounded them. This again resulted 
in greater proximity between the formerly higher and isolated 
humid life zones. During the glacial maxima, the lowering of 
the habitats permitted many avian species to colonize through 
the northern mountain chains along a north-south axis. During 
interglacial periods the life zones retreated to higher altitudes, 
thereby splitting again into isolated "islands." The deep valleys 
became dry, further decreasing the potential for gene flow be- 
tween neighboring high-altitude populations. Thus a species 
whose range had been widespread and nearly continuous during 
maximum glaciation divided during the subsequent interglacial. 
New races and species evolved as a result. 



6 BREVIORA No. 402 

Within the central and northern Andes, two particular cli- 
matic features affected speciation in high-altitude populations. 
First, the western slope from southern Ecuador through Peru 
and northern Chile now receives very little rainfall. Presumably 
this was the same in the past, and the area remained generally 
unavailable to the forest-inhabiting Ochthoeca species. Only 
one species, leucophrys, occurs in this arid region today. Sec- 
ond, two large areas of low, dr\' vegetation occur within the 
range of this genus. The first is the low, semi-arid valley of the 
upper Magdalena River, which lies between the Eastern and 
Central Cordilleras of Colombia, creating a significant barrier 
between a number of montane taxa on either side (F. Vuil- 
leumier, 1969). Five Ochthoeca species now occur in northern 
Colombia and northwest Venezuela. Only that species inhabiting 
the lowest altitudes does not show a major break and/or sub- 
specific differentiation in this area. A second, equally efTective 
barrier occurs in the region of the upper Maraiion River in 
northern Peru. The low mountains in this area create a wide 
break between the extensive high-altitude regions of Peru and 
Ecuador. The ranges of all nine Ochthoeca species reach this 
divide, and six show major breaks at the barrier. Furthermore, 
evidence exists that warmer Pleistocene interglacial tempera- 
tures resulted in even drier conditions still less favorable to alpine 
life than today's climate (B. Vuilleumier, 1971). Hence these 
dry barriers isolated populations more effectixely in the past 
than they do today. These two major divides thus created three 
large, separate areas of montane conditions north of Bolivia 
which appear to have been the primary centers for speciation 
in Ochthoeca, and certainly for other species as well. 

Speciation in Ochthoeca 

Geologists recognize four major world-wide glacial advances 
during the Pleistocene (e.g., Leet and Judson, 1971), but the 
first had relatively less effect on the northern Andes, since they 
were the last to attain their present height (B. Vuilleumier, 
1971). Data on high-altitude species in the northern Andes in- 
dicate that they were derived predominantly from central An- 
dean birds (F. Vuilleumier, 1970). Species originating in these 
central mountains dispersed northward during the three major 
glacial periods affecting the northern mountains. The precursor 
of the genus Ochthoeca seems therefore to have originated in 
the central Andes, and became isolated during the first Pleisto- 



1973 



SPECIATION IN OCHTHOECA 



to 



a 
<u 

a 

o 

a 

a 

c 
c 

o 

•X 




FigTire 3. Schematic evolution of the genus through time, from 
it^ ancestral stock. Shaded areas show dispersal periods during the 
final three glaciations; unshaded areas represent interglacial periods, 
when isolated populations differentiated into new forms. 



cene interglacial period. As are most species today, this was 
probably an open-forest bird. The occurrence of seven of to- 
day's nine species in northwestern Bolivia and southern Peru 
further indicates that the area was probably suitable to the pre- 
cursor. 

The development of the modern genus from this precursor 
is suggested by the structure of the group as a whole. The ex- 
istence of clear-cut species-groups, the subdivision of the groups 
into component species, and the species' divisions into morpholog- 
ically distinct races indicate three separate occasions of dispersal 
each followed by isolation. In the first period, the parent species 
was split into species-group precursors, which in turn split into 
their component species following the second dispersal period. 



8 BREVIORA No. 402 

Finally, after the third period, each species was broken into the 
isolated populations found today. These three dispersal periods 
correspond to the three glaciations following the isolation of the 
progenitor early in the Pleistocene. This pattern is outlined in 
Figure 3. A detailed sequence for this speciation will now be 
presented. 

During the first of the three glaciations, the parent species 
was probably able to colonize northward over the entire Andean 
chain. Colombia's Eastern Cordillera had by this time gained 
its present altitude and could support alpine life, particularly 
during the cold, damp glacial period. At this time there was 
sufficient gene flow throughout the range to inhibit differentia- 
tion and the population remained a single species. 

Upon glacial retreat, the two major geographic barriers, the 
Magdalena and the Marafion valleys, became effective. As il- 
lustrated in Figure 4a, the original species was split into three 
populations that formed the parent stocks of the two species- 
groups and the precursor to the remaining two species. The 
southernmost population, in the highest mountain areas, gave 
rise to the highest-altitude species-group. This precursor shall 
be referred to as "oenanthoides."^ This conclusion is further 
strengthened by the generally southern ranges of these species. 
While this form was confined to Peru and Bolivia, forest birds 

^To avoid confusion between these primitive taxa and the modern 
species to which they gave rise, the precursors will be called by their 
species-group name, within quotation marks. Hence "oenanthoides" 
eventually gave rise to the species now within the oenanthoides 
species-group. 

Figiire 4. (A) Three populations isolated after the Mindel (sec- 
ond) Pleistocene glaciation; "diadema" (north) and "oenanthoides" 
(south) gave rise to the respective species groups; "rufipectoralis" 
(central) differentiated as shown in 4(D). 

(B) Status of "oenanthoides" during and following 
tlie Riss (third) glaciation; leucophi-ys (west) and oenanthoides 
(east) were separated by glacial ice and lakes. O. fumicolor differen- 
tiated in the north. 

(C) Isolation of diadema (north), frontalis (central), 
and pulchella (south) from "diadema" following the Riss glaciation. 

(D) Isolation of cimiamomeiventris east of the Mag- 
dalena River following the Riss glaciation. O. rufipectoralis (cen- 
tral) may have ranged south of the Marafion River by this time. 

Various races have now differentiated following the dispersal of 
the species during the Wiirm (final) glaciation. 



1973 



SPECIATIOX IN OCHTHOECA 







10 BREvioRA No. 402 

in western Ecuador and Colombia differentiated into "rufipec- 
toralis." The population that was isolated in the Eastern Cordil- 
lera of Colombia became "diadema." These latter two forms 
probably remained relatively undifferentiated until the succeed- 
ing glaciation brought them into secondary contact, since even 
today they apparently share similar habitats. 

The second glacial episode again resulted in range extensions 
along the lower valleys. An important occurrence during this 
period is illustrated in Figure 4b: "oenanthoides," which was 
evolving into a higher-altitude, paramo-inhabiting species, ex- 
tended through the central and northern mountains. Its range 
seems to have been divided during the glacial peak by the ex- 
tensive ice and glacial lakes that capped the highlands of Peru 
and Bolivia in complex patterns (B. Vuilleumier, 1971). Thus 
at least three populations were isolated during this glaciation, 
as a result of their high-altitude requirements. Restricted by the 
ice to a portion of the western slope, leucophrys developed into 
a puna-inhabiting species. O. oenanthoides differentiated to the 
east, and the two apparently exclude each other over much of 
their ranges, for sympatry today remains fairly small; they still 
reflect somewhat these primitive ranges. O. piurae also prob- 
ably speciated at this time on an isolated puna region in extreme 
northwestern Peru, never drifting far enough from leucophrys 
to permit extensive sympatry after the glacial retreati. 

Figure 4c illustrates the probable spread and differentiation 
of "diadema" into the species of that group. Again the two 
barriers created three surviving, isolated populations, which 
evolved into diadema in the north, frontalis in the central moun- 
tains, and pulchella south of the Marafion. Characters shared 
by these species are nearly identical wing, tail, and bill measure- 
ments, wingbars, and bright yellow in the eyebrow, as well as 
smoky-grey crowns and backs; their close relationship is quite 
clear. Their current ranges reflect the three primitive ranges, 
with one exception: the range of frontalis now extends south- 
ward to overlap that of pulchella. To judge from collection 
data, however, frontalis appears less common within this south- 
ern range extension. The frequency of pulchella south of the 
Maraiion and its total absence farther north indicate that fron- 
talis did achieve species rank in the north and move south into 
pulchella's range. 

The second glaciation allowed the expansion of "rufipec- 
toralis" along the forests of the northern Andes, possibly through- 



1973 SPECIATION IN OCHTHOECA 11 

out the range of the genus. When the ice retreated, two popu- 
lations were separated. In the forests east of the Magdalena 
River, cinnamomeiventris differentiated, and in the central 
Andes, rufipect oralis reached species rank. This pattern is sug- 
gested by the large gap in the current range of the latter species 
across the Magdalena. This break is not present in the range 
of cinnamomeiventris, which adapted to the moist forests at a 
lower altitude in the Eastern Cordillera. Furthermore, the rela- 
ti\ e frequencies of the two species follow this pattern — cin- 
namomeiventris is common in the north and occurs only at 
scattered localities south of Colombia. O. rufipectoralis is much 
more common in the south though it has established small pop- 
ulations in the north, where cinnamomeiventris originated. Their 
close relationship is supported by their sharing of the white eye- 
brow and the conspicuous rufous-chestnut breast band. O. cin- 
namomeiventris appears to be a darkened form of rufipectoralis 
adapted to forest habits (see Figure 1). Their morphological 
and ecological difTerences certainly increased upon secondary 
contact, and both species eventually dispersed throughout the 
range of the genus. Their phylogenetic relationship appears to 
be analogous to that of the two species-groups. Their status as 
taxonomically distinct species arose from their more significant 
divergence following the second glaciation. 

Thus all current species were present at the onset of the final 
glaciation. Their ranges had already overlapped in several re- 
gions, and they probably had become well differentiated. Dur- 
ing this glaciation the species colonized new areas as far as 
suitable habitat and competition with congeners permitted. The 
northeastern species, diadema and cinnamomeiventris, moved 
south, diadema stopping where pulchella was established. Many 
populations seem to have crossed the Maraiion to invade the 
moist eastern slopes of Peru, resulting in the seven species oc- 
curring there today. O. leucophrys, a strictly high-altitude spe- 
cies, crossed to the east and spread south into Argentina. Upon 
glacial retreat this resulted in a series of isolated populations 
stretching southward along the mountains at progressi\'ely lower 
altitudes that coincided with the preferred habitat. In short, 
each species spread to its ecological limits and was subsequently 
split by the final warming and drying of the valleys. Hence the 
morphological isolates recognized today gi\e evidence for the 
geographic barriers that split these last major glacial range ex- 
pansions. 



12 BREVIORA No. 402 

Current Speciation 

Isolated races of each species now exist in regions which, fol- 
lowing the last glaciation, became sufficiently separated to limit 
gene flow between them. Several interesting exceptions are 
worthy of note. 

The relative scarcity of O. frontalis (I found only twenty 
localities recorded in the literature) makes meaningful discus- 
sion of its subspecific ranges difficult. The apparent large break 
in its distribution in northern Peru seems insufficient to isolate 
the populations on either side, as both are apparently the same 
race (Carriker, 1933; Zimmer, 1937). It appears, therefore, 
that the gap is a product of the rarity of the species, rather than 
a genuine absence throughout the region. 

The subspecific continuity of O. fiimicolor across this same 
low divide is also confusing, for the species normally occurs at 
altitudes well over 3000 meters. In contrast to O. frontalis, O. 
fumicolor is extremely common in the north-central Andes; ap- 
parently the few small isolated paramos within the divide permit 
the maintenance of gene flow across it. Collection localities from 
north to south through the divide are: Bestion (3100 m), Tara- 
guarcocha (3200 m). El Tambo (2900 m), Cutervo (3000 m). 
Hence certain sufficiently common species apparently have the 
potential for gene flow across this divide. 

Among other species, in areas lacking any major geographic 
barriers, several subspecific discontinuities are found which 
would seem to be the result of habitat breaks not marked on 
vegetation maps. In one such area, O. frontalis, O. cinnamo- 
meiventris, and O. rufipectoralis each split into different races 
on either side of the extreme upper Maraiion, while higher alti- 
tude species do not. Excessive taxonomic splitting of clinical dif- 
ferentiation may cause these anomalies. 

In general, the modern ranges of the species, with their many 
isolated races, provide accurate evidence of currently isolated 
vegetation zones. Such regions, as indicated also by B. Vuil- 
leumier (1971), are presumably equally important in many 
other upper montane species. These regions, some of which may 
be areas of current or future speciation, are separated by the 
geographic barriers listed below. The Ochthoeca species split 
apart by the barriers are listed under each. 

1 . Huallaga Valley, southern Peru 

fumicolor, leucophrys, pulchella 



1973 SPECIATION IN OCHTHOECA 13 

2. Maranon Valley and low Peruvian mountains 

cinnamomeiventris, diadema 

3. Upper Magdalena Valley, northern Colombia 

fumicolor, diadema, frontalis, rufipectoralis 

4. Cristobal Valley, separating mountains of Merida, 
Venezuela, from the Eastern Cordillera 

fumicolor, cinnamomeiventris, diadema 

5. Yaracuy River valley, isolating the northern coastal 
mountain ranges of Venezuela 

diadema 

6. Cesar Depression, isolating Nevada de Santa Marta 

diadema, rufipectoralis 

7. Upper Rio Catumbo depression, separating the Perija 
Ridge from the Eastern Cordillera 

diadema, rufipectoralis 

ACKNOW^LEDGEMENTS 

I would like to express my deepest thanks to Raymond A. 
Paynter, Jr. for his very generous time and painstaking edito- 
rial advice during the preparation of this paper. 

Literature Cited 

Carriker, M. a. 1933. Notes on Peruvian birds. Pioc. Acad. Nat. 
Sci. Philadelphia, 85: 23-24. 

De Schauensee, R. M. 1966. The Species of Birds of South 
America and their Distribution. Narbeth, Pennsylvania: Liv- 
ingston Publ. Co. xvii + 577 pp. 

Leet, D. L. and Judson, S. 1971. Physical Geology (fourth ed.). 
New Jersey: Prentice-Hall. 

VuiLLEUMiER, B, 1971. Pleistocene changes in the flora and fauna 
of South America. Science, 173: 771-780. 

VuiLLEUMiER, F. 1969. Pleistocene speciation in birds living in the 
high Andes. Nature, 223: 1179-1180. 

1970. Insular biogeography in continental re- 
gions. Pt. I. Am. Nat., 104: 373-387. 

1971. In Smith and Vuilleumier, Evolutionary 



relationships in some South American ground tyrants. Chap. I. 
Bull. Mus. Comp. Zool., 141: 181-232. 
ZlMMER, J. T. 1937. Notes on the genera Agriomis, Muscisaxicola, 
Myiotheretes, Ochthoeca, ColoJiia, Knipolegiis, Phaeotriccus, 
Flavicola, and Ratnphotrigon. Am, Mus. Novitates, No. 930: 
1-27. 






MAR 1 8 198 



r- 



B R E V I 0-R-A 



useiiiii of Comparative Zoology 



Cambridge, Mass. June 1973 Number 403 

A NEW LATE PALEOCENE PHENAGODONT 

(MAMMALIA: CONDYLARTHRA) 

FROM WESTERN COLORADO 

Bryan Patterson and Robert M. West^ 

Abstract. A new genus and species of phenacodont condylarth, 
Prostheci&n Tuajor, is described from late Paleocene beds of the De- 
Beque Formation of western Colorado. Prosthecion is most closely 
related to the well-known small Paleocene-Eocene phenacodont Ecto- 
cion, although it also shows some similarities to the larger Phena- 
codus. The new genus contributes additional emphasis to the rather 
unique nature of the mammalian assemblage of the Plateau Valley 
local fauna. 

Introduction 

Among the fossil materials collected by Field Museum of Nat- 
ural History field expeditions to the early Tertiary of Mesa 
County, Colorado, in 1932, 1933, 1937 and 1939, are several 
specimens of a new genus of phenacodont condylarth. These 
specimens were recognized as novel by one of us, Patterson, who 
assigned to them a manuscript name, but this was not formalized 
by publication. The genus has been mentioned, however, with- 
out use of the generic name in West (1971: 5). The present 
paper establishes the new taxon in conjunction with West's re- 
view of the family Phenacodontidae. 

We thank Dr. William D. TurnbuU of the Field Museum of 
Natural History for the loan of the specimens described herein. 
The illustrations were prepared by the late John Conrad Han- 
sen. 



'Department of Biology, Adelphi University, Garden City, New 
York 



2 BREVIORA No. 403 

Order Gondylarthra 
Family Phenacodontidae 

Prosthecion^ gen. nov. 

Type species. Prosthecion major sp. nov. 

Included species. Type only. 

Distribution. Late Paleocene, western Colorado. 

Diagnosis. Moderate-sized phenacodont; approximately the 
same size as Phenacodus bisonensis and P. vortmani, larger 
than Ectocion wyoyningensis and E. osbornianum. Upper pre- 
molars with complete cingulum and smaller conules and para- 
style than in Ectocion. P^ shorter and wider than in Ectocion, 
with separate metacone and paracone. Upper molars similar to 
those of Ectocion. Ps with strong paraconid, distinct metaconid, 
rudimentary protostylid, and hypoconid surrounded posteriorly 
by a cingulum; heel broader than in Ectocion. Pi with hypo- 
conid more central in position than in Ectocion, entoconid not 
differentiated, rudiments of protostylid and metastylid present. 
Lower molars intermediate between those of Ectocion and 
Phenacodus, moderately robust and lacking well-developed para- 
conid, especially in posterior molars. 

Prosthecion major- sp. nov. 

Text-figures 1^ 

Type. FMNH P26131, LP^-M' (P^ M^"'-' incomplete), in- 
complete RP'\ M^; two lower Fs, R and LPs— Mi (incomplete). 
Found by William M. Harris. 

Hypodigm. Type plus FMNH PI 5586, part of left ramus 
with Mi-M:,; P15570, part of left ramus with P4-M2; P26129, 
LMi; P15581, RMi; P14940, RM2; P14947, RM3; P15011, 
incomplete RMi; P26066, trigonid of M2; P26128, lower molar 
fragments. 

Horizon and locality. Middle late Paleocene, lower part of 
Atwell Gulch Member of De Beque Formation^; Plateau Valley 
local fauna. Holotype from "Hell's Half Acre," 6 miles SSE of 
De Beque on the De Beque-Mesa road, W 1/2 sec. 14, E 1/2 
sec. 15, and NW 1/4 sec. 23, T.9S., R.97 W., Mesa County, 
Colorado. Remainder from scattered nearbv localities. 



Vpos^fTos additional, plus kluv, pillar; in allusion to the large para- 
conid of P3 and to the relationship with Ectocion. 
''Largest species of Ectocion group. 
^Wasatch Formation in Donnell, 1969. 



1973 A XEW PALEOCENE PHENACODONT 3 

Distribution. Late Paleocene of western Colorado. 
Diagnosis. As for the genus. 

Discussion. Prosthecion is a genus that combines certain 
characteristics of Ector ion and Phenacodus with others peculiar 
to itself. The cingula on the upper premolars readily differen- 
tiate Prosthecion from the other known phenacodonts. The de- 
gree of metacone development is similar to that in Ectocion, 
whereas Phenacodus has only slightly de\eloped parastyles and 
conules. Ectocion premolars are less robust than are those of 
Prosthecion. The upper molars, so far as they are known, are 
Ectocion-\\k&, and also show some resemblance to the earlier 
(and probably ancestral) Tetraclaenodon in the position of the 
metaconule and the small size of M^. 

P.! of Prosthecion is more adxanced than that of Ectocion in 
its possession of a metaconid, a strong paraconid, and a rudi- 
mentary protostylid. Talonid complexity here differentiates 
Prosthecion from Phenacodus. Pi is peculiar in the possession 
of stylids and in the central position of the hypoconid. The rel- 
atively small paraconid is a departure from the basic Phena- 
codus plan, but the paraconid is larger than in P4 of Ectocion. 
The molar paraconids form anterior ridges that are decidedly 
crenulated in P26129, less so in PI 4940, and smooth in the 
rest. Prosthecion is not so extreme as Ectocion, where the para- 
conid is absent and is replaced by a steeply sloping ridge. The 
trigonid basin is open antero-intemally in all specimens except 
P15011 and P15581, in which the paraconid ridge unites with 
the metaconid. Entost\lids are present in PI 5586 (except on 
Ms) and P26129, rudimentary in PI 4940, and absent in 
P15581. External cingula are lacking on most lower teeth, 
vaguely suggested on a few, and definite only on P26129. 

Prosthecion is clearly a close relative of Ectocion, which here 
is considered congeneric with Gidleyina; work in progress by 
West will provide e\-idence in support of this synonymy. This, 
then, lea\es two genera in an ^'Ectocion group" of phenaco- 
donts. Simpson (1937: 251, note 12) mentioned, and disap- 
proved of, a manuscript proposal by Gidley to separate the 
Ectocion group subfamilially from the Tetraclaenodon-Phena- 
codus lineage. While it is apparent that two rather distinct 
e\olutionary trends were being followed by the phenacodonts, 
specialists should be able to discuss such minor groups as these 
without giving them fomial hierarchic rank. 

Thus, with the addition of Prosthecion to the roster of late 



4 BREVIORA No. 403 

Paleocene phenacodonts, there were at least three genera de- 
scended from the common middle Paleocene Tetraclaenodon : 
Phenacodiis, Ectocion and Prosthecion. The first two of these 
are abundantly represented in later faunas, and Phenacodus 
underwent a moderate radiation, producing several size-differ- 
entiated species. Prosthecion, as yet unknown later than late 
Paleocene and outside of western Colorado, was either a mem- 
ber of an unsuccessful, short-lived lineage, or one that occupied 
environments as yet poorly sampled. The latter possibility is 
supported by the rather distinctive nature of the Plateau Valley 
local fauna. 

TABLE 1 

Measurements, in mm, of teeth of Prosthecion major 

P 26131 P 15570 P 15586 



P4 

Ml 

Mo 
Ms 



L 


7.7 






W tri 


4.3 






W tal 


4.5 






L 


8.6 


8.7 




W tri 


5.5 


5.8 




W tal 


5.8 


5.9 




L 




9.0 


9.4 


W tri 




7.5 


6.6 


W tal 




7.4 


7.1 


L 




9.0 


9.8 


W tri 




7.8 


7.4 


W tal 




7.5 


7.2 


L 


9.9 




9.7(e) 


W tri 


6.8 




6.6(e) 


W tal 


5.9 




5.9 


L 


7.4 






W 


7.6(e) 






L 


8.1 






W 


9.4 






e — 


- estimated 







References 

DoNNELL, J. R. 1969. Paleocene and lower Eocene units in the 
southern part of the Piceance Creek Basin, Colorado. U.S. Geol. 
Surv. Bull, 1274-M: Ml-18. 



1973 



A NEW PALEOCENE PHENACODONT 



Simpson, G. G. 1937. The Fort Union Group of the Crazy Moun- 
tain Field and its mammalian faunas. U.S. Nat. Mus. Bull. 169: 
1-287. 

West, R. M. 1971. Deciduous dentition of the early Tertiary 
Phenacodontidae (Condylarthra, Mammalia). Am. Mus. Novi- 
tates, No. 2461: 1-37. 




B 




Figure 1. Lower premolars of Prosthecion major, P26131, tyi^e 
specimen. Line indicates 5 mm. 

A. Occlusal view, P3 and P4. 

B. Ling-ual view, P3 and Vi. 



BREVIORA 



No. 403 




Figure 2. Occlusal view of lower molar dentition of Prosthecion 
major. Line indicates 5 mm. 

A. P15.586, M,-M.. 

B. P26121, type specimen, Mj. 




Figure 3. Upper dentition of Prosthecion major, P26131, type speci- 
men, P"'-M-. Line indicates 5 mm. 



1973 



A XEW PALEOCEXE PHENACODONT 











OS 

o 

o 



'o 






c 



• ^ C 



o 



ti-, = C; 01 



> 




15 

3 


■ss~5 


hr 


ij - » 


r 


C w ^H 




•r: 00 CO 

1— 1 LO ^H 




Lf O 




^- th oa 


■^ 


© HH fl^ 




•»» 






9 
-1 


1 <■ cc 


a 





1 ' r->i ■ 



U ■:■: 



MAR 1 8 1965 

B R E V I O R A 



Museum of Comparative Zoology 



CAMBRrocE, Mass. June 1973 Number 404 

NEW STUDIES ON A MONTANE LIZARD 
OF JAMAICA, A NO LIS RECONDITUS 

Robert Hicks^ 

Abstract, Data on the natural history of Anolis reconditns, an 
arboreal ignanid lizard, living in the montane mist forest of east 
central Jamaica, is presented. The animal uses the crown of trees 
both for sleeping and as a refuge when frightened. It pei-ches fac- 
ing down on open tree trunks, generally about 12 feet off the ground. 
Its displays appear to vary a great deal from individual to individ- 
ual. This animal seems to use a combination of two tyioically anoline 
foraging methods: 1) active searching and 2) motionless perching. 
In addition, reconditus displays a characteristic reported in no other 
Jamaican anole: it does not interrupt its daily activities because of 
rain. The lizard appears to be less specialized than its Jamaican 
congeners. This may be related to its apparent isolation fi-om con- 
generic competition. 

INTRODUCTION 

Of the seven species of anoline lizards in Jamaica, Anolis re- 
conditus is apparently the rarest and certainly the least studied. 
This relatively large {ca. 90 mm snout-vent length), montane 
reptile was first described in 1959 by Underwood and Williams 
from a single male specimen. The animal is usually green, 
brown, and beige. It is mottled and striped laterally and dorsally, 
often with a prominent small green patch just behind each eye. 

The type specimen was collected in 1953 in St. Thomas par- 
ish, on the edge of the Blue Mountain rain forest region in east 
central Jamaica. In 1961 and 1962 A. Stanley Rand collected 
five Hzards at Hardwar Gap, nine miles west of the type locality. 
In 1965, J. D. Lazell captured five additional reconditus, in- 
cluding the first females, in the Hardwar Gap area; he pub- 
lished a paper (Lazell, 1966) that added substantially to the 

'219 Steeplechase Road, Devon, Pennsylvania 19333 



2 BREVIORA No. 404 

scanty data on coloration, squamation, and behavior of this liz- 
ard. Robert Trivers in 1969 collected, marked, and released 
several specimens at the Green Hills field station of the Institute 
of Jamaica, one mile northeast of Hardwar Gap. 

In 1970, I spent the months of April and May in Jamaica 
observing all seven species of AnoHs on the island for varying 
periods of time. I spent two weeks with Robert Trivers, who 
was working on A. garmani, and two weeks with Thomas Jens- 
sen, who was studying A. opalinus, A. lineatopus neckeri, and 
A. sagrei. Between April 26 and May 28, I lived three weeks 
at the Green Hills field station, near Hardwar Gap. During this 
time I caught, marked, and observed 32 reconditus. I saw at 
least another two dozen. The data that follow are the results 
of my study. 

HABITAT AND RANGE 

During my stay at Green Hills, I discovered some specific 
new localities in the Hardwar Gap region which have substan- 
tial populations of this species. The population that Lazell 
(1966) describes as living in the garage near the field station, 
I found in the garage and on a stand of trees just adjacent to 
and east of the garage. I also saw one animal living under the 
eaves of the Denoes' house, which is just east of the field sta- 
tion and across the road. 

One of my main study areas, referred to hereafter as "Steep 
Bush," was located about one-quarter mile southwest of the 
field station along the road to Hardwar Gap. The road is cut 
into the mountainside, which slopes 50° or more. The viney, 
densely wooded "montane mist forest" (Asprey and Robbins, 
1953) is called "bush" by the local people. The area where I 
worked on seven difTerent days is on the top side of the road. 
I concentrated on a roughly square area that runs along the 
road for about 300 yards and extends up the mountainside an 
equal distance. 

The trees there are generally 30 to 50 feet tall, with no 
branches below 20 feet. Several feet of their root systems are 
exposed as they cling to the steep mountainside. Moss covers 
the rocks, much of the tree trunks, and the fallen trees and 
limbs. There are also 15-foot tree ferns, vines, and saplings. 
The canopy is quite dense and everything, including the deep 
leaf litters, is very damp. The trees are generally plane trees. 
Wild yacca (Podocarpus urbanii), milkwood (Sapium jamai- 



1973 



ANOLIS RECONDITUS 




Map 1. Hai'dwar Gap-Green Hills vicinity. Note collecting sites 
for reconditus marked by triangles. Open squares are buildings. 



cense), and trees known to the local people as "fleura" are 
most abundant. 

In Steep Bush, I caught ten lizards and made many observa- 
tions. However, movement for an observer is extremely difficult 
in this "montane thicket" (Asprey and Robbins, 1953), and, 
owing to the mazelike quality, resightings are nearly impossible. 

My best study area was a place, known locally as "Fairy 
Glade," one-quarter mile east of Hardwar Gap. It is on top 
of the ridge that lies to the east of the Gap and is the largest 
area of flat "bush" land that I could find within several miles 
of Hardwar Gap. The glade is about 10 acres in extent, and 
has a path 10 to 30 feet wide running through it which leads 
northeast to St. Catherine's Peak. 

Lazell (1966) quotes Rand as saying he found reconditus in 
"mossy, stunted forest — nearly 'elfin woodland' — at Hardwar 



4 BREVIORA No. 404 

Gap." This is an excellent description of the Fairy Glade. Al- 
though generally found over 5000 feet, elfin woodland occurs 
as low as 2500 feet in the John Crow Mountains, which are 
east of the gap. Asprey and Robbins (1953) regard elfin wood- 
land as a faciation of montane mist forest. Fairy Glade repre- 
sents an intermediate stage between the montane mist forest of 
the steeper, lower slopes and the dwarfed, stunted, typically 
elfin woodland of the higher altitudes. The omnipresent moss 
is often six to eight inches deep on the ground and on the rot- 
ting fallen trees and limbs which litter the ground. 

The trees in the glade are generally of large diameter with 
full foliage but are occasionally twisted, unlike the tall, straight, 
branchless trees of Steep Bush. The growth of the glade could 
certainly be described as lush. An early botanical observer, in 
describing similar Jamaican vegetation, spoke of "the confusing 
abundance of species and the rich plant life in a muggy, glass- 
house atmosphere." (Asprey and Robbins, 1953) . 

I caught and marked 15 lizards along a 200-yard stretch of 
the path which runs through the glade. Most of my data are 
for these lizards. 

I also observed lizards at two other localities in the Hardwar 
Gap region : ( 1 ) in the northwestern part of Holywell Park, 
just west of the gap, and (2) in the "19 Mile Gulley," a steep, 
dark, damp, rocky ravine about four miles southeast of the gap 
and along the road to Kingston, just below the 1 9-miles-f rom- 
Kingston roadside marker. This latter area is several hundred 
feet lower than Steep Bush but has about the same physical 
character and vegetation. In half an hour on my final day in 
the Hardwar Gap area, I located four large males and one fe- 
male in this gulley. In Holywell Park, I saw only one animal, 
a male. 

With the, exception of Lazell's garage (see above), I was 
directed to all the places where I found reconditus by local Ja- 
maicans. Consultation with those local people who frequent 
the "bush" could in the future lead to a better understanding of 
the extent of the lizard's range. 

It seems likely that reconditus extends over a narrow, per- 
haps discontinuous band of montane mist forest running east 
and west of Hardwar Gap. The band is at least two or three 
miles wide in places. The Green Hills area is half to three- 
quarters of a mile north of the gap and the Newcastle area 
one to one and a half miles south of the gap. 



1973 ANOLIS RECONDITUS 5 

SIZE AND SEXUAL DIMORPHISM 

During ni\ month of study, I obscncd at \arious times more 
than 50 diflerent indixiduals. Thirty-two of these were cap- 
tured; 29 were measured. 

Of the 1 7 measured males, the largest had a snout— \ent length 
of 100 mm, the smallest 61 mm. The largest lower jaw (tip of 
snout to back of lower jaw), 33 mm, was that of a 98-mm 
male, the shortest, 20 mm, that of the smallest male (61 mm). 
The longest tail, 220 mm, was recorded on a 96-mm male; the 
shortest unbroken tail was 115 mm, on the smallest individual 
(61 mm) captured. 

Of the twelve females that I measured, the longest snout- 
vent length was 84 mm. This lizard had the longest lower jaw, 
26 mm, of any female. The longest tail, 164 mm, was on a 
77-mm lizard. The smallest female, smallest in each of the three 
measurements taken, was 62 mm snout-vent with a 19-mm 
lower jaw and an 121-mm tail. 

The body length of reconditus males is 1.17 times longer than 
females and the lower jaw of males is 1.19 times longer than 
females. Robert Trivers (personal communication) has found 
that for A. garmani in southwestern Jamaica, the adult males 
are 1.31 times longer than the adult females. 

COLOR 

AVhen describing A. reconditus, Lazell (1966) emphasizes a 
grey component in the coloration of the animal. According to 
my observations, a more accurate description of the lizard is 
that a tan or brown component predominates. I never observed 
any part of a lizard to be black or grey. Often parts of the liz- 
ard would be \ery dark brown. These appeared tan when the 
lizard changed to a lighter color phase. In my field notes I 
once described a cr\ptic male perched on a grey tree trunk as 
"brown, greyish mottled." However, I still maintain that close 
obser\-ation re\-eals only shades of brown, never grev. 

Interestingly, the populations in m\ main stud)' areas, three- 
quarters to one mile apart and ph)sicall\ isolated from each 
other only b)- hea\\ undergrowth and o\ergrown guUeys, seem 
to differ in color. The males of the Steep Bush area appear 
generally light brown with quite dark brown markings and 
stripes. They are green or slighdy bluish green just behind the 
forelegs. They can change to a mossy dark green or to a dark 



6 BREVIORA No. 404 

chocolate. With both these color changes, the markings fade 
out. 

The females of Steep Bush are lighter tan than the males, 
with prominent green patches on the head, just behind each 
eye. There is also a green streak just above each foreleg. The 
females can appear quite beige with the dark markings almost 
completely suppressed. Larger patches of light green on the 
upper body can appear with this lightening. The females can 
also display dark speckled markings over the tan, with green 
remaining only on the head and above the foreleg. The dewlap 
of the female, which is about half the size of that of the male, 
is yellowish green. 

The animals of Fairy Glade are generally darker in colora- 
tion than those of Steep Bush. The glade receives more sun- 
light than the bush forest and this may account for part of the 
color difference. Both males and females have the prominent 
green patches behind their eyes that characterize the females 
of the bush described above. In addition, the lower body of 
the males from just in front of the hind legs to the tip of the 
tail is fairly green, while the mid-body is mostly brown. 

One Fairy Glade female I described in my notes as being 
"a dull, mousey brown" with a bit of muted green on the 
shoulder and behind the eyes. I observed a 74-mm Fairy Glade 
female in a completely chocolate brown phase, with no mark- 
ings evident. She was delicately rosy red behind each leg, just 
where the leg joined the body. Females, as well as males, in 
Fairy Glade seemed also to have the power of males in Steep 
Bush to turn to a cryptic dark chocolate. In this phase the ani- 
mals have no markings evident. All color except the brown is 
suppressed. 

In Holywell Park, about a mile west of each of the two other 
areas, the one male I saw was very green over much of his 
body. As is typical of reconditus, his head and legs were basical- 
ly brown, but his trunk reminded me of the bright green of 
A. garmani. Markings were evident as patches of darker green 
on his back and sides. The striking shade of most of his body 
marked him as a member of possibly a third distinct population. 

These data are too scant for speculation about whether re- 
conditus occurs in distinct demes each with a specific color pat- 
tern. In view of the broken terrain, I may have been sampling 
separated segments of a continuum. More work is needed before 
a positive statement can be made about the local distribution 
and differentiation of reconditus. 



1973 



ANOLIS RECONDITUS 




Figure 1. Male perching- (Holywell Park). Note proximity of his 
perch to the hole in the trunk. See escape behavior. 



8 BREVIORA No. 404 

PERCHES 

Although I observed reconditus on a N^ariety of perches in a 
variety of positions on the perches, there was one tableau which 
I observed most frequently. The lizard would be about 12 feet 
up on a tree trunk or large-diameter branch (more than 4 
inches) "in a head-down position, with the neck bent so that 
the head was nearly parallel to the ground" (Lazell, 1966). 
A. S. Rand first reported this stance to J. D. Lazell, but re- 
ported the perch height as from three to seven feet. 

I have two dozen observations of height and diameter of spe- 
cific perches, 1 1 for females and 1 3 for males. According to 
these I saw only four females and two males perched under 
seven feet. On the other hand, 16 of the perches were ten feet 
or over. The highest was about 16 feet. I have four additional 
entries with only perch diameter. I made many additional ob- 
servations for which I failed to record perch height or diameter. 
In such cases the lizard was perched, facing down, between 10 
and 14 feet on a four-inch or larger diameter perch. I had 
come to expect them to be at this height and diameter and 
onlv if thev were not there did I make note of it. 

As stated above, the standard perch was of fairly large diam- 
eter. I recorded at least 14 different perches of eight inches or 
larger. The largest was about 18 inches. 

There were a few exceptions to the standard downward-fac- 
ing posture. I saw one male and one female who were perched 
for a long period of time facing up. I saw several perched cross- 
wise, coming out of rotten furrows or at a crook in a tree branch. 
In some cases, crosswise or upward-facing perches would be 
assumed for a minute or two as transitional positions between 
two different downward perches or between periods of mo\e- 
ment. 

One striking and possibly unique element in the site that re- 
conditus chooses as a perch is that in a great number of cases 
it is near a hole in a tree trunk or branch (See Fig. 1 ) . In the 
Hardwar Gap area with its frequent rainfall and lush vegeta- 
tion, it is common for a tree to have a partially exposed root 
system and many dead moss- and vine-covered limbs. Holes 
into the heartwood of a tree where a branch used to be are 
quite frequent. I counted no less than eight separate perch 
sites each within a few inches of prominent trunk or branch 
holes. One of these had three marked lizards, two females and 
a male, perched by it at different times. Another hole had two 



1973 ANOLIS RECONDITUS 9 

marked animals, one male, one female, who regularly used a 
perch next to it. Onh one lizard at a time was ever seen at any 
of the holes. 

Several times I saw lizards perched half in and half out of 
these furrows or holes. Rand ( 1967) reports that A. valencienni 
lays eggs communally in tree-trunk holes but apparently' no 
other anoline lizard uses them in just the same way as recondi- 
tus. It is possible that the ho'es are related to the foraging activ- 
ity of reconditiis (see foraging). They are certainly hiding 
places (see escape behavior). 

The fact that more than one individual uses the perch near 
a particular hole raises questions concerning the nature of the 
territoriality of this species. Perhaps there is only one male per 
hole and the females who use the perch are part of a group 
that resides within his territory. Robert Trivers (personal com- 
munication) has obser\ed for A. oarmani that indixidual males, 
especially large ones, seem to police territories containing several 
females. Each male appears to maintain exclusive copulatory 
rights with the females in his territory. Perhaps there is a sim- 
ilar situation with reconditus (also see home range and ter- 
ritorial structure). 

The holes raise another question. If a lizard is going to es- 
cape into a hole, instead of up a tree, he has to be sure that 
the hole is not already occupied. Indeed, the intimacy of a hole 
is a great deal more than that of the crown of a tree, and a liz- 
ard cannot see from a distance if he will ha\e company in a 
hole, the way he can at the top of a tree. Therefore it seems 
likely that at any time the lizards on a particular tree must have 
a good idea of where the other lizards on that tree are perched 
and where to go in case of danger. 

HOME RANGE AND TERRITORIAL STRUCTURE 

In an attempt to learn something about the habits of A. re- 
conditus, especially about its territoriality, I noosed and marked 
individual animals. After its capture, the sex of each lizard was 
determined, it was measured, given a number, and marked in 
two ways. For a permanent mark, part of one or two toes was 
clipped (see marking system at end for details). This allowed 
me to give each indi\idual a unique identification number. 
Also stripes of paint of various colors were painted on the 
lizard's back so that different indi\'iduals could be distinguished 
at a distance. The location of the perch where the lizard was 



10 BREVIORA No. 404 

caught was recorded and then the animal was returned to this 
site. The lizards were rarely held capti\e for more than fifteen 
minutes. 

During my study I resighted nine marked lizards. One, a 
62-mm female, I saw seven separate times. All the resightings 
were in Fairy Glade. Although my data is scanty, it points to- 
ward a large male having a home range shared by two or more 
females, three large trees (30—40 feet X 12—18 inches) and a 
few saplings. In the heavy growth of Fairy Glade, the ground 
area of a range might be only 500 to 750 square feet. 

I have no data on whether reconditus defends these areas or 
any parts of them. Robert Trivers (personal communication) 
has found that males of A. garmani do defend territories, slightly 
larger in extent than the reconditus home range described here. 
Further study may reaveal whether reconditus also defends a ter- 
itory. 

On two very large adjacent wild yacca trees, each of which 
had two 12-18-inch trunks, I regularly saw a 74-mm female, a 
77-mm female, and an 88-mm male. The larger female I saw 
on six different occasions, five times on the same tree, four times 
on the same perch. Once I saw her on the adjacent tree, oc- 
cupying the perch on which I had often seen the male. The 
male I also saw half a dozen times, four times on "his" tree 
and twice on the perch where I usually saw the female. The 
small female I saw only twice. The second time she was on the 
perch on the "male's" tree which had been used by both other 
lizards. 

The smallest female (62 mm) that I caught I saw six times 
at exactly the same perch. This was a small tree (four-inch 
diameter), five feet away from the tree on which I first caught 
her. She was always perched four feet ofT the ground next to 
the same furrow in the tree trunk. 

I saw an 81 -mm male three times, always within ten feet of 
where I first caught him. He was caught on a straight, branch- 
less tree {ca. 30 feet X 6 inches) and resighted on two adjacent 
trees, one a sapling (one- to two-inch diameter) and one a small 
tree (three-inch diameter) . 

I repeatedly saw a 95-mm male and a 75-mm female on or 
in the vicinity of two medium-sized ( 8 inches X 35 feet each ) , 
well-foliaged, adjacent trees. They spent most of their time at 
one perch, nine to ten feet up, next to a hole in the tree in 
which I first caught the male. One day I would see one on the 
perch, another day I would see the other. 



1973 ANOLIS RECONDITUS 11 

One large male (95 mm s-v; 30 mm jaw) I saw on three 
successixe days reveal either an enormous home range or rather 
atypical beha\ior. I caught him one day about 8 a.m. along 
the path, close to the entrance of Fairy Glade. Three hours later, 
I saw him again 12 feet further up the path. He must have 
crossed se\eral large trees on his way to this new perch. The 
next day I saw him on the other side of the path, about halfway 
between the two trees on which I had seen him the day before. 
Four hours later he was nowhere to be found. 

On the third day I spotted him at noon on the side of the 
path on which I originally caught him but ten feet further up 
the path than I had e\er seen him. I proceeded to follow him 
and photograph him for two hours as he covered an 80-foot 
stretch which contained many large trees. My field notes that 
describe this remarkable journey follow. 

May 17, 1970 
Fairy Glade 

1 2 : 00 noon — ^ # 5 (95 mm s— v ; 30 mm jaw ) 

Spotted 10 feet further up path (on side he was orig- 
inally caught on) than seen prexiously. 

12:10 — Following #5. Moved up beyond tree where #11 
(62-mm female) was caught. 

12:35 — Much head bobbing after assuming new perch. Many 
^ inch diameter sapling perches. 

1 2 : 45 ■ — Ate some ants. 

12:50- — Dewlapped 3 /a times. 
1:10 — 5 dewlaps preceded by head bobbing. 
1:15 — Moxed further up path and dewlapped 4 times with 
head bobs. Sitting out on tree fern fronds, 10 feet up. 
1 : 25 — Taken up downward facing perch 1 2 feet up on 3 inch 
diameter tree. Brown, greyish mottled shading very 
much color of the bark. 
1:50 — Still perched. Light rain. I left. #5's run started 
about 30 feet further up path than tree he was caught 
on. Run documented abo\e roughly followed path up 
farther into glade. Lizard covered 80 feet in an ap- 
proximately linear fashion. In that 80 feet were about 
30 trees greater than Ij/o inches in diameter and 
many, many smaller trees. On other side of path 
where #5 was observed perching on May 16, this 
same 80-foot stretch contained six large trees. The dis- 



12 BREVIORA No. 404 

tance between the two rows of trees, i.e., distance 
across the path, was about 40 feet. 

Clearly much more work has to be done to establish a rea- 
sonable knowledge of reconditus' home range size or territorial 
structure. I caught no other Hzards, except the 62-mm female, 
in the vicinity of the trip of the large male described above. It 
is not impossible that all the ground he covered was "his." It 
is also true that my best resighting data is on females who seem 
to occupy one or two adjacent perches quite regularly. You 
might expect a female to spend more time in a specific spot in 
a male's home range while the male himself roams around more 
within the area. Perhaps I did not see a male regularly because 
he was much of the time out of my view, in some other part 
of his quite large home range. 

It is difficult to get an idea of the size of the reconditus popu- 
lation of Fairy Glade. If I had a good estimate, I would under- 
stand the dispersal of the species better. I only caught fifteen 
lizards, eight males and seven females, along the 200 yards of 
the path which I used for my study area. If that represents 
half or even possibly a third of the resident males, then it is 
possible that reconditus males have a large home range, some 
even as big as that of the particular large male I have described. 
I am quite certain that there were many more females in my 
area than I was able to catch because they are smaller, more 
cryptic, and have a less obvious display. Owing to the cryptic 
nature of this species, the general lush growth of the habitat, 
and the apparently quite timid character of the animals once 
they have been exposed to man (see escape behavior), 
I think it is very likely that I missed many more lizards than I 
caught. This would indicate that my conservative estimate of 
home range size, at the beginning of this section, is a reason- 
able one. 

INTRASPEGIFIG INTERACTION 

I rarely had more than one reconditus under observation at a 
time. On only two occasions did I observe interaction between 
two lizards, in both cases between a male and a female. 

The first interaction was one morning in Steep Bush. Initially 
the male ( ca. 90—95 mm ) was ten feet up on a tree trunk ( eight 
inches in diameter), facing down; the female (84 mm) was 
about seven and a half feet up on the same tree, also facing 
down, about 120° to the right of the male. After I had been 



1973 ANOLIS RECONDITUS 13 

watching them for half an hour, the male dewlapped once. 
Fifteen minutes later he ran up the tree two feet, dewlapped 
three times and assumed an upward-facing perch. After an- 
other {\\c minutes, the male ran down about a foot and toward 
the female slightly. He gave three head lx)l)s, paused and gave 
three more. His downward-facing perch was now about three 
and a half feet above and 60° to the left of the female. 

Ten minutes later, the female finally moved, running down 
and around the back of the tree. A bit of vine fell behind the 
tree and then she returned. Probably she was eating. For the 
next forty minutes the two lizards remained motionless. Then 
the male moxed down a few inches toward the female, but also 
to the left, putting more lateral distance between himself and 
her. He dewlapped three times, the second and third being from 
a half open position of the dewlap. Fixe minutes later the fe- 
male opened her mouth. After another fixe minutes, the male 
dewlapped again three times in a manner similar to his earlier 
series. 

Although the female appeared to display no direct response 
to the male, it looked to me as if the male was courting her. 
Neither before this occasion nor after did I see two lizards so 
close together. Because of the extreme peripheral vision of these 
lizards, it seems likely that the female cou'd have been watching 
the male's displays despite his position abo\'e and behind her. 
Although I was quite close at the time, nothing about the male's 
behavior suggested that he was reacting to me. He mav of course 
have been making territorial displays either in general or to spe- 
cific other hzards that I could not see. 

On the second occasion the object of the male's displa\' was 
e\'en more aml:)iguous. I caught a 74-mm female about ten feet 
from where an 88-mm male, previously caught and marked, 
was perched. Both lizards were on different trunks of the same 
large tree. 

A minute or so after I returned the newly marked female to 
the perch from which I had captured her, the male began one 
of the most elaborate displays I ha\'e e\er seen. He dewlapped 
five times, head-bobbed quickly twice, dewlapped once again, 
head-bobbed several more times, and then dewlapped once more 
while bobbing. Finalh he head-bobbed and then dewlapped 
an eighth time. 

Five minutes after this display the female assumed the exact 
perch and position she had occupied before I captured her. In 
the next fifteen minutes both the male and the female rotated 



14 BREVIORA No. 404 

their perches through 90°. Then twenty-five minutes after 
his grand display, the male resumed his original position and 
dewlapped two and a half times. 

The male's elaborate display was certainly in response to 
either myself or the female. If it was an aggressive display to- 
ward me, it was a minute or two after I had put the female 
back on the trunk and returned to my close-by observation post. 
Normally, if a lizard saw me he would sit and wait motionless, 
and then rapidly move away if I got too close. This makes me 
tend to think that the male was reacting to the female to whom 
I had specifically drawn his attention. The movements of the 
male and female described above after the male's display may 
also have been part of the courtship interaction. 

DISPLAY 

I observed no pattern to the displays of reconditus. In gen- 
eral the displays are, in regard to specific body movements, 
much like those of other Jamaican anoles. There is head-bob- 
bing; four-leg pushups; a front-legged, half pushup and pulled 
back, slightly cocked head preceding dewlapping; dewlapping 
combined with head bobbing; and sometimes, at the end of a 
series of dewlaps, a half dewlap pulse. On characteristic of the 
species is that the displays are slower and seemingly more 
"methodical," as I called them in my field notes, than those of 
other Jamaican lizards. 

I saw females display only three times; each time the lizard 
I was observing head-bobbed on assuming a new perch. One 
of these females I watched for nearly four and a half hours 
and saw only the single head-bob display. 

Males displayed much more frequently, generally dewlapping 
intermittently while they maintained a specific perch. One in- 
dividual that I watched for three hours dewlapped five or six 
times at various intervals, sometimes every ten to twenty minutes, 
sometimes every 30 to 40 minutes. Two different displays of 
possible courting males are discussed in intraspecifig inter- 
action. Although I recorded that a 95-mm male head-bobbed 
and dewlapped each time he assumed a new perch (see home 

RANGE AND TERRITORIAL STRUCTURE ), I also obsCrVCd malcS 

take new perches without display. Occasionally males directed 
what were probably aggressive displays toward me. These con- 
sisted of dewlapping once or twice. 

The elaborate display of one male just before he "went to 



1973 ANOLIS RECONDITUS 15 

bed" is worth describing. I had been watching this lizard con- 
tinuously for three hours. He had not moved for the last hour 
while darkness fell. Suddenly he shifted his downward-facing 
position on a three-inch trunk 30° to the right and began a 
rapid display: three head bobs, a dewlap, two more bobs, a 
dewlap, one bob, and finally two dewlaps. Seven minutes later 
he ran up two feet on the trunk, assumed a downward position, 
and again displayed: four pronounced head bobs followed by 
three dewlaps. Three minutes later he ran up two feet more, 
stopped and looked around. After a brief pause, he ran up six 
feet more, stopped, paused, and then continued up the trunk, 
disappearing in the dark foliage. 

It seems likely that males, and females to a lesser degree, dis- 
play to announce their position in their home range or territory 
to other members of the species in typically anoline fashion. 
But I could ascertain no pattern to this type of display, and my 
data is not extensive enough to identify consistent features of 
aggressive or mating displays. 

INTERSPECIFIC INTERACTION 

On only one occasion did I see reconditus in close proximity 
with a lizard of another species. In the stand of imported yac- 
cas, next to the garage where Lazell worked, I saw a large male 
A. garmani perched 20 feet up on a tree. A foot away on an 
adjacent tree, perched slightly lower, was a male reconditus 
that I was trying to catch. Contrary to LazelFs speculation 
(1966), it seems that the niches of garynani and recoriditus do 
overlap at least occasionally. 

On the steep path up to Fairy Glade from the Hardwar Gap 
road I also once saw a male garmani perched 30 feet up on a 
tree. I never saw any reconditus on the trail to the glade, but 
this garynani was less than 100 feet below the glade. It is pos- 
sible that the two species share the same habitat at the outer 
edge of Fairy Glade. I saw neither A. opalinus nor garmani in 
Steep Bush, inside Fairy Glade, or in the Nineteen-Mile GuUey, 
the three areas that I located with substantial reconditus popu- 
lations. It may be significant that of all the anoles in Jamaica, 
only reconditus lives for the most part without congeners (see 
discussion). 

ESCAPE BEHAVIOR 

Lazell (1966) describes the "wary and timid behavior" of 



16 BREVIORA No. 404 

reconditus. I too found this to be true, but usually only after 
the lizards had become "wild," as the local people refer to the 
lizards after thev ha\e become fris:htened of man. Generallv. 
"naive" lizards, individuals who have had no contact with man, 
are quite easily approached, and with a little care easily caught. 
Owing to their cryptic nature and restricted range, they are 
often hard to find, but once you find them, you usually can catch 
them. 

However, after you have caught, measured, and marked them 
once, even if you can find them again, which is dif^cult, you 
will have an incredibly hard time recapturing them. One day 
in Fairy Glade I saw a 75-mm female which three days pre- 
viouly had been caught and marked. She edged her way into 
the trunk hole next to her perch. I was 20 or 30 feet away \vhen 
this happened, and it suddenly occurred to me that the reason 
I was resighting few marked lizards was because they would see 
me coming and hide. 

I only made one recapture, that of a 78-mm female whom I 
had originally caught and measured, but who had escaped 
before she was marked. I attempted to recapture her half a 
dozen times before I succeeded. She grew more and more warv. 
It is not surprising at all that Lazell ( 1 966 ) had trouble catch- 
ing the lizards in the garage and under the house. Most Ja- 
maicans fear lizards and try to kill them or drive them away 
whenever they encounter them. Lizards are especially dishked 
around human dwellings. LazelFs lizards were probably quite 
afraid of humans. 

I often saw reconditus next to the garage he describes in his 
paper, in a stand of imported yacca trees. Although I tried 
more than a dozen times to catch some of them, they would in- 
variably escape up the straight 40-foot trees. I finally caught 
one 77-mm female on the roof of LazelFs garage. I was very 
lucky indeed to do so. I heard reports that the sons of the care- 
taker of this garage threw stones at the lizards on the trees next 
to the building. The boys were afraid of the harmless animals, 
and I could not persuade them of the folly of their actions. The 
lizards around the garage were definitely "wild." 

Although Lazell in his paper agrees with Underwood and 
WilUams (1959) that reconditus usually escapes "under objects 
and into holes," I found this to be only partially true. In the 
great majority of cases, I observed reconditus, when attempting 
to evade my noose, to go up the tree in typical Anolis fashion. 
On four specific occasions, three times with males and once with 



1973 ANOLIS RECONDITUS 17 

a female, I watched lizards go into trunk holes, conveniently 
next to and at least as high off the ground as their perches, to 
hide from me. I nexer chased a lizard down to the ground or 
under the exposed roots of a tree, seemingly an adequate hiding 
place. They invariably went up. 

REACTION TO WEATHER 

The "moisture-laden easterly trade winds," which blow all 
year long bring the Hardwar Gap area of the Blue Mountains 
over 200 inches of rain annually. During much of the fall and 
spring, there is daily rain. Although there are occasional dry 
davs, it usually rains for several hours during the day and some- 
times for several days continuously. The area is described as 
"montane mist forest" because of the mist or fog which envelops 
it every day. The mist remains even during the rain and lasts 
for 30 to 70 percent of the daylight hours, often forming shortly 
after dawn and remaining until an hour before sunset (Asprey 
and Robbins, 1953). 

Because of the weather conditions and also because of the full 
canopy of the forest in which the lizards live, it is very difficult 
to get any idea of how reconditus reacts to heat or sunlight. In 
all my observation time, I very rarely saw a lizard perched in 
direct sunHs^ht. This was not because the lizards chose to a\oid 
it by picking shady perches, but because the sun was rarely shin- 
ing; if it was sunny, the canopy usually prevented direct sun 
from striking the lizards. On the few occasions when I did 
see sunlight on the lizards, they seemed to ha\e no apparent 
reaction to it. Basking would have been futile since the sun 
rarely shone more than a few minutes, if that, in one place and, 
as I said, seldom shone at all. Further study should be carried 
out during the summer months, which according to the local 
people are drier, often 80° or more, and sunny. It would be 
interesting to see if the sun ever gets hot enough in the bush to 
drive the lizards to seek shade and damp. 

The local name for reconditus is "water lizard." The animals 
seem to prefer the damp, often living in mountain-side gulleys 
where water flows during heavy rains. But to truly grasp the 
significance of the local name, you have to watch one in a drench- 
ing downpour: water streams off the body of the lizard, while, 
with no apparent concern, it remains passively perched. On 
three separate days of heavy rain I watched a total of seven in- 
dividuals display no apparent reaction to the precipitation. Four 



18 BREVIORA No. 404 

of these were females; three were males. One of the females 
was active during an entire hour of downpour, but her move- 
ments seemed to have nothing to do with seeking shelter. She 
was also quite active before the rain and during its tapering- 
off stages. 

On numerous other occasions I saw lizards in light rain show 
no concern for the weather. Only once did I see a female run 
down a tree and under the exposed roots during a heavy rain. 
She stopped every few feet during her descent to head-bob. A 
few minutes before, I had unsuccessfully tried to capture a large 
male and chased him up into the crown of the same tree. It is 
not clear whether the female's actions were related to the heavy 
rain or to the frightened male I chased up the tree. 

In any case the tolerance for rain which rcconditus seems to 
have is unusual in anoles. Robert Tri\'c^=, in his extensixe studies 
of A. garmani and A. vaJencienni in jcii.iaic?, reports (personal 
communication) that it is nearly impossible m rainy days to 
find even one individual in an area w'tli a laige, well-studied 
population. However, Thomas Schoencr (personal communi- 
cation) has seen A. sagrei perched in the rain. Still, every ob- 
server of anoles that I have talked to has corroborated the usual 
pattern as avoidance of the rain. 

Since A. opal inns is smaller and A. garmani is larger than the 
"water lizard,'' apparently size cannot be the determining factor 
in tolerance for rain. The habitat inxoh'ed is undoubtedly im- 
portant. The tolerance that reconditus has seems to me a rea- 
sonable adaptation in an environment where wet weather is so 
frequent. A great deal of energy would be wasted if the lizard 
had to mo\'e to shelter every time it rained. 

FORAGING 

In my many hours of obser\'ation, I saw only three instances 
of feeding bv reconditus. 

An 84-mm female was perched se\en and a half feet off the 
ground, facing down, on a 45-foot X 8-inch tree. I had been 
watching for half an hour when suddenly she turned 180° 
around and seemed to take something off the bark with her 
mouth. Then she "chewed" (opened and closed her mouth) 
four times and then turned back around to resume the stand- 
ard downward-facing perch. Half an hour later she ran around 
the back of the tree and down. A bit of \ine fell from the back 
of the tree. Then she returned. All of this took place in about 
a minute. Presumably she had gone after a bit of food. 



1973 ANOLIS RECONDITUS 19 

On another occasion I watched a 74-mm female traverse six 
feet of a rotten furrow in a horizontal tree trunk. The trunk was 
about eight inches in diameter and four feet al)o\c the ground. 
The Hzard licked the rotten wood inside the furrow, perhaps 
"tonguing" ants. Then, after assuming a horizontal perch fac- 
ing out of the furrow at one end of the trunk, the female snapped 
out of the air a mosquito which was buzzing around her mouth. 
Ten minutes later, she caught another, larger, flying insect, pre- 
sumably a fly. Half an hour later she withdrew from this perch 
back into the furrow and proceeded to tra\erse the length of 
the trunk again, inside the furrow. She made the trip quite 
slowly, moving from side to side as if searching for something. 

The final instance of feeding I observed was by a 95-mm male. 
I saw him eat some ants while he was making a long, relatively 
fast-mo\'ing journey of 80 feet through densely forested Fairy 
Glade (see field notes in home range and territorial struc- 
ture ) . The eating of ants appeared at the time to be a digres- 
sion from the business at hand, that is, his long, steady trip up 
the path. He ate only once in two hours. However, it is not 
impossible that the whole trip was a food search. 

Robert Trivers has shown (personal communication) that A. 
valencienni feeds by acti\'ely searching for prey. Generally Ja- 
maican anoles maintain a perch until they see a food item, and 
then go after it. Perhaps the feeding habits of reconditus include 
both searching and waiting. More study is necessary to deter- 
mine whether this is true. 

SLEEPING 

It seems likely that reconditus, just as A. garmani and se\'- 
eral other Anolis species, sleeps in the foliage of the crown of 
trees. The first lizard that I caught at Hardwar Gap was seen 
at 7:50 a.m. walking down the trunk of a 45-foot tree. The 
sun was up, of course, but had not yet reached the incredibly 
steep, densely forested northwest mountainside of Steep Bu'^h. 
It is possible that this 96-mm male was coming down from his 
sleeping perch to a lower daytime perch. On two later occa- 
sions, both at sunset, I put individual lizards "to bed" to see 
where they went. 

The first time, I watched a large male {ca. 90 mm) for more 
than three hours. He maintained until sundown his perch place 
12 feet above ground with only occasional sideways movements 
from the downward-facing perching position. Then, in a period 



20 BREVIORA No. 404 

of 12 minutes, as darkness was rapidly falling, he went through 
a flurry of activity and elaborate displays (see display). This 
culminated in his disappearance into the higher foliage. 

Two weeks later, I watched a 62-mm female at sundown. 
She suddenly changed to a crosswise perch from the unchar- 
acteristic upward-facing perch she had maintained all day. The 
perch was four feet above the ground on a small tree. Ten 
minutes later, she jumped to a nearby straight sapling and looked 
around. Four minutes later she ran up a few inches and looked 
around. Then she disappeared up the small tree. It seems rea- 
sonable that a female of this size, indeed the smallest I caught, 
would sleep in a sapling such as this, a great deal smaller than 
the large tree, where, as noted above, it is likely that a big male 
sleeps. 

DISCUSSION 

Anolis recondilus is unique and at the same time unspecial- 
ized. The lizard is apparently isolated from congeneric compe- 
tition. This may be one of the keys to understanding the mor- 
pholog)', ecology, and behavior of the animal. E. E. Williams 
(Williams et al, 1970) makes such a point in his discussion of 
the solitary Venezuelan lizard Anolis jacare. 

Recondilus does not occupy any of the typically anoline struc- 
tural habitats defined by Rand and Williams (1969). Of the 
six categories they set up, it seems most like a "trunk anole." 
But it is unlike that category in that it often goes farther up the 
tree and occasionally down to the ground. Of course, being 
isolated as it is, there is no reason for it to be restricted to any 
one structural habitat, as is necessarily the case when several 
species live syntopically. In addition, the perch that recondilus 
most frequently takes seems a very reasonable adaptation for a 
solitary species. From 10 or 12 feet up on a trunk, the animal 
gets about as good a view of the surrounding forest as is possible 
and can reach either the ground or crown with equal facility. 

In the Lesser Antilles, islands with only one Anolis species 
are common. However, on the larger islands of the Greater 
Antilles, each of which supports several to many species, there 
is much overlap of congeneric range. Recondilus in Jamaica 
is in a situation similar to that of A. jacare on the continent 
of South America. It is on a "mainland island, just like 
jacare," at its Blue Mountain retreat. A detailed comparative 
study of A. recondilus in its montane mist forest and A. jacare- 



1973 ANOLIS RECONDITUS 21 

in its "Premontane Humid Forest" ^Williams cl al, 1970) might 
yield some interesting insights into the adaptations of isolated 
species. 

Thomas Schoener (1969 and unpublished data) shows that 
the male snout— vent length (mean of the largest third of his 
sample) for lizards inhabiting islands of the Lesser Antilles which 
ha\e only one species of Anolis ranges between 60 and 94 mm. 
Reconditus, at 97 mm, would fall only a bit beyond the high 
end of this range. From the few indi\iduals of jacare that have 
been measured, male size seems to be about 75 mm. 

Reconditus is possibly the only Jamaican anole which actively 
searches as well as motionlessly perches in attempts to obtain 
food. Other than A. sagrei, it is unique among anoles in its 
ability to function normally in the rain. It does not fit into a 
Rand and W^illiams' perch category, but instead ranges over se- 
veral of the structural habitats that they define. No other anole, 
except A. valencienni (Rand, 1967), is known to crawl into 
large holes in the tree trunks on which it perches. 

These features of the beha\ior of reconditus suggest perhaps 
that the animal is adapted to its habitat in a more general way 
than some of its aparently more specialized congeners. Recon- 
ditus appears less restricted in foraging methods, in ability to 
cope with the weather, and in perch and refuge space than al- 
most all other anoles. Further study of these and additional 
aspects of the lizard's behavior and ecology might reveal whether 
it is indeed less specialized than its congeners. 

Possibly because reconditus is isolated from congeneric com- 
petition it is free to fully exploit its habitat. If this is the case, 
it may be that within its restricted montane mist forest habitat, 
it can, living alone, use more of the resources of the environ- 
ment through generalized rather than specialized adaptations. 

AN EXPLANATION OF THE MARKING SYSTEM 

I marked lizards at the two main study areas. Steep Bush 
and Fairy Glade (see habitat and range and map for loca- 
tion of these places). In each area I used a numerical mark- 
ing system for permanently identifying individual lizards. A 
single toe or, in the case of the large toes, just the terminal joint, 
was clipped off the lizard's foot. 

The toes were numbered as described below : 

Starting with the lizard on its \-entral side with its dorsal 

side facing you, the toes on the front left leg (your left) were 



22 BREvioRA No. 404 

numbered consecutively 1 to 5 from smallest to largest. The 
toes on the right front leg (your right) were labelled 6 to 10 
from smallest to largest. The toes on the left hind leg were 
labelled 11 to 15 from smallest to largest. Finally the toes 
on the right hind leg were labelled 1 6 to 20 in a similar man- 
ner. 

Should anyone catch any of these marked animals at Hardwar 
Gap, I would greatly appreciate his sending me size data 
(lower jaw, snout-vent, and tail lengths) and any other inter- 
esting features about the lizards. I can be reached care of Pro- 
fessor E. E. Williams, Museum of Comparative Zoology, Cam- 
bridge, Massachusetts. 

ACKNOWLEDGEMENTS 

I want to thank especially Ernest E. Williams for introducing 
me to the study of anoline ecology and behavior. This project 
would not have been possible without his constant advice and 
support throughout all stages of the work. 

Robert Trivers' contributions were invaluable. Not only did 
he help with details such as initial identification of reconditus 
in the field but also he generously contributed his ideas and his 
field techniques. Thomas Jenssen gave me much help and ad- 
vice while I was in the field. Preston Webster and A. Stanley 
Rand both provided extensive and valuable criticism on earlier 
drafts of this paper. 

I want to thank C. B. Lewis, Director of the Institute of 
Jamaica, for allowing me to use the Institute's field station at 
Green Hills for the duration of my study. 

This study was partly supported by NSF grant B 019801x 
to Ernest E. Williams. 

REFERENCES 

AsPREY, G. F., AND R. G. RoBBiNS. 1953. The vegetation of Ja- 
maica. Ecol. Monogr., 23: 359-412. 

Lazelx., J, D., Jr. 1966. Studies on Anolis reconditiLS Underwood 
and Williams. Bull. Inst. Jamaica, Sci. Ser., 18 (pt. 1): 1-15. 

Rand, A. 1967. Communal egg laying in anoline lizards. Herpeto- 

logica, 23(3) : 227-230. 
, AND E. E. Williams. 1969. The anoles of La Palma: 

aspects of their ecological relationships. Breviora, No. 327: 

1-19. 



1973 ANOLIS RECONDITUS 23 

SCHOENER, T. W. 1969, Size patterns in West Indian Ayiolis liz- 
ards: I. Size and species diversity. Syst. Zool., 18: 386-401. 

Underwood, G., and E. E. Williams. 1959. The anoline lizards of 
Jamaica. Bull. Inst. Jamaica, Sci. Ser., 9: 1-48. 

Williams, E. E., O. A. Reig, P. Kiblisky, and C. Riv^ero-Blanco. 
1970. Anolis jacare Boulenger, a "solitary" anole of the Andes 
of Venezuela. Breviora, No. 353 : 1-15. 



" «>• c. 



Ob- 






B R 



^ 



J 



V I 



MAR 1 8 )9g5 




iTr" 



iiseuiii of CoBiparal" 



TJ^ 



d" 



y 



us ISSN 0006-9698 



Cambridge, Mass. September 20, 1973 Number 405 

A FOSSIL TRTONYGHID TURTLE 
FROM SOUTH AMERICA 

Roger Conant Wood^ 

and 

Bryan Patterson 



Abstract. A hyopkistron of u larg-e trioiiychid turtle from the 
middle Pliocene (Huayquerian) Urumaco Formation of northern 
Venezuela constitutes the first indisputable record of a member of 
this family from South America. The reported occurrence of a tri- 
onychid from the late Cretaceous or Paleocene of Patagonia {Trionyx 
argentina Ameghino 1899, iiomen dubiuni) cannot be substantiated. 
The species represented by the Venezuelan fragment was presumably 
a waif immigrant from Central America that was unsuccessful, very 
possibly owing to competitive exclusion, in establishing an enduring 
foothold in South America. 

Living trionychid turtles are known from Africa, Asia, the 
Indo- Australian archipelao;o and North Americ;i. Paleontologi- 
cal e\'idcnce indicates that this group formerly had an c\'en more 
extensixe distribution. Fossil forms are known from parts of 
Africa, Asia, and North America where they no longer occur, 
and remains are found in European Tertiary deposits as well. 
Until now, with one dubious exception (discussed below), no 
members of the family, living or fossil, have ever been encountered 
in South America. Discovery of an uncjuestionable trionyc;hid 
fragment from late Tertiary sediments in Venezuela during the 
summer of 1972 is therefore of considerable interest. 

The specimen (Museo de Ciencias Naturales, Caracas, no. 
238) was reco\ered from the upper part of the lower member 
of the LIrumaco Formation about twcKe kilometers WS^V of 
Urumaco and .75 kilometer north of Kilometer 153 on the oil 
pipe line running from Punta Gorda to the Paraguana Penin- 

'Stockton State College, Pomona, New Jersey. 



2 BREvioRA No. 405 

sula. It was found on the surface, together with parts of a 
pelomedusid that was in the process of weathering out. The 
Urumaco Formation (Gonzales de Juana ct al., 1970: 612—613) 
is now believed to be of Huayquerian age (Pascual and de 
Gamero, 1969; Wood and de Gamero, 1971). Outcrops are 
restricted to a relati\'ely small area in the northern part of the 
state of Falcon. The vertebrate fauna, for the most part not )et 
described (see, in addition to the publications cited above, Royo 
y Gomez, 1960, and Sill, 1970), consists of numerous aquatic 
reptiles (some half dozen species of both crocodilians and tur- 
tles), a testudinine, fish, and \'arious kinds of mammals. The 
depositional environment was in the main a near-shore marine 
one, with fluctuations involving a variety of paralic and near- 
shore continental conditions. 

The find consists of a large, well-preserved hyoplastron (figure 
1 ) . Its sculptured external surface, covered with anastomosing 
ridges, is characteristic of trion)chids. These rugosities, coupled 
with the distinctive outline of the bone, leave no doubt regarding 
the familial reference. Dimensions are as follows: width along 
the hyo-hypoplastral suture = 22.2 centimeters; length at nar- 
rowest part = 7.3 centimeters; greatest dorsoventral thickness 
along hyo-hypoplastral suture '= 2.8 centimeters. From the con- 
dition of the sutural surface, it is clear that the hyo- and hypo- 
plastra were distinct bones rather than fused as is the case in 
various genera of the family. The h)oplastron appears to be 
unusually thick for its size, the thickness/ width ratio being 0.126. 
Measurements of shell thickness are scarcely ever given in the 
literature. A few T/\V ratios can roughly be estimated from 
the figures and sporadic measurements given by Hay ( 1 906 ) 
for North American fossil trionychids; these range from 0.083 
to 0.019. The hyoplastra of those living forms available to us 
are as follows regarding this ratio: Lissemys punctata, 0.087; 
Chitra indica, 0.081 ; Trionyx hurutn, 0.067; T. sinensis, 0.079; 
T. cartilagincus, 0.090; T. spinifcr (6), 0.107-0.064; T. ferox 
(4), 0.081-0.070; T. muticus (2), 0.077-0.049. This, of course, 
is a wholly inadequate sample, but as far as it goes it confirms 
our impression that the Urumaco hyoplastron is unusually thick. 
Further, the sample suggests that thickness decreases, relatively, 
with age; in all species of which we have more than one speci- 
men the smallest indixidual has the highest ratio and the largest 
the lowest. The thickness of the Urumaco specimen is not, we 
accordingly suspect, due to size. 

This hyoplastron is sufficiently distinctive for reasonable as- 



1973 



SOUTH AMERICAN TRIOXVCHII) TIRTI.E 





-*^>'.-':'ic 







Figure 1. Trionychidae, gen. et sp. indet. Ventral and posterior 
views of hyoplastron. Scale in centimeters. 



4 BREVIORA No. 405 

surance that we are dealing with a representative of the group 
sometimes recognized as the subfamily Trionychinae (group I 
of Lo\'eridge and Williams, 1957: 414), but quite inadequate 
for further taxonomic refinement. Assessment of precise relation- 
ships within the family must await discovery of more complete 
material. The geographically nearest li\ing species of the family 
are Trionyx spinifer, whose range extends from the United States 
down into northeastern Mexico, and T. ater, which is confined 
to the Cuatro Cienegas Basin of central Coahuila, Mexico 
(Webb, 1962). Neither attains great size; in fact, the Urumaco 
form was clearly larger than any of the living American species. 
Some fossil trionychids {e.g., the late Paleocene Paleotrionyx 
quinni Schmidt, 1945) are known from North America, how- 
ever, whose size is comparable to that of the Venezuelan speci- 
men. 

The earlier record of a South American trionychid must now 
be considered. Ameghino (1899: 10) published a very sum- 
mary "diagnosis" of Trionyx argentina from the "Cretaceous of 
Patagonia (Guaranitic Formation)." In his only subsequent 
mention of it, he stated that Tryonix [sic] argentina had come 
from the "Sehuenian" (1900: 216). No locality was ever given. 
His "Guaranitic Formation" included marine and continental 
strata ranging from Cretaceous to early Oligocene ( Deseadan ) , 
and his "Sehuenian" formed part of it. On the basis of his last 
statement concerning the age it could be assimied that 7". argen- 
tina came from the type area of the "Sehuenian" (now the 
Mata iVmarilla Fonnation, Coniacian; Leanza, 1972: 695, 701) 
on the Rio Shehuen (or Sehuen or Chalia) in the west-central 
part of the province of Santa Cruz, and Feruglio (1949a: 257) 
belie\ed that it had. This is not certain, however. 

In 1898 (p. 121) Ameghino included in his "Sehuenian," or 
"Piso Sehuense," deposits occurring in regions other than the 
type area, among them the Rio Chico del Chubut. It would 
seem possible that T. argentina could have come from this gen- 
eral resrion, and there is some indication that it mav ha\e. In 
the introduction to his paper of 1899 Ameghino stated that all 
species proposed in it had been collected by his brother Carlos 
since the manuscript of his paper of 1898 (to which 1899 was 
a supplement) had been dehvered to the printer, which was 
done at the end of Julv, 1897 (letter from F. Ameghino to D. 
de la Fuente, August 7,' 1897; in Torcelh, 1935: 688). By that 
time Carlos Ameghino's collecting activities had centered on 
northern Santa Cruz and Chubut. Reporting on the results of 



1973 SOUTH AMERICAN TRIOXYCHID TURTLE 5 

his expedition of 1898-99 to the Golfo dc San Jorge, he \vrote 
(letter to Florentine, February 15, 1899; m Torcelli, 1935: 106) 
that he had found "more or less in the center of the gulf . . . 
characteristic fossils (sharks, chelonians, crocodiles, etc., the 
same as those of Sehuen and Lake Argentino)." These che- 
lonian remains could have included the material on which T. 
argentina \va.s based. Florentino's statement (1899) that a 
trionychid had just been found ("acaba de encontrarse" ) could 
be so interpreted. At any rate, years later, Staesche (1929) 
described turtle fragments collected, together with crocodilian 
remains, by von Huene near the center of the gulf, "mainly at 
Cabo Peligro," and se\'eral of these ans\ver to Ameghino's "diag- 
nosis" of T. argentina in one respect, \-ermicular sculpturing. 
Staesche believed the age of these specimens to be Late Creta- 
ceous, but the bed from which they came was the banco negro 
inferior at the base of the Rio Chico Formation or perhaps at 
the top of the Salamanca Formation Feruglio, 1949b: '11. 
fig. 87), both of which are now considered to be Paleocene in 
age. Turtle remains, according to Feruglio, commonly occur in 
this bed or beds in the region of the gulf. Against the possibility 
that T. argentina may have been reco\ered from this area, how- 
ever, is the fact that by 1900 (p. 119) Ameghino had begim to 
make a distinction between the "Sehuenian'" and the beds in 
Chubut containing Ostrea pyrotherionun (now the Salamanca 
Formation). As in the case of Niolamia argentina (Simpson, 
1938), also described in 1899, no conclusion can now be reached 
as to the precise age and locaHty of T. argentina. although we 
regard Paleocene and the gulf as being the more likely alternative. 

The affinities are even more uncertain. Ameghino"s "diag- 
nosis" reads, in free translation, "Trionxx argentina n. sp., size 
small, surface of shell with \-ermicular sculpturing, ossification 
incomplete, without scute sulci. "^ No information was given as 
to the type material. This is wholly inadequate to define a spe- 
cies of Trionyx or e\en to demonstrate the familial assignment. 
Other turtles have what could be termed \'ermicular sculpturing ; 
\vhat is meant by "incomplete ossification"" is not clear; and the 
absence of sulci is not a certain indication that scutes were lack- 
ing. The specimen or specimens on which the taxon was based 



"'. . . de talla pequeiia y superficie de la coraza con esculptura 
vermicular pero de osificacion incompleta y sin surcos externos que 
indiquen la presencia de escudos ccrneos." 



6 BREVIORA No. 405 

would appear to have been lost; Wood failed to encounter them 
in the course of an examination of the turtle remains in the 
Ameghino collection. Consequently, we regard "Trionyx argen- 
tina" as a no?nen dubium and we feel that the existence of a late 
Cretaceous or Paleocene South American trionychid has not been 
demonstrated. 

The specimen described here is thus the only definite record 
of a trionychid for the continent. Representati\es of the family 
did obtain a foothold there, but for how long and in what man- 
ner? During 1970 Wood spent four months in South America 
examining collections of fossil turtles ranging in age from Cre- 
taceous to Pleistocene and nowhere saw so much as a fragment 
that could be identified as a trionychid. The extensive collec- 
tions made in the late Ohffocene and later Miocene of Colombia 
by University of California parties include turtle remains JDUt no 
representative of this family. Among the hundreds of turtle 
specimens in all stages of completeness seen by us in the Uru- 
maco Formation only this one trionychid fragment was found. ^ 
On the face of it, therefore, it would seem possible that the in- 
vasion \vas a Huayquerian event and only briefly successful in 
the geological sense of the word. Why this should ha\e been 
so is puzzling, for the great ri\er systems of South America 
encompass habitats surely suitable for members of the family. 
Simpson (1943: 423) has suggested that trionychids and che- 
lids may be ecologically incompatible. Chelids are clearly an 
old South American group in the sense of Dunn (1931). \Vood 
has in hand undescribed material from the early Eocene Casa- 
mayor of Argentina." Undescribed chelids have recently been 
discovered at Tremembe, in Brazil, now known to be a deposit 
of early Oligocene, Deseadan, age (Paula Couto and Mczzalira, 
1971). Specimens occur in the Oligocene and in the Miocene 
of Colombia (Stirton, 1953: 614) and have long been known 
from Parana. We have good material of Chclus from the Uru- 
maco Formation, and additional undescribed material of this 
same genus has been found in Pliocene deposits along the upper 
reaches of the Jurua River in the territory of Acre, Brazil. The 
Chelidae was not the only family that could have provided 



'It was collected midway in the season and a very close watch for 
others was maintained thereafter. 

^Ameg-hino "diag-nosed" three species of Platemys from his "Guar- 
anitico" in 1899, but these records are vominn dnbia and the speci- 
mens evidently lost. 



1973 SOUTH AMERICAN TRIONYCHID TURTLE 7 

competition to inxading aquatic turtles.^ Pelomedusids, also Old 
South American forms, are known in the continent from the late 
Cretaceous on and are \ery numerous in the Urumaco deposits. 
The demonstrated failure of trionychids to maintain themselves 
in South America could be interpreted in terms of competitive 
exclusion. Chelids and pelomedusids ha\'e evolved together in 
South America throughout the Cenozoic, and in the course of 
this time no doubt parceled out between them a wide variety of 
habitats. That they would have completely saturated the con- 
tinent is unlikely, but such diversification would have lengthened 
the odds against an invader. The species represented by our 
Urumaco trionychid may well have had the misfortune to en- 
counter upon arrival a comparably adapted resident and have 
failed to compete successfully with it. It was, indeed, similar in 
size to the commonest Urumaco pelomedusid, Podocnertiis ven- 
ezuelensis Wood and de Gamero. 

Such mammalian remains as occur in the Urumaco Forma- 
tion are all of South American type, a finding in accord with 
Argentinian evidence suggesting a Montehermosan date for the 
joining of the American continents. On this basis, then, the 
\^enezuelan trionychid would appear to have been a waif immi- 
grant carried by ocean currents. Members of the family may 
well be rather prone to such accidents. There is some evidence 
that several of them enter brackish or even salt water (summary 
and references in Neill, 1958: 26-27). These would be par- 
ticularly liable to hazards of this kind, but tolerance for saline 
waters need not necessarily be prerequisite. The Afiican Trionyx 
tri unguis also occurs in rivers and swamps along the eastern 
Mediterranean littoral. Flower (1933: 754) has suggested that 
this extension results from individuals being swept to sea by Nile 
floods and carried eastward by the prevailing current. The 
ancestor or ancestors of the Venezuelan trionychid presumably 
came from Central America, the nearest land area, and arri\ed 
in South America at approximately the same time as did the 
extinct Cyonasua group of the Procyonidae. No trionychid now 

^At the present time there is one area in which chelids and tri- 
onychids come together. This is southern New Guinea where the 
wide-ranging- trionychid PelocJichjs bihroni has established itself — 
i-ather recently in Darlington's opinion (1957: 210). There it is in 
contact wdth several chelids. So far as we are aware no infonnation 
is available on possible interactions, but direct competition would 
seem unlikely. P. bibrojil is a large species, much larger than any 
New Guinean chelid. 



8 BREVIORA No. 405 

lives in Central America — it is a decidedly odd fact of distribu- 
tion that all living North American species of Trionyx, in con- 
trast to a number of their Old World congeners, are extratropi- 
cal. None has been recorded fossil there, but in view of our 
abysmal ignorance of the Tertiary vertebrates of the region no 
weight can at present be given to their absence.' 

This note records a momentary success in range extension 
resulting from waif dispersal. Such transitory events must surely 
have been commoner than those more enduringly successful, but 
in the nature of things the chances are against their entry into 
the fossil record. They could have played a part in furthering 
distribution. In the case of organisms such as freshwater turtles 
each new toehold gained, for however brief a time, in a suitable 
environment would have provided a potential base for further 
waif dispersal. The distribution of pleurodirans, for example, 
may have been to some degree forwarded in this way. 

ACKNOWLEDGMENTS 

Our field work in Venezuela was made possible by National 
Science Foundation Grant no. GB-32489x to Patterson, and by 
the generous cooperation of the Escuela de Geologia, Universi- 
dad Central de Venezuela, and the Ministerio de Minas y 
Hidrocarburos. We are indebted to Profesora Maria Lourdes 
Diaz de Gamero, Dr. Clemente Gonzales de Juana, Dr. A. 
Bellizzia G., Professor John Gibson-Smith, Dr. Edgardo Mon- 
dolfi, Sr. Bias Gonzalez, and our companions in the field : Messrs. 

'Webb (1962: 584) has credited Mullerreid (1943) with recording 
a fossil trionychid of unknown provenance from Mexico. Actually, 
that author imentioned seeing the internal mold of an unidentified 
turtle. The owner of the specimen, which has probably been lost, 
stated that he had found it on the Mesa de Capolla, a short distance 
SSW of Tuxtla Gutierrez, Chiapas. Cretaceous and early Tertiary 
sediments are present there, and Mullein-eid believed that the mold 
had come from the latter. With the exception of the Oligocene or 
Miocene testudinine Geochelone costarricensis (Segura) (Auffenberg, 
1971), this is the only published record of a Tertiary turtle from 
Central America. (In addition, we know of undescribed material 
from two Tertiary localities in Panama; no trionychids are included 
among the specimens.) Aguilera (1907: 241) listed Tryonix [sic] 
sp. as occurring- with marine invertebrates in Turonian deposits at 
Peyotes, Coahuila. This is the only North American fossil record of 
the family south of the United States. It requires confirmation, and 
is in any event within the present range of T. spinifer emoryi. 



1973 SOUTH AMERICAN TRIONYCHID TURTLE 9 

Arnold D. Lewis (the finder of the specimen here discussed), 
Daniel C. Fisher, Robert W. Repenning, and Michael F. Stan- 
ford. The participation of Messrs. Fisher, Repenning, and 
Stanford was due, respectively, to the generous provision of a 
summer scholarship from the Department of Geological Sciences, 
Har\-ard Uni\ersity, and of grants from the Shell Oil Company 
and the Creole Petroleum Corporation. W^ood's work in South 
America during 1970 was supported by the National Geographic 
Society. 

REFERENCES CITED 

Aguilera, J. G. 1907. Apergu sur la geologie du Mexique pour 

servir d'explication a la carte geolog'ique de I'Anierique du Nord. 

Compte Rendu lOieme Sess. Cong. Geol. Internat., Mexico: 227- 

248. 
Ameghino, F. 1898. Sinopsis geologico-paleontologica de la Re- 

publica Argentina. Segundo Censo de la Republica Argentina, 

1: 111-255. 
. 1899. Sinopsis geologico-paleontologica. Suplemento 

(adiciones y correcciones). La Plata: Imprenta la Libertad. 13 

pp. 

1900. L'age des formations sedimentaires de Pata- 



gonie. An. Soc. Cient. Argentina, 50: 109-130, 145-165, 209-229 

(part). 
AUFFENBERG, W. A. 1971. A new fossil tortoise, with remarks on 

the origin of South American testudinines. Copeia, 1971: 106-117. 
Darlington, P. J., Jr. 1957. Zoogeography: the Geographical 

Distribution of Animals. New York: John Wiley, i-xiii + 675 

pp. 
Dunn, E. R. 1931. The herpetological fauna of the Americas. 

Copeia, 1931: 106-119. 
Feruglio, E. 1949a. Descripcion geologica de la Patagonia. Tomo 

I. Buenos Aires: Coni. i-xv + 334 pp. 

. 1949b. Descripcion geologica de la Patagonia. Tomo 

II. Buenos Aires: Coni. 349 pp. 

Flower, S. S. 1933. Notes on the Recent reptiles and amphibians 
of Eg>T)t, with a list of the species recorded from that kingdom. 
Proc. Zool. Soc. London, 1933: 735-851. 

Gonzalez de Juana, C, A. Salvador, R. M. Stainforth, G. A. Young, 
F. DE RiVERO, C. Martin Bellizzia, C. Petzall (eds). 1970. 
Lexico estratigrafico de Venezuela (segunda edicion). Rep. 
Venezuela, Minist. Min. Hydroc, Dir. Geol., Bol. Geol. Pub. Esp. 
No. 4: 1-757. 

Hay, 0. P. 1906. The fossil turtles of North America. Publ. Carne- 
gie Inst. Washington, No, 75: 1-568. 



10 BREVIORA No. 405 

Leanza, a. F. 1972. Andes patag-onicos australes. In. A. F. Leanza, 
Ed., Geologia Reg-ional Argentina. Cordoba: Academia nacional 
de Ciencias, pp. 689-706. 

LOVERIDGE, A., AND E. E. WiLLlAMS. 1957. Revision of the African 
tortoises and turtles of the suborder Cryptodira. Bull. Mus. 
Comp. Zool., 115: 163-557. 

MULLERREiD, F. K. G. 1943. Fosiles raros de Mexico III — Una tor- 
tuga fosil del estado de Chiapas. An. Inst. Biol. Univ. Nac. 
Mexico, 14: 623-624. 

Neill, W. T. 1958. The occurrence of amphibians and reptiles in 
saltwater areas, and a bibliogi-aphy. Bull. Marine Sci. Gulf 
Caribbean, 8: 1-97. 

Pascual, R., and M. L. Diaz de Gamero. 1969. Sobre la presencia 
del genero Eumegamys (Rodentia, Caviomorpha) en la foiTn- 
acion Urumaco del Estado Falcon (Venezuela). Su significacion 
cronologica. Bol. Infonnativo, Assoc. Venezolana Geol. Min. Pet., 
12: 369-387. 

Paula Couto, C. de, and S. Mezzalira. 1971. Nova conceituagao 
geocronologica de Tremembe, Estado de Sao Paulo, Brasil. An. 
Acad. Bras. Cienc, 43, Suplemento : 473-488. 

ROYO Y Gomez, J. 1960. Los vei'tebrados de la formacion Urumaco, 
Estado Falcon. Mem. Ill Cong. Geol. Venezolano, 2: 506-510. 

Schmidt, K. P. 1945. A new turtle from the Paleocene of Colorado. 
Fieldiana: Geol., 10: 1-4. 

Sill, W. D. 1970. Nota preliminar sobre un nuevo gavial del Plio- 
cene de Venezuela y ima discusion de los gaviales sudamericanos. 
Ameghiniana, 7: 151-159. 

Simpson, G. G. 1938. Crossochelys, Eocene horned turtle from 
Patagonia. Bull. Amer. Mus. Nat. Hist., 74: 221-254. 

— ■ . 1943. Turtles and the origin of the fauna of Latin 

America. Amer. Jour. Sci., 241: 413-429. 

Staesche, K. 1929. Schildkrotenreste aus der oberen Kreide pata- 
goniens. Palaontographica, 72: 103-123. 

Stirton, R. a. 1953. Vertebrate paleontology and continental 
stratigraphy in Colombia. Bull. Geol. Soc. America, 64: 603-622. 

TORCELLI, A. J. (ed.) 1935. Obras Completas y Correspondencia 
Cientifica de Floi'entino Ameghino. Volumen XXI. Corres- 
pondencia Cientifica. La Plata: Taller de Impresiones Oficiales. 
935 pp. 

Webb. R. G. 1962. North American Recent soft-shelled turtles 
(family Trionychidae). Univ. Kansas Publ. Mus. Nat. Hist., 
13: 429-611. 

Wood, R. C, and M. L. Diaz de Gamero. 1971. Podocnemis vene- 
zitelensis, a new pelomedusid (Testudines, Pleurodira) from the 
Pliocene of Venezuela and a review of the history of Podocnemis 
in South America. Breviora, Mus. Comp. Zool. No. 376: 1-23. 



vlpO^ 



-,«■•' ■ 



^^^I'S. CO MP. ZOOL 
LfPRARY 



B R 



A 



V I WW A 



iiseiim of Comparative Zoology 



us ISSN 000(1- '.)(i'.)S 



Cambridge, Mass. September 20, 1973 Number 406 



TWO NEW LYGOSOXIINE SKINKS FROM 

NEW GUINEA WITH COMMENTS 

ON THE LOSS OF THE EXTERNAL EAR 

IN LYGOSOMINES AND OBSERVATIONS 

ON PREVIOUSLY DESCRIBED SPECIES 

Allen E. Greer, Jr. 

AiiSTKAcr. Two new species of the jasciatus species group of Spheno- 
murplnis are desciibcd from New Cuiiica: S. anotus is unique among its 
relati\es in lacking an external ear opening and i. inicrutympanus can be 
disiinguislietl 1)\ tlie greatly reduced size of its tvnipanuni. 

The earless l)gosomines are reviewed and it is noted that, while the 
loss of the external ear opening is apparently a prerequisite for a burrowing 
way of life, ii is inicertain whether the loss of the external ear in burrowcrs 
is a piimary adaptation to burrowing or whether it is a preadaptation 
inherited fioui a nonburrowing ancestor. 

Certain aspects of the morphology of the previously described but poorly 
known sjjecies Splicuoinorplius forbcsi, S. olii^olcpis and i'. schuUzei are 
discussed antl photographs of type specimens are provided. 

In 1964 members of the Se\enth Archbold Expedition col- 
lected two small scincid lizards on the Huon Peninsula which 
differed from all other Icnown skinks in New Guinea in having 
a scaly auricular depression instead of the more external ear 
opening. These were thus the first "'earless" skinks to come out 
of New Guinea. The specimens were sent to Dr. Richard 
Zweifel at the American Museum of Natural History, but he 
was unable to identify them and he put them aside in the iiope 
that more specimens would be forthcoming. 

Five years later two more very similar skinks were collected 
by Angus F. Hutton at Garaina and gi\en to Dr. Zweifel during 
his 1 9(59 trip to New Guinea. Dr. Zweifel recognized the great 
similarity between these two specimens and the Huon Peninsula 
specimens but, still being unable to identify them, he kindly 
turned them o\'er to me for further study. 



2 BREVIORA No. 406 

On close examination the two specimens from the Huon 
Peninsuhi pro\e to be members of the fasciatus species group 
of Sphenomorphus (^ Greer and Parker, 1967), but they difTer 
strikingly from all known members of this species group in being 
"earless." The two Garaina specimens are also clearly members 
of the fasciatus species group and are indeed similar to the Huon 
Peninsula specimens. They differ from the Huon Peninsula 
specimens, however, in having a relatively small, but nonethe- 
less distinct, tympanum instead of a scaly auricular depression, 
and they differ from all pre\iously described members of the 
fasciatus species group in the extreme reduction in the size of 
the tympanum. The Huon Peninsula and Garaina specimens 
are thus distinct enough from each other and from their closest 
relatives in the fasciatus species group to be described as new. 

Description of Two New Species 

Sphenomorphus anotus new species 

Figure 1 

Hololype. American Museum of Natural History 95880; an 
adult collected on 5 May 1964 by Hobart M. Van Deuscn and 
Stanley O. Grierson in the Morobe District of the Territory 
of New Guinea at MASBA GREEK (Figure 4) at an elevation 
of approximately 2000 feet. For an account of this locality see 
Van Deusen (1966). 

Paratype. AMNH 95881; a badly mangled young juvenile 
collected by Van Deusen and Grierson on 7 May 1964 at the 
same locality as the holotype. 

Diagnosis. This species is a typical representative of the 
fasciatus species group of Sphenomorphus^ (Greer and Parker, 
1967) except that it has a scaly auricular depression instead of 
the more usual external ear opening. In other words, it is the 
only "earless" member of the fasciatus species group known to 
date. 

Etymology. The species name anotus calls attention to the 
absence of an external ear {an — without and otus — ear). 

'The diagnostic features of tlie fasciatus species group of Splieiioniorphus 
are as follows: digits and limbs usually well developed but the limbs 
generally not overlapping when adpressed to the body; frontal in contact 
with the two anteriormost supraoculars; generally four supraoculars; a 
single anterior loreal; no supranasals; usually a series of two or more 
paired nuchal scales; generally 36 or fewer scales around midbody, the 
scales of the paravetebral rows being larger than the scales of the more 
lateral rows; generally a postorbital bone that is usually long and thin. 



1973 



EARLESS SKINKS 





B 



Figure 1. Dorsal (A) and lateral (B) view of the head of the holotype 
of Sphenomorphus anotus (AMNH 95880) from Masba Creek, Huon 
Peninsula, New Guinea. 



Description of the holotype. A small attenuate skink mea- 
suring 48 mm in snout-\ent length with a complete tail mea- 
suring 68 mm in length; head bluntly conical; hmbs short, 
pentadactyl, widely separated when pressed against the body; 
uniformh' brown above and light yellowish brown below (in 
preser\ati\e ) . 

Rostral about as deep as wide and projecting well onto dorsal 
surface of snout; nasal large with external naris situated well 
forward and ventral in nasal scale; no supranasals; frontonasal 
wider than long and forming a short suture with rostral and a 
slightly wider suture with frontal; prefrontals separated; single 
anterior and posterior loreals; frontal slightly longer than wide 
and in contact with two anteriormost of four supraoculars; lower 
eyelid movable and scaly; frontoparietals and interparietal dis- 
tinct, approximately subequal in size, and measured together 
along midline, about equal in length to frontal measured along 
midline; parietals meet behind interparietal; two nuchal scales on 
left side and four on right; six supralabials, fourth situated most 
directly below eye ( Fig. 1 ) . 



4 BREVIORA No. 406 

External ear opening lacking and in its place an auricular 
depression completely lined with small scales; body scales smooth 
and in 26 longitudinal rows at midbody; scales of paravertebral 
scale rows slightly wider than other dorsal scales at midbody and 
numbering 71-72 in distance from parietals to midpoint of in- 
sertion of hind legs; medial pair of preanal scales enlarged; 
medial row of scales on underside of tail only slightly larger 
than adjacent lateral rows; 8-9 obtusely keeled lamellae beneath 
fourth toe; fourth toe covered dorsally by a single row of scales 
on distal third of length, by two rows of scales over medial third, 
and by three rows over proximal third (Group III of Bron- 
gersma, 1 942 ) . 

Color of the holotype. No color notes were made on the speci- 
men in life. In preservative, however, the dorsum is a uniform 
light chocolate brown while the venter is a light yellowish brown 
anterior to the vent and a slightly darker brown posterior to 
the vent. There is no sharp transition line between the dark 
dorsal color and the light ventral color. There is only the 
slightest trace of scattered dark spotting on the throat. 

Variation in the paratype. The single paratype of Spheno- 
morphus anotus is a small (snout-vent length = 24 mm), badly 
mangled specimen obviously of very young age. In the char- 
acters that can be evaluated it differs but little from the holo- 
type: there are 26 midbody scale rows, the fourth supraocular 
lies most directly beneath the eye, there are 3^ nuchals, and 
most importantly, there is a scale-lined auricular depression 
instead of an external ear opening. In color the paratype is 
similar to the holotype but it lacks the yellowish wash to the 
venter. 

Distribution. Sphenomorphus anotus is known only from the 
type locality on the Huon Peninsula of New Guinea (Fig. 4). 

Habitat. The Masba Creek locality where the two types were 
caught is in a "stretch of unbroken rain forest" (Van Deusen, 
1966) and both animals were taken as the litter was being 
scraped level for the tents and work flys. Thus it would seem 
that, like other members of its species group, S. anotus is a 
cryptic burrower in the litter. 

Relationships. S. anotus appears to be very closly related to 
the following species, but a discussion of the relationships of both 
forms is deferred to the end of that species' description. 



1973 



EARLESS SKINKS 





Figure 2. Dorsal (A) and lateral (B) view of the head of the para- 
type of Sphenomorphus micro tympanus (MCZ 132767) from Garaina, New 
Guinea. 



Sphenomorphus microtympanus new species 
Figures 2, 3, and 5 (top) 

Holotype. AMNH 104076; an adult collected on 7 July 
1969 by Angus F. Hutton in the Morobe District of the Terri- 
tor\' of New Guinea at GARAINA (Fig. 4) at an elevation of 
approximately 2300 feet. 

Paratype. MCZ 132767; same data as the holotype. 

Diagnosis. S. microtympanus is a member of the jasciatus 
species group of Sphenomorphus and in that it lacks an ecto- 
pterygoid process to the palatine running along the outer edge 
of the palatal ramus of the pterygoid it is most similar to the 
jasciatus subgroup of that species group (Greer and Parker, 
1967). It differs from all members of its species group, how- 
ever, in having the tympanum both much reduced in size and 
decidedly more opaque (thickened?). 

Etymology. The name microtympanus calls attention to the 
relatively small size of the tympanic membrane of the species. 

Description. Since S. microtympanus is so similar to S. 
anotus just described, I will only give specific counts and mea- 



BREVIORA 



No. 406 




1973 KAKI.KSS SKIXK.S 7 

surcincnts f(jr >S'. yn'urolyynpdnus and will tl('s(iil)e im\\ tfiovc 
aspects of the species" niorpholog) in whicli it dilTers noti( cabU 
from S. anolus. 

Both the holotxpc and j:)aratype ha\e a snout \ent len^ili of 
45 mm; in the paratype the tail is broken but in the t\jxj ii i.s 
complete and measures 60 mm. Both specimens of .V. tnirru- 
tympanus are a richer chocolate brown abo\c than .S'. anulus 
and in preser\'ati\e both lack the yellowish wash on the under- 
sides shown b\- the holotypc of S. anolus. These color differences 
may, however, be an artifact of preservation. 

There is a scaly auricular depression \er\- similar to the auri- 
cular depression of S. anolus, but at the bottom of the depression 
there is a small, opaque t)mpamnTi instead of scales a.s in 
S. anolus (Fig. 2). 

Both type specimens of S. microlympanus have four pairs of 
nuchal scales, and, in three out of the four cases, there are six 
supralabials with the fourth situated most directly below the 
eye; on the rioht side of the head in the paratype there are 
seven supralabials and the fifth is under the eye. There are 26 
scale rows at midbody and the scales of the two mid-dorsal 
rows number 75 in the paratype and 71 in the holotype when 
counted from the parietals to the midpoint of the insertion of 
the hind legs. The subdigital lamellae on the fourth toe number 
8-10. 

Color. In addition to being richer brown above and lacking 
the yellowish wash below, the type and paratype of S. micro- 
lympanus differ from S. anolus in ha\"ing a very noticeable 
brown wash on the throat and chest instead of a ver\' faint 
brown wa-sh limited to the throat as in the type of S. anolus. 
This wash is much more pronounced in the paratype of .?. 
microl\mpann'i than in the holot)pe. 

Dislribulion. S. microlympanus is known at present only 
from the type locality (Fig. 4). 

Habilal. According to Dr. Zweifel (letter, 14 February 1973), 
the ''undisturbed habitat around Garaina is rain forest on river 
terrace and foothills." 



Figure 3. '1 lie liolt)i\pe of Sl)linio»ini plnis niii rolynijxniu.s lAMXH 
104076) from Garaina. Xcw Ciuinca. 1 lie specimen has a snout-veni length 
of \'i mm and a tail length of HO mm. 



8 



BREVIORA 



No. 406 



^::z^ 




^> 



Figure 4. Map of eastern New Guinea showing the type localities of 
Sl)ln')Wiii()i j)hus aiiuliis (Masba Creek = dosed star) and of S. iiiicro- 
t\)iil>auus (Garaina = open star) . Tiie two species are known only from 
their type localities. 



Relationships between S. microtympanus and S. anotus. These 
two species are so similar in all aspects of their external mor- 
phology, sa\e for the nature of the external ear, that they are 
almost certainly each other's closest known lixing relati\es. 
Indeed, it looks as if S. anotus could have ea.sily e\ol\ed from 
a microtxm panusAxkt ancestor simply by ha\ing the scales on 
the sides of the auricular depression extend down and oxer the 
\ery small tympanum at the bottom of the depression. 

The relationship between these two taxa appears to be so 
close that I onginally thought it might be possible to describe 
them as the same species. To do so would ha\c required only 
that one bclicxe that the \ariation shown in the external ears 
of the specimens exists within a single species. But this kind of 
variation is unknown in better studied skink species [e.g., the 
skinks of the earless genus Heyniergis or the many species of 
Lerista, which ha\e minute ear openings), and to conclude 



1973 EARLESS SKINKS 9 

that it exists within or between the populations represented by 
these four specimens seems presumptuous.' 

Relationships with other skinks. Wiiiiin the fasriatus species 
group of Sphenomorphus there are only four other previously 
described species that arc like ynierotympanus and anotus in 
possessing the following suite of characters: relatively small size 
(maximum snout-\'ent length 55 mm or less) ; more or less 
uniformly dark dorsal color; a moderate number of midbod\- 
scales (overall range, 20-28; range of modes, 24-26), and a 
low number of subdigital lamellae on the fourth toe (upper 
limit of range not exceeding 16). All iouv species occur in New- 
Guinea and in the order discussed below they are forbesi, 
sehultzei. beaujorti, and olioolepis. Comparati\e data for these 
four species plus ynicrotxmpdnus and anotus are presented in 
Table 1. 

Forbesi. On the basis of palatal morphology, forbesi seems 
quite distant from ynierotympanus and presumably also anotus, 
although I ^vas not able to examine the palate of this last species. 
In forbesi there is an ectopterygoid process to the palatine 
which excludes the palatal ramus of the pterygoid from a posi- 
tion on the infraorI:)ital \'acuitv. In rnierotympayius and pre- 
sumably also in anotus there is no ectopter\goid process and the 
palatal ramus of the pter\ooid enters the infraorbital vacuity. 
The presence or absence of the ectopter\goid process may reflect 
a basic e\olutionar\' dichotomx' in the fawiatus species group 
' soloynonis subsirotip \s. the jasciatus subgroup — fide Greer 
and Parker. 1967) and on this basis alone I would exclude 

'Thcic is f)ne pooilv .iii.il\/r<l |)iftc(lcnt foi iiu hiding skinks lioth \\itli 
nnd withoiii an cMcinal opening in liu- same- -pec ics. Fiilm n'lfiO) lins 
trcalcd \\\v earless .} hlrlil/in ii ^ ;Mrnvn/)/s as ,i snhspecies of A. j)a)nioriiriis, 
a s|)e(ies wiili a minnle (xieiiial ear opening. His Iiaid-roie e\i(len(e for 
this la\onf)niic nio\e is appaunlh (onlained in llie following sc-niente: 
"Mortens nOGt. in litl.) records also spefiniens of \fglian .-1 . jmnnniiinis 
popnlaiions wiili no ear openings uoU. l)i. K. I indhcig) ." lint >rei tens' 
(MHi'ii pnhlislied e\iden(e is nothing more than a hrief deseiiption and 
iinich disnission of one spednien ({\\v onl\ one fi(ini that lo(alit\) wliirli 
he said looked like ji(niii(nii( ii\ hut whiili lacked an e\lemal ear opening, 
the ke\ (haiacler of '^vaxdiim. I nlin himself exainiiu'd a total of oid\ six 
s|)((iinens of hoth spetinicais (llirte jxiiiiKitiif ii\ fiom one lo(alit\ and 
tlnee ^i(i\fmii<: ta<h fiom a dilieien: loralilv). none of wliidi lie re])Oits 
as heing inuisnal nith ie<;ai(l to thi' eMeinal eai. !'eiha|)s mote evidence 
exists in \reriens' in lill. ( onimnii i( at ion. hut until that is foi throming. 
the (ase for inti aspei ifle \aiiation in the ])Ks-,ii(e or ahsencc of an external 
e.ii in skinks is. at hest. on s]iak\ giouiul. 



10 



BREVIORA 



No. 406 



•2 



u 






— 5J 



"^ 3 



tn 


— 






r^ 


tr. 








■r. 






r^ 


Zj 






^ 


U 









g - s 

= 2 bc 



o 







V) 




o 








^ 




c 








>~ 




_>- 








0! 












i-l 








o 


o 


o" 


o 






z 


Z 


Z 


z 


>■ 


>i 



.^ o ~ 



ir. — 

X X 



1<5 

3 



1^ cr-, 
— X 






C-. 






■^ X 



!M 






?i ?. s 






1973 EARLESS SKINKS 11 

forbcsi from close relationship with micrulyrnpanus and anolus. 

Schultzei. The palate of schullzei is similar to that of micrn- 
tyyyipanus in that it lacks an ectoptervgoid process, but other 
features of its morphology cause mc to exclude it from the 
close relatives of ynicrotympanus and anotus. The most notable 
of these features is the unique fusion of the first supralabial 
and nasal scales (see below), the medially meeting prefrontals 
(in most specimens), and the proportionately longer legs. 

Beaujorli. As far as I can tell, beauforti is known only from 
the type specimen (de Jong, 1927) and I have not seen this 
specimen. I feel, howexer, that the absence of nuchal scales 
and the medially meeting prefrontals are enough to make beau- 
forti an unlikely near relative of microtympanus and anotus. 

OUgolepis. In contrast to the preceding three species, 
oligolepis is in every way a perfect candidate for the closest living 
relative of ?nicrotympanus and anotus. The palate of oligolepis 
is very similar to that of rtiicrotyyyipanus and presumably also 
to that of anotus; all three species are similar in size and body 
proportions, and there is no significant difference in the general 
details of squamation. Oligolepis differs markedly from micro- 
tympanus and anotus only in having a well-defined external 
ear opening and ear canal at the bottom of which is a trans- 
lucent tympanum (Fig. 5), but since this kind of ear was 
undoubtedly primitive for the microtympanus^ anotus line, 
it simply serves to make oligolepis the closest living species, 
morphologically, to the ancestor of that line: oligolepis^ 
microtympanus-^ anotus. 

Comments on the Loss of the 
External Ear in Lygosomines 

With the desecription of Sphenomorphus anotus, the total 
number of known "earless" lygosomines comes to 33. This is 
about 5 percent of the total number of known species in the 
subfamily.^ 

Taking a \ery conservative view of the species relationships, 
I beliexe that these 33 species represent no fewer than ten dif- 
ferent lineages. Or, to put it another wav, the external ear has 
been lost at least ten difTerent times in the ex'olutionarv history 



'This pcKcntase is vci v low compared to the other three subfamilies 
of skinks. All of the fcylinines (4 species) and acontines H") species) lack 
an external ear opening and just under 2") percent of tlu- scincines, of 
wliich there are a total of approxiuKUclv 182, are also "earless." 



12 



BREVIORA 



No. 406 






1973 EARLESS SKINKS 13 

of the living lygosomines. The species in these ten groups, along 
with their distributions and other pertinent data, are listed in 
Table 2. 

Unfortunately, it is difficult to say anything very conclusive 
about why the skinks in these different groups have Tost the 
external ear, but I can make a few comments and suggestions 
for further research along these lines. 

First, there is probably no one unifying reason for the loss 
of the external ear in all ten groups since there is nothing in 
the biology of these skinks beside the absence of an external 
ear that sets them apart from other lygosomines. It is true that 
all the earless species are in some sense cryptic in their habits 
but this is the rule rather than the exception for skinks. 

Second, the only outstanding ecological feature of any of the 
species in the list of earless lygosomines is that certain of the 
species, i.e., Isopachys, most of the australis group and perhaps 
the sumatrense group, appear to be the most confirmed burrow- 
ers among lygosomines. This fact indicates that the absence of 
the external ear is probably a prerequisite for an in-depth evolu- 
tionary commitment to burrowing life, but there is no way of 
knowing whether the loss of the external ear in these skinks 
was achieved as a primary adaptation to burrowing life or 
whether it was a preadaptation, i.e., originally evolved for other 
reasons in nonburrowing ancestors. The large number of ear- 
less lygosomines that show no exceptional procUvity to a bur- 
rowing life, e.g., the quadrivitattum group, Anotis mariae, and 
Ablepharus grayanus, would argue that the loss of the external 
ear could be as much a preadaptation to burrowing life as it 
is a primary adaptation. 

Third, Minton (1966) has suggested that, along with the 
ablepharine eye, the absence of an external ear opening in Able- 
pharus grayanus is a protective adaptation that allows this 
species to feed unmolested on the ants that are said to form 
much of its diet. This is an interesting idea, but to be con- 
\incing, it will have to be shown more rigorously than it can 
now be shown that A. grayanus is more of an ant specialist 

Figure 5. Lateral view of the head of Sphenomorphus microtympanus 
(top; paratype: MCZ 132767) , S. oligolepis (middle; syntype: BMNH 
1946.8.3.47) . and 5. solomonis (bottom; syntype: BMNH 1946.8.34-37) . Note 
the relatively small external ear opening and small tympanum of S. micro- 
t\>iij>n)i)is compared to its dose relative S. oligolepis and its more distant 
species gioup relative S. solomonis. 



14 



BREVIORA 



No: 406 






c" 



c 

E 



^ 


O 


C 


5 






O 




o 


n 




■^ 


^ 


c 


Cj 


.^T" 










^ 


;^ 


~ 


W 


^ 


T" 


3d 


J'. 


r: 


O 






"Eh 




~ 


^,, 








rj 


t/: 


^1^ 


c 


^ 


o 








"c? 




S^ 


be 


C 


= 


O 


%/i 










c 


CJ 


o 


«J 


o 




^ 


r* 


o 


— 


*- 


*^ 


F— < 


^J 


I_ 


C 


c 




c 




r: 


. 


o 


















be 


V3 


E 


Hf 


C 




'^ 




r3 


^ 


O 


-;^ 


u 


f/: 


c 


c 



n o 



rs — O 



r^ 


o 


o 


w 


• M 




c 


7^ 


•:<: 








Vl 


_' 


^— H 


rt 




c 


c 


O 










'e 


E 


^ 


■"■ 


o 


— 


r: 


z^ 


(/: 






o 


c 


, . 


c/: 


^ 


be 


■y: 


^ 


Sc 








?— t 


*-< 


r: 










u 
















> 


c« 








«! 


O 






o 




y: 


— 










^ 




tf: 




rt 








o 


E 


'^ 


'X 


* 


c 


— 


o 










c 


ly; 






O 






;/; 




.iii 


^ 


Cj 


Z. 


-f^ 


^J 


'u 




^ 


./: 


o 


^ 




O 


a, 


"w 


'm. 


sz 


w 


..■ 


^ 


■^^ 




5" 


<;*. 






r- 






(« 




V 


>-> 


c^". 


t/: 




*w 


o 


^ 


^ 


~ 



CM 






^ o 



r- be ? 



c 
o 






a ~ 



- v: - - be - - - ;£ .- C;^ - ^ r: 



= . ■ CO -c - — •- a" -S 



^"^ 

_ ^ 



—. 


ly: 


^ 


_ = 


o 


v-' 


j; 


o 


.s 


~ 


c 


r^ 




o 


jr 


= 




5 




cr. 




.r: 


* 




— 




""^"^ 


^ 


^' 


y. 














1 














>-. 


Cj 




■~ 


:^ 


be 




C/j 


1 


1 


"~ 


' z' 


'• ^ 


be 




•^ 


be 


^- 


._= 


■< 


.E 


— ■ 


- 


^ 


^^ 


''^- 


'"" 


r; 


_o 


*; 


r3 


cr 


"C 


^ 


v. 


p 


< 


■^ 


^ 


■§ 


o 


o 




o 


c 


5c 


"5: 



? s^ ' - i 



s a 5 s :s, 

Ch C3 •'-- be t/3 



c 






C 


^ 


c/2 






c/; 


c 


S 


z 


;z 


> 


■-/; 


o 






o 


i* 


;z 




'o 


Z 


^ 
o 


\^ 


r3 


rt 




;z 


z 


'^ 


rz 


S 


z: 



n - -z 



o 






o 

CJ 



= c 



C> J-. z:l ^> ^ 



^> — /-«/ w 



c ^ 



c/; 



C 



O 

an 



"^ n.' ;: o ;; v 
fa' ^, = Z = 2 

Z J7: U U 



o 






C 

;j 
o 
>^ 



c/; :/2 















oc 














■Xl 












O 


00 












1^ 












00 

00 


GO 


'/) 


-J" 


= 




X 


00 


X 


'k- 


^ 


— 


^ 


^^ 




— 


o 








■ ■ v: 


->: 


o 


^^ 


~ 


■r- 


^ 


<- 


o 


^, 
















^ 


s 




^ 

C- 




2 


2' 
'be 


5. 


£/: 


ti" 


i^ 




c 




■^- 






ri 




K 


i 



00 — _ 



-^ -2 2 



•^ -w -o 



o 
c 



CO 




so 


C-5 
1^ 




„ C"' 


00 




1^ 

00 


oo 




CO — 


^" 








w 


?. ■-!_ 


o 






'■J 


o 





t ~ •£ -^ 



c 
o 



a, 






o 






rt 



o 



^ i o S 



Cm C: CI. 



o 

O ^ 

be to 



1973 



EARI.KSS S KINKS 



15 



o ;•<■ 
"!;! > -' 



C o -c 



c "^ - 



O r- 



cr. z 
_— -7, 



4J Y', 






2 _• ^ 



!r 2 c - 



c "5 



tiC <y. 



■^, .£r- CL .? 



rt 



~ ~ ^ S -5 S 
5 2 -5 ^" c £ 



"- O r: 



C ^ ^^ 



v; 



C/3 



b o- o £ o u 



o 



f ■'-■ -. — c^ - 



o 5 



o 



o 



C3 

£ 

iy5 



2 "? 



o 



C 
rt 



rt 

C 



O 



— .2. -r J2 
-J: "3 CJ E 



.2^ U 












-) -^ -J-. 






/^ 






c 
c 



h 



1^ 

I- i --- i 

•o 

X - 

— iJ 



1- GC 



h-. 



\~- ~\ 



•=. "T. ■Z 



& br.-^ ^ 



~ - — = -^ 



be r;- 



br. 



^ .^ -r -^ 



Hi BREVIORA No. 406 

than its close relati\es that have external ear openings. I know 
of onl)- one careful study on the feeding habits of an earless 
lygosomine (Smyth, 1968, on Hemiergis peroni), and this 
species showed no special fondness for ants or any other small 
insects that might pose a serious threat to an exposed tympanic 
membrane. 

The most profitable next step in the study of the e\olution 
of earlessness in lygosomines would be to take a group of earless 
lygosomines that are known to have close relatives with external 
ears and make detailed ecological comparisons between the two 
groups. The three best groups for this kind of study now are 
1 ) the quadrivitatturn group and its close relatives ■ — the sharply 
light and dark striped skinks in the genus Leiolopisma of south- 
east Asia, the Philippines, the Indo-Australian archipelago, the 
Palaus and the New Guinea area; 2) Ablepharus grayanus and 
its congeners in eastern Europe and southwestern Asia, and, 
perhaps somewhat impractically because of the distribution, 
3) Anotis mariae and its two congeners on New Caledonia. 

Observations on Sphenomorphiis forbesi, S. oligolepis 

AND S. schultzei 

Sphenomorphus forbesi Blgr. 1888 and S. oligolepis Blgr. 1914 

In spite of the fact that S. forbesi and S. oligolepis have an 
osteological difference in the palate (see pages 9, 10) that serves 
to separate them rather distantly in terms of relationship, the 
two species are extremely similar in their external morphology. 
They are so similar, in fact, that Miss A. G. C. Grandison of 
the British Museum and I both agreed, after an initial examina- 
tion of the type specimens, that the two forms were conspecific. 
It was only after a second and more detailed look with more 
specimens that I was finally convinced that the two forms are 
good species. It was not, however, until after I had decided that 
the two forms were good species on the basis of external mor- 
phology that I discovered the confirming osteological difTerence 
in the palate. Thus the two species can be distinguished with- 
out resorting to the sometimes rather destructive process of open- 
ing the mouth in preserved specimens and examining the palate. 
Because the two species are so similar externally they are re- 
viewed here together. 

Sphenomorphus forbesi was described by Boulenger (1888) 
from a single specimen collected at Sogere (= Sogeri) by H. O. 
Forbes at an elevation of 1750 feet on his expedition into the 



1973 EARLESS SKINKS 17 

Owen Stanley Range behind Port Moresby. The only other 
specimen to be reported since the original description is a single 
specimen from Bara Bara, Milne Bay, collected by L. Loria 
(Boulenger, 1897). 

Sphenomorphus oligolepis, also described by Boulenger 
(1914), was based on two specimens from the Mimika River 
collected on the British Ornithologists' Union Expedition, and 
the only new locality information published on this species since 
its original description has been de Rooij's (1915) listing of a 
specimen from the Lorentz River. 

Both forbesi and oligolepis are very similar to each other in 
terms of general squamation and color pattern, but they can 
be distinguished from each other in terms of size, small details 
of squamation and subtle differences in color pattern. These 
differences have been worked out from an examination of the 
following specimens: the type of forbesi (BMNH 1946.8.3.13) ; 
ten topotvpic or nearlv topotypic forbesi (MCZ 118845-47, 
118851-53: Sogeri, 2000 feet; MCZ 118848-50: Sogeri road, 
2 miles east of Rouna Falls, 1500 feet; American Museum of 
Natural History 103602: Sogeri, 460 meters) ; and three speci- 
mens that appear to be forbesi on comparison with the type 
(MCZ 13357-58; AMNH 105626: Wipim) one syntype of 
oligolepis (BMNH 1946.8.3.47); and 25 specimens which I 
identify as oligolepis after comparing them with the svnt)'pe 
(MCZ 118857: Soliabeda, 1800 feet; MCZ 109330-47; 
118854-56: Oroi; 130716: Matkomrae; MCZ 130717: Men- 
dua, and MCZ 130718: Bikim, 500 feet). 

The significant differences betv/een the two species are as 
follows : 

Size. Forbesi is a smaller species than oligolepis; the largest 
oligolepis I examined (including the syntype) measured 55 mm 
in snout-vent length, whereas the largest forbesi I examined 
measured only 44 mm. 

Squaynation. In forbesi the fourth supralabial is centered be- 
neath the eye (in all 28 cases provided by the 14 specimens) 
whereas in oligolepis it is the fifth supralabial that is more 
usually centered beneath the eye (the fifth in 43 out of 52 cases 
and the fourth in 9 out of 52 cases). 

In forbesi the first infralabial is only about 1/2 to 2/3 the size 
of the second infralabial, whereas in oligolepis the first and 
second infralabials are about the same size. 

Color. In preservative forbesi is generally golden brown to 
lifjht brown above with small dark blotches and vermiculations; 



8 



BREVIORA 



No. 4()r) 




■ilk 







Figure (). Lateral view of tiie holotvpe of Splivui'inorphus forhcsi (top: 
BMNH H)4().H..S.l,'?; snout vent length = 40 mm) from Sogcii aiul S. olioo- 
lejiis (hoiiom; RNFNH 1946.8.3.47; snout-vent length = 55 mm) from the 
Mimika Ri\tr. Note that the dark longitutlinal lateral lines contrast with 
the (lark color of the dorsum in forhcsi hut not in oligolepis. 



oligolepis, on the other hand, is a more uniform and darker 
brown above and lacks the contrasting darker blotches or \ermi- 
culations. In both forbesi and oligolepis pigment tends to con- 
centrate in longitudinal lines running between the lateral scale 
rows, and in forbesi these longitudinal lines stand out becau.se 
the pigment is generally darker than the general ground color 
of the back, whereas in oligolepis the longitudinal lines do not 
stand out because the pigment is no darker than the dorsal 



ground color 



Fig. 6 



1973 



EARLESS SKINKS 



19 



(PZ^-v 




^^ 



Figure 7. Map of eastern New Guinea showing the known localities for 
SpJienornorphiis forbesi. 




<%^- 



Figure 8. Map of New Guinea showing the known localities for Spheno- 
morphus oligolepis. 



20 BREVIORA No. 406 

Two of the MCZ forbesi which I examined were gravid. One 
specimen with a snout-\-ent length of 40 mm contained a single 
thin-shelled egg in the right oviduct and the other specimen, 
which measured 43 mm in snout-\-ent length, contained two 
thin-shelled eggs, one in the right oviduct and one in the left. 
To judge from the texture of the egg shells in both specimens, 
it is likely that the species is oviparous. 

Five of the MCZ oligolepis were gravid: four with a thinly 
shelled egg in the right and left oviduct and the fifth with a 
single large ovum in each ovar\\ It would thus appear that 
oligolepis is oviparous. The smallest of these gravid females had 
a snout-vent length of 43 mm and the largest had a snout-vent 
length of 53 mm. 

The known distribution of jar be si and oligolepis is shown 
in Figures 7 and 8 respectively, 

Sphenomorphus schultzei Vogt 1911 

The two types of Sphenomorphus schultzei (Berlin 22135) 
were examined because the species seems to be a member of 
the fasciatus species group and within this group it combines 
a moderate to low midbody scale count with a low fourth toe 
subdigital lamellae count (Table 1). 

Theodore Vogt (1911) described this species on the basis of 
two specimens captured by Leonard Schultze on an unnamed 
mountain at an elevation of 1570 meters in the region of the 
Sepik River below 5° latitude. Four years later de Rooij (1915) 
listed the Sermowai River as an additional locality for the 
species, but beyond this no new records ha\'e been published 
for the species. 

In examining Vogt's two syntvpes (Fig. 9) I immediately 
discovered that the first supralabial and nasal scale are fused 
into a single scale. That this fusion is not an anomaly is proved 
by the fact that it exists in the 1 1 schultzei from four different 
localities which have recently been added to the MCZ collec- 
tions through the efforts of Fred Parker ^ MCZ 89897-99: 
Bomai, Tive Plateau, 3500 ft.; MCZ 124037-40: Tifalmin, 
4300 ft.; MCZ 124041-43: Wangbin, 4800 ft.; MCZ 124044: 
Imigabin, 4200 ft.). These and other known localities are 
shown in Figure 10. 

Apparently neither Vogt nor de Rooij noticed the scale fusion 
for neither author mentions it. There is a crease between the 
two scales that may have been mistaken for a suture with 
early optical equipment, but with good light and modern optics 
there is no doubting that the scales are fused. 



1973 EARLESS SKINKS 21 




1 ^ 3 <4 . ^ 



Figure 9. Lateral view of a syntNpe of Sjihnioinoi pints schultzei ('Berlin 
22135; snout-vent length = 34 mm) from tiie region of the upper Scpik 
River. 



To my knowledge no other lygosomine skinks ha\'e a fused 
first supralabial and nasal scale, and thus this character pro- 
\ides a blessedly certain method of identifying at least one 
species of a notoriously difficult ''generic"' assemblage of 
Ivgosomines. 

X^ariation in the taxonomically important characters of the 
13 schultzei that I ha\e examined (12 intact and one decapi- 
tated) ma\' be summarized as follows: The two types ha\e 20 
midbody scale rows, but the 1 1 MCZ specimens ha\-e from 22 
to 26 midbody scale rows. The modal number for all 13 speci- 
mens is 24. The range in the number of subdigital lamellae on 
the fourth toe for all specimens is 8-13 ( a\g. = 10.9). The 
mmilier of nuchal scales on the left and right side of the mid- 
line in the 1 2 intact specimens ranges from 0-0 to 3-2. In all 
but two of 1 2 intact specimens the prefrontals meet medialh' 



00 



BREVIORA 



No. 406 




Figure 10. Map of New Guinea showing the known localities for Spfieno- 
morphus schultzei. 



and form a broad suture; in the remaining two specimens the 
right prefrontal is lacking in one and the prefrontals are sepa- 
rated in the other. In all 12 intact specimens the fourth supra- 
labial is situated beneath the eye. The largest specimen mea- 
sures 39 mm in snout-vent length. 

In two of the MCZ specimens I found a single large, heavily 
shelled egg in the right oviduct. And in the only specimen of 
these two in which I looked for a left o\iduct, I could find none, 
although there was a left ovar)-. It would appear, therefore, 
that the species is oviparous with a clutch size of one, and that 
it may lack a left o\'iduct. 

Fred Parker has very kindly summarized his field notes on 
schultzei for me, and I have extracted the following information 
nearly verbatim from his notes. In the Bomai area (Fig. 10) 
schultzei is found in dense rain forest where it li\-cs under decay- 
ing logs and \egetable matter on the forest floor in damp \alleys. 
It is a fairly slow moving species. It is subject to rapid de- 
hydration if not kept damp. 

In life the iridescence of the scales almost conceals the color 
pattern. The dorsal surfaces are mottled pale and dark brown. 
There are some fine white specks on the lips and face. The 
flanks are brown with paler spots. The ventral surfaces are pale 
translucent yellow with some grey spots in the \entrolatera] 



regions. 



1973 earless brinks 23 

Acknowledgments 

I thank Dr. Richard G. Zweifel of the American Museum 
of Natural History for making the specimens of anolus and 
microtympanus available to me, and also for providing me with 
the maps of New Guinea used in Figures 4, 7, 8 and 10. Dr. 
Zweifel also kindly helped me locate certain localities in New 
Guinea. 

Miss A. G. C. Grandison of the British Museum (Natural 
History) was very gracious in lending me the types of oU gale pis 
and jorbesi, not once, but twice, and in making comparisons 
for me in the British Muesum. Dr. Giinther Peters of the Zoo- 
logisches Museum (Berlin) was equally generous in lending me 
the types of schultzei. 

Mr. Fred Parker generously gave me his field notes on 
schultzei. 

Mr. Laszlo Meszoly did the drawings for Figures 1 and 2, 
and a grant from the Milton Fund administered through the 
Harvard Medical School financed the photographs of the type 
specimens in Figures 3, 5, 6 and 9. 

Literature Cited 

Arnold, J. M. 1966. A taxonoiiiic study of the lygosomid skinks of Queens- 
land. >r.S. Thesis, University of Qucenslantl. 246 pp. 

BoL'LENGER, G. A. 1888. Descriptions of new reptiles and batrachians ob- 
tained by Mr. H. O. Forbes in New Guinea. Ann. Mag. Nat. Hist., 
Scr. 6, 1 (5) : 343-346. 

■ . 1897. An account of the reptiles and amphibians col- 
lected by Dr. L. Loria in British New Guinea. .\nn. Mus. Civ. Stor. 
Nat. Geneva, Ser. 2, 17: 694-710. 

. 1900a. Descriptions of new reptiles from Perak, Malay 



Peninsula. Ann. Mag. Nat. Hist., .Ser. 7, 5(27) : 306-308. 
. 1900b. Descriptions of two new lizards from Selangor. 



J. Eombav Nat. Hist. Soc, 1.^(2) : 333-334. 
. 1903. Report on the batrachians and reptiles: pp. 131- 



176 in N. .4nnandalc and H. C. Robinson. Fasticuli Malayensis. Zoology. 
Part 1, .303 pp. 

1908. Report on the Gunong Tohan Expedition, May- 



Septcniber, 1905. IIL Fishes, batrachians and reptiles. J. Fed. Malay 
States Mus., 3: 61-69. 

1914. An aiuiotatcd list of the batrachians and reptiles 



collected by the British Ornithologists' Union Expedition and the 
■Wollaston Expedition in Duidi New Guinea. Trans. Zoo). Soc. London, 
20 (.-)) : 247-274. 



24 BREVIORA No. 406 

BoLRRET, R. 1937. Notes hcrtctologicjucs siir I'lndochine frangaisc. XV. 

Lc'-zarcls ct serpents rcgiis au Laboratoire des Sciences Natuielles de 

rUnivcisite au cours de I'annee 1937. Descriptions de deux espcces et 

de deux varietes nouvelles. Bulletin General de I'lnstruction Publique 

(Hanoi) , No. 4: 57-80. 
. 1939. Reptiles et batraciens requs au Laboratoire des 

Sciences Naturclles de lUniversitc au cours de I'annee 1939. XVIII. 

Descriptions de quatrc especes et dune variete nouvelles. Bull. Gen. 

Instr. Publq. (Hanoi) , No. 4: 5-39. 
Brongersma, L. D. 1942. On the arrangement of the scales on the dorsal 

surface of the digits in Lygosoma. and allied genera. Zool. Meded., 24 

(1-2) : 153-158. 
Brown, W. C., and A. C. Alcala. 1961. A new sphcnomorphid lizard 

from Palawan Island, Philippines. Occ. Pap. California Acad. Sci.. 

No. 32: 1-4. 
Bustard, H. R. 1964. Reproduction in the .'\ustralian rain forest skinks, 

Siaphos eqiinlis and Sphenomorphiis tryoni. Copeia. 1964(4): 715-716. 
Copland. S. J. 1952. A mainland race of the scincid lizard Lygosoma 

truncatum (Peters). Proc. Linn. Soc. New South Wales, 77(3-4): 126- 

131. 
Dumerii,, a. 1851. Catalogue nicthodique de la collection des reptiles. 

Mus. Hist. Natur. Paris, iv -\- 224 pp. 
FuHN, I. E. 1969. Revision and redefinition of the genus Ablepharus 

Lichtenstein, 1823 (Reptilia, Scincidae) . Rev. Roum. Biol. (Zool.), 

14(1) : 23-41. 
Gray, J. E. 1825. Synopsis of the genera of reptiles and amphibia, with 

a description of some new species. ,\nn. Phil., Ser. 2, 10: 193-217. 
Greer, A. E., and F. Parker. 1967. A new scincid lizard from the northern 

Solomon Islands. Breviora, No. 275: 1-20. 
GuNTHER, A. 1873a. Notes on some reptiles and batrachians obtained 

by Dr. Adolf Bernhard Meyer in Celebes and the Philippine Islands. 

Proc. Zool. Soc. London, 1873: 165-172. 

. 1873b. Notes on, and descriptions of, some lizards with 

rudimcntarv limbs in the British Museum, .^nn. Mag. Hist.. Ser. 4, 

12(68) : 145-148. 
Jong, J. K. de. 1927. Reptiles from Dutch New Guinea. Nova Guinea, 

15(3) : 296-318. 
KorsTEiN. F. 1926. Reptilien von den Molukken und den benachbarten 

Inseln. Zool. Meded.. 9: 71-112. 
Longman, H. A. 1916. Snakes and lizards from Queensland and the 

Northern Territorv. Mem. Queensland Mus., 5: 46-51. 
Mertens, R. 1965. Bemerkungen iiber einige Eidechsen aus .Afghanistan. 

Senck. biol., 46(1) : 1-4. 
MiNTON, S. A., Jr. 1966. A contribution to the herpetology of West 

Pakistan. Bull. Amer. Mus. Nat. Hist.. 134(2): 27-184. 
Peters. W. 1867. Herpetologische Notizen. Monatsber. Berlin Akad. Wiss., 

1867: 13-37. 



1973 EARLESS SKINKS 25 
. 1873. Einc Mitlheilung iibcr neue Sauricr (Spaerioclactylus, 



Anolis, Phrynosoma, Tropidolepisma, I.ygosoma, Ophioscincus) aus 
Ccntralamtiica, Mexico unci Austialien. Monalsbcr. Berlin Akad. W'iss., 
1873: 738-744. 
. 1S76. Uber die von S.M.S. Gazelle mitgebrachten Amphibien. 



Monatsber. Berlin Akad. Wiss., 1876: 528-535. 
Rooij, X. DE. 1915. The Reptiles of the Indo-.Austialian .\rchipelago. I. 

Lacertilia, Chelonia, Eniydosauria. Leiden: E. J. Brill Ltd. xiv + 384 

pp. 
Smyth, M. 1968. The distriI)utioii and life history of the skiiik. Uemiergis 

peronii (Fitzinger) . Trans. Roy. Soc. South .Australia, 92: 51-58. 
Taylor, E. H. 1963. The lizards of Thailand. Univ. Kansas Sci. Bull., 

44(14): 687-1077. 
TwEEDiE, M. W. F. 1940. Notes on Malayan reptiles. Bull. Raffles Mus., 

No. 16: 83-87. 
Van Dei sen, H. M. 1966. The seventh Archbold Expedition. BioScience, 

16 (7) : 449-455. 
Vinciglerr.\, D. 1892. Rettili e batraci di Engano raccolti dal Dott. Elio 

Modigliani. Ann. Mus. Civ. Stor. Xatur. Genova, Ser. 2, 12(32) : 517-526. 
Vis, C. W. de. 1888. A contribution to the herpetology of Queensland. 

Proc. Linn. Soc. New South Wales. Ser. 2, 2(4): 811-826. 
VocT. T. 1911. Rcptilicn und amphibien aus New-Guinea. Sitzungsber. 

Geselisch. NaturforsMi. Ircundc 1911. No. 9; 410-420. 



^'^^-'^. COMP. ZOOL 



B R E V I «>'* A 



MARVARO 



useuni of Comparative^^iMlogy 



us ISSN 0006-9698 



Cambridge, Mass. September 20, 1973 Number 407 

THE GHANARES (ARGENTINA) 

TRIASSIG REPTILE FAUNA. 

XIX. POSTGRANIAL MATERIALS OF THE 

GYNODONTS PROBELESODON AND 

PROBAINOGNA THUS 

Alfred Sherwood Romer and Arnold D. Lewis 

Abstract. Descriptions are given of postcranial materials of Probelesodon 
and Proba'nwgnallnis and a restoration of Probelesodon lewisi is attempceu. 

Because of the phylogenetic position of the cynodont therapsids 
as the probable ancestors of mammals, their structure is of great 
importance in the story of vertebrate evolution. A number of 
excellent studies have been made of cranial structures of c)no- 
donts. As regards the postcranial skeleton, there have been 
numerous descripti\e papers, but few which ha\'e attempted a 
broad study of cynodont skeletal materials from an evolutionary 
or functional viewpoint. Early essays of this sort were those of 
Watson (1917), Gregory and Camp (1918), and Romer 
(1922) ; a recent comprehensive work is that of Jenkins (1971; 
cf. also Jenkins, 1970). 

Postcranial remains of African cynodonts ha\'c been compre- 
hensively studied and summarized by Jenkins (1971); some 
data on Permocynodon of Russia have been given by Konjukova 
(1946). Of the South American cynodonts, descriptions of 
gomphodont skeletons ha\e been given by Bonaparte ( 1 963 ) 
for Exaerctodon, and by Jenkins (1970) for Alassetognathus, 
and Huene ( 1 944 ) has described a limited amount of post- 
cranial material of Traversodon. For the carnixorous c\nodonts 
of South America, all so far published has been the description 
of a partial skeleon of Belesodon by Huene (1944), and of a 
limited amount of material of Chiniquodon by Romer (1969). 

The purpose of the present paper is to place on record such 
data as are available on the postcranial skeleton of the car- 



2 BREVIORA No. 407 

nivorous cynodonts of the Chafiares Formation — Probaino- 
gnathus and Probelesodon. Together with their relatives, 
Chiniquodon and Belesodon, from the Santa Maria Formation 
of Brazil, they include the latest in time and most advanced 
of therapsids leading in a mammalian direction. As Jenkins 
has noted (1971), the postcranial skeleton of cynodonts shows 
in general a remarkable consistency of pattern, and hence little 
novelty is to be expected from the description of the Chafiares 
genera. Collection and preparation of Probelesodon and Pro- 
bainognathus were made possible by grants from the National 
Science Foundation. 

Postcranial Material of 

Probelesodon lewisi 

A major source of information is MCZ 3781 (field no. 79), 
the remains of a nodule that contained a nearly complete and 
mostly articulated specimen of this form, with a skull approxi- 
mately the size of the type. This is preserved as a slab, prepared 
on both surfaces. Photographs of the two sides (which we will 
call the obverse and reverse sides) are shown in Figures 1 and 2. 
Outhnes of the materials seen on the two surfaces are shown 
in Figures 3 and 4. Figures 3 and 4 were derived from the 
photographs and, owing to perspective, the two are not com- 
pletely superposable. Unfortunately, some parts of the speci- 
men were lost before collection; further, the specimen had 
undergone considerable weathering and penetration by grass 
roots, with the result of obscuring much detail and rendering 
interpretation difficult. 

MCZ 4002 (field no. 98), in addition to a good skull, 
included a fair amount of mostly disarticulated postcranial 
material; this material is, in general, in better shape than that 
of MCZ 3781. Several other concretions include postcranial 
materials that may be of Probelesodon, but for the most part 
such materials either add little to the data available in MCZ 
3781 or MCZ 4002 or are of doubtful assignment. We may 
note, however, that MCZ 3801 includes a melange of bones, 
certain of which rather surely pertain to Probelesodon, notably 
a pair of excellent femora. 

Axial skeleton. At burial the specimen constituting MCZ 
3781 appears to have had a complete vertebral column, articu- 
lated for the most part. However, before collection, a con- 
siderable portion of the dorsal vertebrae had been lost, leaving 



1973 



CYNODONT POSTCRANIAL MATERIALS 







Figure 1. Obverse side of the slab, MCZ 3781, containing a skeleton of 
Probelesodon leivisii. Slightly less than l^ natural size. 



BREVIORA 



No. 407 




Figure 2. Reverse of the slab shown in Fig. 1. 



1973 



CYNODOXT POSTCRAXIAL MATERIALS 










Figure 3. Diagram of the structures on tlie obverse side of the slab 
shown in Figs. 1 and 2. Slightly less than I3 natural size. Abbreviations 
for Figs. 3 and 4: /, left; r, right; d, clavicle; /. femur; fib, fibula; /;, hu- 
merus; ic, iiiterclavicle; t, ilium; is, ischium; r, radius: sc, scapula; SI, first 
sacral; S5, fifth sacral; t, tibia; u, ulna. 



BREVIORA 



No. 407 



a 




^nxU^SS 



Figure 4. Diagram of the reverse side of the same slab. Slightly less 
than y^ natural size. Abbreviations as in Fig. 3. 



1973 CYNODONT POSTCRANIAL MATERIALS 7 

a small number of cendcal vertebrae present anteriorly, then, 
follo\sing a gap, posterior dorsals, "lumbars," sacrals and part 
of the tail. Although the articulated condition of the posterior 
part of the column and the forward continuation of a rib 
series in seemingh natural position indicates that much of the 
colunm was articulated at the time of burial, it is ob\ious that 
the column had been broken at about the posterior end of the 
cervical series. Se\'eral isolated \ertebrae, ob\iousl\- from the 
region of the break, are present abo\e and behind the skull. 

It is impossible to be certain of the exact presacral \ertebral 
count. As noted belo^v•, there is some question as to the position 
of the first sacral, but the articulated series anterior to this 
appears to include 14 posterior dorsals and "lumbars." For the 
most part the ribs associated with these vertebrae are preser\ed, 
and anterior to the \ertebrae present, this rib series is con- 
tinued for eight segments further to the 22nd segment anterior 
to the sacrum. The most anterior rib is not complete at its 
head, but has a moderately long shaft, and hence can be no 
farther forvvard than the posterior end of the cer\ical series. 
If 27 presacrals — 3. typical cynodont number — were present, 
this rib would pertain to \ertebra 6. Since presumabh the 
anterior cer\ical ribs were short, no more anterior position for 
this rib seems probable and we ha\e hence assumed for purposes 
of restoration the typical count of 27 presacral \ertebrae. 

In general the presers-ation of the vertebrae present is not 
good, but as far as can be seen the structure is of the general 
cynodont type. Remains of three \ertebrae are present close 
to the back end of the skull. The most anterior vertebra pre- 
ser\'ed has a neural spine (broken abo\'e) of considerable length 
anteroposteriorly, suggesting that we are dealing with the axis. 
Little more can be made out as to details of structure on the 
three further cervicals present in MCZ 3781 except for the 
presence of normal zygapoph) sial regions, well de\eloped trans- 
verse processes slanting outward, backward and downward, and 
in two of the three, neural spines that are relatively narrow 
anterposteriorly (the other — apparently the fourth in this series 
— -appears to have a widened spine). Isolated vertebrae that 
appear to be cervicals are present in MCZ 4002. As preserved, 
the centra are subcircular in end view, with a diameter of 
11-12 mm, and a central length of perhaps 10 mm. Transverse 
processes are well de\eloped and extend strongly out from the 
arch bases. 

Posterior cer\icals and much of the dorsal region is missing 



8 BREVIORA No. 407 

in MCZ 3781, although represented by a few scattered vertebrae. 
A number of dorsal vertebrae are present in MCZ 4002. Those 
best preser\ed have central lengths of about 12 mm, and height 
of centrum of 12-13 mm. They are deeply biconcave; the cen- 
trum, as seen in end \iew, is oval in outline, with a width some- 
what less than the height. Both anterior and posterior margins 
of the centra are somewhat thickened ; presumably the capitulum 
was carried on the anterior rim, but there is little indication 
of a discrete articular facet. The transverse processes are stout 
but short, extending but little outward beyond the level of the 
arch base, and face strongly downward. The posterior zyga- 
pophyses are almost directly above the base of the transverse 
processes; anteriorly the arch bases extend far forward, so that 
the spine extends upward about opposite the front margin of 
the centrum. 

Of the series of 14 presacrals preserved in MCZ 3781, and 
mainly \'isible on the reverse surface of the slab, the second to 
fourth are seen in side view and are fairly well preser\'ed. The 
neural spines are relati\'ely short, rising to about 15 mm above 
the level of the zygapophyses, and broad anteroposteriorly. 
Centrum lengths are 12-13 mm. Beyond this point the \-erte- 
brae are seen from above, and the neural arches and spines 
are for the most part absent, revealing the neural canal on the 
surface of the slab. The series is continuous, except for the last 
two presacrals, which are disarticulated and damaged. 

Beyond this point the sacrals and anterior caudals are articu- 
lated. Here the vertebrae are seen from abo\e and somewhat 
to the left side. Eleven caudals and part of a twelfth are 
present. The neural arches, still well developed on the posterior 
sacrals, decrease in height and become more slender as we 
continue along the caudal series; the transverse processes, as 
preser\^ed and seen on the left, likewise decrease in length 
posteriorly. Length of centra, about 12 mm in the sacrals. 
decreases to about 8 mm on the last caudal preser\'ed. On the 
ob\'erse side are seen the rounded \entral surfaces of some of 
the vertebrae in the series. Beneath the articulated caudal series 
are obscure remains of additional caudal \ertebrae. On MCZ 
4002 is found a series of seven articulated caudals, five of which 
are well preserved. The centra have lengths of 8 mm each; they 
are apparentlv somewhat compressed vertically. The height of 
the centra is al^out 5 mm; the neural spines are very short, and 
as preserved there is little trace of any transverse process. It 
seems certain that we are dealing with elements distal to the 



1973 



CYNODOXT POSTCRANIAL MATERIALS 



26 25 24 23 22 21 20 19 18 17 16 15 14 13 12 II 10 9 8 7 6 



4 



Figure 5. Lengths of right ribs of Probclcsodon lewisi, MCZ 3781, 
from presumed \ertebra 6 to 26. Measurements are direct between the two 
ends of the parts preserved; rib lieads are complete only on ribs 14 and 15. 
X 2/3 natural size. 



tail "stump" seen on the slab, which strongly suggest the presence 
of a fairly long tail. 

No interpretable remains of anterior cersical ribs are pre- 
served. Of ribs of the left side, a few dorsals are present in 
isolated fashion on the reverse surface of the slab. Posteriorly, 
a few stumps of left rib are present in the posterior dorsal- 
lumbar series; these have a broad head but rapidly decrease 
to the base of a slender shaft. There is no indication of the 
interlocking expansions seen in various other cynodonts. 

The ribs of the right side are better preserved. Onh- on two 
ribs, which we interpret as presacrals 14 and 15, are the heads 
preserved; these are expanded proximalh' for capitular and 
tul^ercular attachments to the \ertebrae, but taper rapidly dis- 
tally, with no trace of proximal overlapping expansions. A 
proximal series of right ribs, commencing with that which we 
assume pertained to vertebra 6, are seen on both obverse and 
reverse surfaces posteriorly. The ribs attributable to \crtebrae 
in the posterior presacral series are seen on the reverse .surface, 
their heads concealed beneath the vertebrae. In Figure 5 we 
have plotted the lengths of ribs as preserved (measured directly 
between the two ends ) . Although we cannot be sure how much 



10 BREVIORA No. 407 

has been lost proximally or distally in most cases, we see a rea- 
sonable distribution in lengths, with an increase to a maximum 
in the anterior dorsal region followed by a gradual decrease as 
we enter the lumbar region and approach the sacrum. 

On MCZ 4002 a number of dorsal ribs are preser\'ed in their 
entirety, including the heads; they are similar in nature to 
numbers 14 and 15 on the slab and appear to be mid-dorsals 
of both sides with average lengths of about 10 cm. Although 
the usually imperfect nature of trans\'erse processes makes diffi- 
cult the interpretation of the orientation of ribs to the back- 
bone, the rib structure strongly suggests a broad, barrel-shaped 
abdomen. 

The lumbar ribs as preserved are slender (except for the 
presumed first presacral) and although the heads are concealed 
beneath the centra, there obviously was none of the interlocking 
found in the lumbar series of certain other cynodonts. 

In the sacral region some seven ribs on the left side show 
distal expansion (the first broken off from its \'ertebra). As 
the length of the associated right ilium indicates, fixe members 
of this series are sacrals; but which five? Possibly incorrectly, 
we assume that the first (whose vertebra is disarticulated from 
the sacral-postsacral series) is the last presacral (and we have 
used this assumption in making a count of presacral vertebrae). 
The first four, particularly, of the assumed sacral series are 
expanded at their distal surfaces for iliac articulation ; two sacrals 
are visible in ventral view, which better shows this prominent 
distal expansion. A slender rod of bone, extending fonvard, is 
attached without evidence of sutural separation to the outer end 
of the presumed fourth left sacral; this appearance, however, is 
probably due to the peculiar type of preser\'ation in this speci- 
men, and this rod is perhaps a fragment of the left ilium. 

Limbs a7id girdles. Remains of the shoulder girdle are present 
in MCZ 3781. The interclavicle is present, and on the obverse 
are the lower ends of the two clavicles. The interclavicle as 
preserv'ed (slightly incomplete anteriorly) measures 57 mm in 
length. A better preserved interclavicle (slightly incomplete dis- 
tally) is present in MCZ 4002 (Fig. 6b). This is of similar 
size, with a length of 59 mm, a width of 33 mm across the 
head, and, following a constriction, to a width of 1 1 mm back 
of the head, a modest increase in width distally. The structure 
is that typical of cynodonts in general. The anterior "quad- 
rants" for clavicular reception are deeply incised; a median 
keel is present between these quadrants but there is little de- 



1973 



CYNODOXT POSTCRANIAL MATERIALS 



11 




Figure 6. Probelesodon lewisi, a, left clavicle in lateral view; b, inter- 
clavicle in ventral view; c, ulna; d, radius of right side seen from extensor 
surface. All from MCZ 4002. X 1. 



\-elopment of a median ridge more posteriorly. The left cla\icle 
is well preser\"ed in MCZ 4002 (Fig. 6a). It is of typical 
cynodont build. Its length, measured directly between the ends, 
is 51 mm. The relati\ely thin shaft curves about 45° between 
proximal and distal ends. The dorsal end is somewhat expanded 
and excavated medially for acromial apposition. A well-defined 
ridge descends the posterior margin of the shaft and continues 
on to the posterior margin of the \-entral expansion for articu- 
lation with the interclaxicle. 

Of the primary girdle in MCZ 3781, an obscure fragment 
of the left scapula is present; the girdle of the right side is 
present, but is poorly preserved and incompletely seen (Fig. 7). 
On the re\erse side is visible the glenoid region and incomplete 
coracoids, in poor condition: on the ob\'erse is the medial sur- 
face of the scapular blade. This is, as in other chiniquodonts, 
imusually tall and narrow, with a length (probably incomplete) 
from the glenoid articulation with the coracoid of 55 mm, and 
a width at mid-height of 1 1 mi-i. The exposed medial surface 
of the blade is strongly convex in the transverse plane and 
obviously the external surface was strongly ridged longitudinally 
on both front and back margins. In MCZ 4002 a scapular 
blade was present, but is represented bv little but an imoression 



12 



BREVIORA 



No. 407 




Figure 7. Probelesodon lewisi, right scapula and coracoids; the anterior 
coracoid poorly jsreserved, the acromial region of the scapula incomplete. 
From MCZ 3781. X 1. 



indicating that it was elongate and slender. Both coracoids are 
present, the posterior element complete and well preserved; the 
anterior coracoid is incomplete. In MCZ 3801 there is a 
scapular blade of Probelesodon type, narrow and deeply con- 
cave, externally with strong ridges posteriorly and (especially) 
anteriorly. As preserved, the blade is 48 mm tall, 12 mm wide 
at the summit, with the width decreasing to 6 mm below. 
Unfortunately, the scapula is incomplete ventrally, but there 
was obviously little acromial dexelopment and no indication of 
any beginning of a supraspinous surface. 

Both humeri are present in MCZ 3781, and are essentially 
complete although somewhat crushed, and a good left humerus 
is present in MCZ 4002 (Fig. 8). The bone in general corre- 
sponds well with that of typical cynodonts, as described in detail 
by Jenkins (1971). Lengths of the humeri of MCZ 3781 as 
preserved are 72 and 66 mm, that of MCZ 4002 68 mm. The 



1973 



CYNODONT POSTCRANIAL MATERIALS 



13 




Figiac 8. Probclesodon lexi'isi. Right humerus iu distal ventral, lateral, 
and distal dorsal views. From MCZ 4002. X 1. 



head of the bone is unusually expanded anterodorsally, so that 
the articular area extends into a "lip"' overhanging the antero- 
proxinial corner of the dorsal surface. The deltopectoral crest 
is somewhat more expanded than in typical cynodonts. The 
distal end of the bone is unusually broad; its width in MCZ 
4002 is 43 mm, thus being about 63 percent of the length, in 
contrast with lower figures in most cvnodonts. This increased 
width appears to be associated with a greater expansion than 
usual of the entepicondyle. On the ventral distal surface the 
areas for radial and ulnar articulations are well ossified in MCZ 
4002. 

In MCZ 3781 the right radius and ulna and left ulna and 
partial radius are present, but poorly preserved. In MCZ 4002 
the right radius and ulna are well preserved (Fig. 6, c, d). In 
this specimen the radius has a length of 53 mm, the ulna 56 mm. 
Both bones conform well to typical cynodont structure. The 
radius is essentiallv columnar in shape, but as preserved some- 
what flattened between extensor and flexor surfaces. Proximally 
the l-)one is expanded and deeply cupped for humeral articula- 
tion. Distally the width of the bone is increased toward the 
medial margin and thickened on the flexor surface to make 



14 



BREVIORA 



No. 407 




Figure 9. Probelesodon leivisi, right ilium and ischium, composite, MCZ 
4002 and MCZ 3781. X 1. 



possible a broadly oval distal surface for carpal articulation. As 
in Thrinaxodon, a ridge, perhaps for biceps attachment, is 
present proximally on the flexor surface; a flattened area at 
its summit was presumably for apposition with the ulna. A 
ridge with a rounded summit descends the distal third of the 
medial margin, presumably marking the boundary between 
extensor and flexor muscle areas. 

The ulna has the typical cynodont pattern. As usual, the 
olecranon is unossified except for its base, the sigmoid notch, 
represented only by its basal portion. At the expanded dorsal 
end of the bone the medial area is somewhat concave for radial 
apposition; lateral to this area a pronounced ridge is present 
proximally, fading out rapidly below. The proximal part of the 
flexor surface is markedly concave, bounded externally l^y a 
distinct ridge. Below, the shaft of the bone narrows, to be only 
moderately expanded distally for carpal articulation. Deep to 
the trans\'erse plane of the shaft, however, a very strong ulnar 
crest is present, as in other cynodonts; this is conspicuously 
de\'eloped along the distal half of the bone. 

Remains of the manus of both limbs are present in MCZ 3781, 
but in poor condition. Little certain data can be made out 
concerning the carpus, except for the presence of three promi- 



1973 CYNODONT POSTCRANIAL MATERIALS 15 

nent distal carpals. All metacarpals are present on the left 
side, four on the right. Exact measurements are meaningless, 
owing to the imperfect nature of the material, but the a\erage 
length of metacarpals II-V is about 18 mm. A limited number 
of poorly preserved phalanges are present but in no case is there 
a complete digit. The digits are, however, sufficiently preser\'ed 
proximally to show that the vestigial "extra" phalanges present 
in primiti\e cynodonts were absent. 

In MCZ 3781 the incomplete upper margin of the left ilium 
is seen on the reverse side of the slab; the vertically oriented 
complete right ilium is seen on the margin of the slab, and a 
nearly complete left ilium is present in MCZ 4002 (Fig. 9). 
The iliac length is about 70 mm. For the most part the struc- 
ture is typical of that of cynodonts generally, but the posterior 
part of the blade is much more slender than in most cynodonts, 
suggesting that Prohelesodon is ad\'ancing toward the mamma- 
lian condition of forward migration of the gluteal musculature. 
The right ischium is present in MCZ 3781, and has a normal 
cvnodont form. Except for the area of iliac articulation of the 
right pubis, there are no identifiable remains of the pubis in 
available m.aterial. 

Hind limb material is poorly preserved in MCZ 3781. Only 
the proximal part of the left femur is present and the right femur 
(with a probable length of about 80 mm) is badly crushed and 
liT-oken. 0'"lv fragments are present of the left tibia and fibula. 
The rioht tibia and fibula are present but in such poor condition 
that little can be said exceot that they appear to conform 
generally to the usual cvnodont pattern. .A.s preserved their 
lengths are 74 mm and 63 mm (the fibula is incomplete 
proximally). 

In the melange of material in field number 188 (MCZ 3801 ) 
are t^vo well-preserved cynodont femora which seem quite 
surelv to belong to Probelesodon leivisi (Fig. 10). With a 
length of 63 mm each, they represent an animal smaller than 
MCZ 3781, but they are too large to pertain to Probelesodon 
minor or Probainognnthiis, and they differ in structure from 
the gomphodonts in the fauna. In nearlv all respects these 
femora are closelv comparable to the "PCynognathus" femur 
illustrated by Jenkins (1971, fig. 48) — even to the presence, 
part-way down the shaft, of a groove of vmknown nature. These 
femora, however, difTer markedly from all hitherto described 
cvnodont femora in the position of the greater trochanter. In 
all cvnodonts hitherto described the orcater trochanter lies some 



16 



BREVIORA 



No. 407 




Figure 10. Probelesodon lewisi, left femur in dorsal, ventral and medial 
views. From MCZ 3801. X 1. 



distance down the posterior margin of the femur and faces as 
much laterally as proximally. In this specimen it is placed 
definitely farther toward the proximal end of the bone and its 
somewhat expanded tip faces nearly directly proximally. We 
have here a position splitting the difference between typical 
cynodonts and the mammalian condition. A femur of the chini- 
quodontid Chiniquodon described by Romer (1969, fig. 9C) 
shows a somewhat comparable proximal movement of the 
greater trochanter, but the specimen is not too well preserved. 
The feet are poorly preserxed. Of the left foot, a few dis- 
articulated metatarsals, phalanges, and tarsal remains are seen 
on the reverse side. The right foot is nearly complete but 
difficult to interpret. Astragalus and calcaneum can be made 
out on the obverse side. Two complete and two incomplete 
metatarsals are present; the two complete — perhaps the third 
and fourth — ^ are 17 mm in length as preserved. Although too 
much reliance should not be placed on measurements of such 
poor material, it would appear that here, in contrast to certain 
other cynodonts, front and hind feet were of approximately 
equal size. A number of phalanges are present, but except for 



1973 



CYNODONT POSTCRANIAL MATERIALS 



17 




u 

a, 



18 BREVIORA No. 407 

the certain presence of two phalanges on supposed digit IV 
(and the obvious absence of "vestigial" phalanges), we cannot 
be sure of their arrangement. 

Restoration. On Figure 11 we have attempted a restoration 
of the skeleton of Probelesodon. We believe that this restoration 
gives correctly the general appearance of the animal, although, 
as noted in the description above, knowledge is lacking in a 
number of regards, such as cervical ribs, the distal portion of 
the tail, the pubis, and the distal phalanges; even the probable 
presacral count of 27 \ertebrae is uncertain. The body was 
surely stockily built, and although we have included in the tail 
only those \'ertebrae in MCZ 3781, the evidence indicates that 
Probelesodon (and not unlikely other cynodonts as well) had a 
tail of rather good length. A notable departure from the situa- 
tion in many cynodonts is the almost complete absence of the 
overlapping of the ribs, seen at its height in Thrinaxodon. 
This type of structure has been discussed by Jenkins (1971: 
76ff ) ; unless we assume that such a late cynodont as Probele- 
sodon has evolved independently from the very base of the 
cynodont stock, the situation, as Jenkins believes, is one in which 
there has been a return to "normal" rib structure. 

Presumably such cynodonts actively pursued their prey, but 
their limbs would seem to ha\e been relatively inefficient for 
speedy locomotion as compared with later mammals or con- 
temporary archosaurs (although an improvement over the 
pelycosaur condition). As in early tetrapods generally, the hind 
legs were longer than the fore, the combined length of femur 
plus tibia being about 30 percent greater than that of humerus 
plus radius. Again, as in primitive forms generally, the podials 
are longer than the epipodials; the humerus is about 30 percent 
longer than the radius, the femur about 15 percent longer than 
the tibia. 

Probelesodon may be compared with a cariid, Vulpes, of 
siinilar size, and with presumably somewhat comparable habits. 
We may use as a crude iDase for comparison the length of dorsal 
v'ertebrae (approximately the same length in Vulpes and Pro- 
belesodon). To give a rough estimate of comparative limb 
length we may use the length of humerus and radius for the 
front leg, of femur and tibia for the hind leg. In Vulpes this 
front leg measurement is 1 9 times the length of a dorsal vertebra, 
in Probelesodon, little longer than the len<yth of nine vertebrae. 
The hind leg of Vulpes is 21 vertebrae long, of Probelesodon 
] 2 \'ertebrae. It is immediately obvious that for speedy loco- 



1973 



CYNODOXT POSTCRANIAL MATERIALS 



19 





Figure 12. One surface of a block containing material of the shoulder 
region and front leg presumably belonging to Probelesodon minor. On this 
surface are seen the clavicle and interclavicle, most of the primary girdles, 
and part of the left humerus and ulna. The left scapulocoracoid is seen 
in external view, extending dorsaliy as far as the acromion. The right 
scapulocoracoid is seen in posterior view. MCZ 4164. X I. 




Figure 13. The opposite side of the block seen in Fig. 12, showing the 
left humerus, ulna, incomplete radius and part of the primary girdle. X 1. 



20 



BREVIORA 



No. 407 





Figure 14. Probelesodon minor, a, external view of the right scapulo- 
coracoid; b, ventral view of the left humerus. MCZ 4164. X 1. 



motion, the limbs of Vulpes are far superior to those of Pro- 
belesodon. The contrast is still strong-er if it is noted that Pro- 
belesodon had not completed the shift from the sprawled 
pelycosaurian limb position to the fore-and-aft mammalian pat- 
tern. Carnivorous cynodonts are still moderately abundant in 
the early Middle Triassic Chanares beds, but become rarer in 
the formations succeeding this horizon; meanwhile there are 
appearing various thecodonts of a seemingly more sprightly 
nature, and it is not to be wondered at that carnivorous cyno- 
donts were reduced in competition with these rivals. 

POSTCRANIAL MATERIAL OF 

Probelesodon minor 

There is no positi\e association of postcranial materials with 
skulls of Probelesodon minor. However, in field number 40, 
a nodule containing a "mixed grill" of material, there is present 
the pectoral region of a small cynodont (MCZ 4164) which is 
of a size appropriate for this form and in which the scapula 
is of the unusually tall slender type seen in Probelesodon lewisi. 
It is highlv probable that MCZ 4164 pertains to P. minor 
(Figs. 12, '13, 14). 

On one surface of the block there is preserved an articulated 
pair of clavicles and the interclavicle. The cla\'icles, about 
28 mm in length, are gently curved \entromedially from the 
presumed acromial articulation. The shaft is slender; there is 



1973 CYNODONT POSTCRANIAL MATERIALS 21 

a slight expansion proximalh at the acromial articulation, and 
a greater distal expansion, so that the two cla\icles cover the 
anterior quadrants of the head of the interclavicle. The shaft 
exhibits a marked posterior ''twist" below the acromial region. 
The intercla\'icle, sHghtly imperfect posteriorly, has a length of 
3 1 mm. It has a typical cynodont shape — relatively broad and 
flat, with an anterior expansion for reception of the clavicles, a 
modest narrowing behind the head, and a distal expansion. The 
longitudinal ridge is little de\eloped. 

Both primary shoulder girdles are present. As in Probelcsodon 
lewisi, the scapula is tall and very slender, with a height from 
the glenoid fossa of 38 mm, and a width at half-height of but 
4.5 mm. The upper end widens to about 8 mm ; below, it widens 
again to a distinctly developed acromial process (present on one 
scapula, broken ofT on the other). The scapula is cur\'ed longi- 
tudinally to about 45° from summit to glenoid fossa. It is strongly 
concave externally in cross section, with both anterior and 
posterior borders outtumed — the former more strongly so. 
There is no development of a supraspinous fossa. At its base 
the posterior margin thickens to present a well-developed upper 
portion of the glenoid ca\'ity, which appears to have faced 
nearly directly ventrally in life. The anteroposterior breadth 
of the scapula increases ventrallv to give a broad area of articu- 
lation with the two coracoids. Sutures between the scapula and 
the coracoids are not clear. The posterior coracoid is incomplete 
on the left girdle; as preserved on the right side, it is of normal 
cynodont type. The posterior coracoid bears the lower half of 
the glenoid; it appears to extend to a relatively greater distance 
posterolaterally than in typical cynodonts, with a V-shaped 
surface, somewhat convex in cross section. The anterior cora- 
coid is of unusual shape, in apparent contrast to that of P. 
leivisi. It attaches to the anterior edge of the scapula, and its 
upper margin cur\'es strongly forward from this point, rein- 
forced by a thick strut of bone internally. The anterior margin 
of the bone is broad and essentially a straight dorsoventral line, 
sharply set off from the upper and lower margins. The pro- 
coracoid foramen is large. A notch on the ventral margin indi- 
cates the point of di\-ision betw-een anterior and posterior 
elements. 

Both humeri are present, and radius and ulna are preser\^ed 
on the right side. These elements conform to the general cvno- 
dont pattern, and differ from those in P. lewisi mainly in a 
more slender Iniild cf)rrclatcd with smaller size. Humeral 



22 BREVIORA No. 407 

lengths are 37 and 35 mm; proximal widths 17 mm and 
16 mm; distal widths 19 mm. As in P. lewisi the deltopectoral 
crest is highly developed. The slender radius is incomplete 
distally; its length as preserved is 25 mm, the width of the 
head 6 mm. The ulna is 30 mm in length, its width proximally 
8 mm, distally 5 mm. As usual in cynodonts, the olecranon is 
unossified. 

The block of material in which this shoulder region is im- 
bedded includes considerable further bony material; it is, how- 
ever, quite uncertain that any of it belongs to this same form, 
and some of it certainly does not. 

POSTCRANIAL MATERIALS OF 

Probainognathus jenseni 
Because of the advanced nature of Probainoonathus (at least 



to' 



in the type of jaw articulation) a description of the postcranial 
skeleton of this genus would be of considerable value. Regret- 
tably, little can be done. In no case is postcranial material 
articulated with a Probainognathus skuU. In a number of con- 
cretions containing skull materials of this genus there is present 
cynodont postcranial remains that may pertain to this genus. 
But since such concretions usually include a melange of ma- 
terials of different forms, attribution of specific elements to 
Probainognathus is in general highly questionable. 

In one instance, however, pectoral and pelvic assemblages 
can be reasonably assigned to Probainognathus (Figs. 15, 16, 
17). As mentioned earlier, Chahares field number 40 is a 
concretion including some remains of Alassetognathus, some 
thecodont elements and a shoulder region described above as 
probably belonging to Probelesodon minor. Still further, how- 
ever, it includes (MCZ 4021) a shoulder region of a small 
cvnodont definitelv differing: from that of Probelesodon minor 
and appropriate in size for Probainognathus, and, nearby, a 
pelvic region also of appropriate dimensions. 

The scapula of this specimen is quite different from that of 
Probelesodon. being relatively short and broad. Its height is 
28 mm, the distal width 10 mm. The external surface is con- 
cave in cross section but anterior and posterior borders are not 
as sharply outturned as in Probelesodon. There is a well- 
developed acromion. The ventral anterior portion of the bone 
is not completely preserved, but there is a welI-de\'eloped area 
extending forward just below the acromion, suggesting the 



1973 



CYNODONT POSTCRANIAL MATERIALS 



23 




Figure 15. Shoulder girdle and front limb material of a specimen prob- 
ably pertaining to ProbainognatJius jenseni. MCZ 4021. X 1. 




Figure 16. Pelvic region and hind leg material as seen on a block per- 
haps pertaining to Probainognathus jenseni. MCZ 4201. X 1- 



24 BREVIORA No. 407 




Figure 17. Left femur and acetabular region of the pelvis as seen on 
the opposite surface of the block illustrated in Fig. 16. X 1. 



initiation of a supraspinous fossa. The posterior coracoid (seen 
from the upper surface) is present, extending 19 mm back from 
its articulation with the anterior coracoid. This latter element, 
partially concealed within the block, extends well forward, with 
an anteroposterior length of about 13 mm. The coracoid fora- 
men is unusually large, but this may be due to accidents of 
preservation or over-preparation. Well-preserved clavicles and 
a nearly complete interclavicle are also present, the three articu- 
lated. These dermal elements are of normal cynodont structure. 
The interclavicle, as preserved, is 26 mm long and is relatively 
shorter than in Probelesodon minor. 

The left humerus is complete with a length of 49 mm. It is 
unusually long and slender, the length being nearly half again 
that of Probelesodon minor, an animal of roughly similar size. 
The entepicondylar foramen, as prepared, is of unusually large 
size. Both radius and ulna of the left side are complete, with 
lengths of 35 mm and 34 mm, respectively. As usual, there is 
little ossification of the olecranon. 

Various other limbs and axial skeletal elements are present 
in the block close to this girdle and limb assemblage, but there 
is no guarantee that any of them pertain to the same animal. 

Not far removed from this shoulder assemblage are articu- 
lated remains of the pelvic region and hind legs of a small 
cynodont. There is here even less guarantee that this material 
is assignable to Probainognathus than in the case of the shoulder 
region, but the elements present definitely differ from those of 
the Chanares gomphodonts, and to some degree differ from 



1973 CYNODONT POSTCRANIAL MATERIALS 25 

those of Probelesodon; hence they quite probably pertain to the 
present genus. 

An articulated series of \-ertebrae includes the last presacral, 
five sacrals and four proximal caudals. The transxerse processes 
and ribs of the last presacral are de\eloped in much the same 
fashion as the sacrals, and are somewhat expanded distally, 
but show no evidence of connection with adjacent elements. The 
first sacral is ob\'iously the strongest of the series and most 
expanded distally, the further sacrals decreasing in strength 
posteriori)-. 

The left and right ilia have lengths as preserved of 37 mm 
and 42 mm respectively. Unfortunately, their upper margins 
are imperfect, so that the possible contrast between the expanded 
anterior portions and narrower posterior portions of the ilia 
cannot be clearly made out. The acetabulum is deeply exca\ated 
and the supra-acetabular buttress is strongly developed. There 
are no remains of pubis and ischium other than the areas 
immediately adjacent to the acetabulum. 

The right femur is nearly completely preserved; the left (seen 
mainly on the under surface of the block) is incomplete distally. 
The length of the right femur (slightly imperfect at its head) 
is 58 mm; the bone is quite slender, the shaft width at mid- 
length being only about 4 mm. The head is well seen on the 
left femur on the under surface of the block. As in the femur 
attributed to Probelesodon the greater trochanter is somewhat 
more proximally placed than in most cynodonts, but is not as 
thickened nor as laterally divergent as in that form, nor is there 
any distinct separation of trochanter from the articular surface 
of the head. There is no line of di\ision of the intertrochanteric 
fossa from the more distal portion of the ventral surface. The 
lesser trochanter is highly developed and extends far down the 
shaft. 

The right fibula, with an apparent length of 47 mm, is repre- 
sented for much of its length by an impression only. The slender 
tibia has a length of 51 mm, with a width at mid-length of but 
4 mm. As preserved the shaft is strongly concave in transverse 
section. Astragalus and calcaneum are imperfectly preserved. 

References Cited 

Bonaparte, J. F. 1963. Descripcion del csquelcto postcraneano dc E\aere- 

todon. Acta Geol. Lilloana, 4: 1-52. 
Gregory, ^\'. K., and C. L. Camt. 1918. Studies in comparative myology 

and osteology. No. III. Bull. Aiiier. Mus. Nat. Hist., 38: 447-563. 



26 BREvioRA No. 407 

HuENE, F. V. 1944. Die fossilen Reptilicn des siidamerikanischen Gondwa- 

nalandes. Munich: C. H. Beck'sclie Verlags. 332 pp. 
Jenkins, F. A., Jr. 1970. Tlie Chailaies (Argentina) Triassic reptile fauna. 

VII. The postcranial skeleton of the traversodontid Massetognathus 

pascuali (Therapsida, Cynodontia) . Breviora, No. 352: 1-28. 
1971. The postcranial skeleton of African cynodonts. 

Bull. Peabody Mus. Nat. Hist., Yale Univ., 36: 1-216. 
KONJUKOVA, E. D. 1946. New data on Permocynodon sushkini Woodw., a 

cynodont member of the Northern Dvina fauna. Dokl. Akad. Nauk 

URSS, 54: 527-530. 
RoMER, A. S. 1922. The comparison of mammalian and reptilian cora- 

coids. Anat. Rec, 24: 39-47. 
1969. The Brazilian Triassic cynodont reptiles Belesodon 

and Chiniquodon. Breviora, No. 332: 1-16. 
Watson, D. M. S. 1917. The evolution of the tetrapod shoulder girdle 

and fore-limb. Jour. Anat. Physiol. London, 52: 1-63. 



.y:r. 



■..+v^' 



roS' 



B R E V I 0-R-A 

iiseuni of Comparative Zoology 

us ISSN 0006-9698 



Cambridge, Mass. September 20, 1973 Number 408 

A TAXONOMIC COMPARISON OF THE 

AMERICAN UPOGEBIA 

(DECAPODA, TH ALASSINIDEA) , 

INCLUDING TWO NEW SPECIES 

FROM THE CARIBBEAN^ 

David Thistle" 

Abstract. This study examines morpholog-ical variability as a 
means of establishing taxonomically useful characters of American 
Upogebia. The variability of measured characters was analyzed by 
regression; the variability of meristic characters was considered in 
a nonstatistical manner for U. affinis and U. omissa. The analyses 
make it possible to further differentiate these two species. Relatively 
invariant and therefore useful characters were combined with char- 
acters from the literature to delineate the known species. Three 
Eastern Pacific-Western Atlantic species-pairs are indicated and inter- 
preted as being- the result of speciation by geographic isolation caused 
by the closing of the Central American seaway. Two new species, 
U. jamaicensis and U. annae, are described. Ujiogebia rostrospinosa 
Bott is redescribed and figured. 

Tntroductiox 

Twelve species of the burrowing mud shrimp genus Upogebia 
are known from North and South America. Two species, U. 
operculaia and U. rugosa, are morphologically distinct: the 
remaining ten species are \er\- similar. A study of the variability 
of characters in two sympatric species, U. omissa and U . affinis, 
was made to find characters of low \ariability which might be 
suitable for distinguishing among the ten species. 

Upogebia ha\e been found from mean low water to 229 m 
depth. Thev occur most often in mud flats but are known from 

"This study was submitted as a senior thesis at Harvard College. 
•Department of Biology, Harvard University, and Scripps Institu- 
tion of Oceanography, La Jolla, California 92037. 



2 / BREVIORA No. 408 

coarser substrates. Their burrow openings are marked by 
mounds of material remo\'ed during excavation. The animal 
digs by using the third maxillipeds and carries the particles to 
the burrow opening with pereopods 1 and 2 (Stevens, 1928: 
346 ) . Burrows are Y-shaped or may be more complex warrens. 
The animal feeds by creating currents in the burrow by fanning 
its pleopods. Food particles are removed by a basket of setae 
on the inner surfaces of the anterior pairs of pereopods. Bur- 
rows contain several individuals. In North Carolina, Peai^se 
( 1 945 : 305 ) repeatedly found egg-bearing females and juve- 
niles in the same burrow. 

Twelve species of this genus are known from the Americas: 
Upogebia affinis from Massachusetts to southern Brazil; U. an- 
nae n. sp., U. jamaicensis n. sp. and U. operculata from the 
Caribbean; U. o?nissa from Panama and Brazil; U. noronhensis 
and U. brasiliensis from Brazil; U. pugettensis from Alaska to 
Lower California; U. rugosa, U. rostrospinosa and U. longi- 
pollex from the west coast of Central America; and U. spinigera 
from the west coast of Nicaragua to Columbia. 

To provide a quantitative estimate of the variability expect- 
able in this group of similar species, I examined in detail two 
species, Upogebia affinis and U . omissa, using characters selected 
from the literature and from ni)- own preliminary survey. On 
the basis of this examination, it was possible to clarify the dis- 
tinctness of these two species, whose morphological similarity 
could have been a source of confusion. Also, by assuming that 
characters useful in separating U. affinis and U. omissa were 
likely to be useful in separating other related species, I con- 
structed a diagnostic matrix comparing the members of the 
species-group. This matrix of characters made apparent the 
close morphological similarity of two Pacific-Atlantic species- 
pairs, U. rostrospinosa and U. omissa, U. spinigera and U. 
noronhensis. Upogebia rugosa and U. operculata are distinct 
from the other American species of Upogebia and were not 
analyzed in detail but they apparently fomi a third species-pair. 
The occurrence of these pairs of species appears to be the result 
of the separation of populations by the closing of the Central 
American seaway and subsequent differentiation of the isolated 
segments of each original population. 

In the course of this study, two new species were recognized 
{Upogebia annae and U. jamaicensis) and were analyzed with 
those previously known. Their descriptions as well as a rede- 
scription of U. rostrospinosa Bott are given as an appendix 



1973 AMERICAN UPOGEBIA 3 

along with a dichotonious key to the American members of the 
genus. The s)non\niy of U. sturgisae Boone with U . spmigera 
and of U. calif ornica (Stimson) with U. pugettensis after Hol- 
thuis (1952: 3) and Stevens (1928: 318) respectively is fol- 
lowed. 

Materials and Methods 

This study is based on alcohol-preserved museum collections 
of Upogebia a /finis (Say, 1818) and U. omissa Correa, 1968 
(see Table 1). Material was obtained from the following 
sources: Dr. H. W. Levi, Museum of Comparative Zoology, 
Har\'ard University; Mr. H. B. Roberts, United States National 
Museum; Dr. L. B. Holthuis, Rijksmuseum van Natuurlijke 
Historic, Leiden; Dr. Thomas Biffar, Old Dominion L'niversity, 
Norfolk, \^irginia; and Dr. A. L. Castro, Museu Nacional, Rio 
de Janeiro. Dr. Richard Bott, Senckenberg Museum, loaned to 
me four paratypes of Upogebia rostrospinosa. I would like to 
express my thanks to these gentlemen for their kind cooperation. 

Specimens were examined with the use of a dissecting micro- 
scope. Drawings were made with a camera lucida. Overall 
length was measured from the tip of the rostrum to the posterior 
edge of the telson by rotation of the specimen in a clear dish 
along a rule. This method is accurate to d= 2 mm. Other 
measurements were made with the use of an ocular grid cali- 
brated with a stage micrometer. Table 2 summarizes the char- 
acters used and gives the manner in which they will be referred 
to in the text. Figure 1 shows the meaning of these characters 
on diagrams of the animal. The abbrexiations used in the text 
and tables, PI, P2, etc., refer to the first pereopod, second pereo- 
pod, etc. 

The measured characters were analyzed by regression. This 
procedure eliminated the effect of variability introduced bv dif- 
ferences in the sizes of individuals and allowed the setting of 
confidence limits, which permitted statistical comparisons. The 
method used was a nonparametric, graphic procedure which is 
efficient on small, non-normal samples (Tate and Clelland, 1957, 
78-82). In all cases the dependent variable was regressed on 
overall length. Comparisons between species were made by the 
use of 90 percent confidence limits, but since the procedure 
decreases in efficiency with distance from the median, all com- 
parisons were made at the point midway between the x-axis 
medians of the two lines to be compared. In all cases one is 



BREVIORA 



No. 408 



c 2 


•-^ 




























— c 


6 




























range 
ipecim( 


of 
ength 

e (m 


CM 


O 


CO 


CO 




eo 


O 


■* 


C^I 


o 






o 


^-1 1— 


m 


O 


■* 


eo 




eo 


CO 


Tt> 


CO 


^ 






CO 


a 




1 


1 


1 


1 


CO 


1 


1 


1 


1 


1 


en 
csr 


o 


1 


T, 0; 


E « 5 


Tf 


in 


t^ 


Ol 




CI 


Cf 


■* 


t^ 


ift 






to 


"* - S 


en 


•*! 


CM 


eo 




GO 


CO 


CO 


01 


c^x 






CM 


ll 


e< ^ ^ 




























o - 




























rt ^ 


V3 





























3 !n 



XI 
u 



1) 

_bC 

'> 
O 

o 






2 - 

OJ 

o c 

a; 
U <n 
(U <u 

c -S 
." t« 

QJ (U 

2 :^ 



c/:i 






OJ 

S " 

3 
c/5 



gO^tjj^OC^jCs,^^ 



OJ 

c id 






in 

(M 



IT) CO CO O O •— I to 



^ ^ Tf ,^ O CM 05 



Si 

Si 

o 



n CI O O CO 



o .ti 



C .i^ 



OJ «3 

a, 



^ -41 









05 OJ 
Si 



C2 S 



rt 



Si 






O X! 



rt 



S O Si 

s ^ '^ 

3 "t; « 

c/5 be S5 

. C '5 



u •:= 



X! 



be 

o 



a, ^ 



tn 



t^ -1 00 CM CM CM 00 



to ;i 



tj 




Ct! 


_r- 


t« 




CS 


^ 


§ 


o 

;2 


4-t 


^ 


S^ 


^ 


o 


o 


S 


<M 


"u 


3 


^ 





* 



tj .2 



o 
U 



3 
O 
i-l 



E 9- 



TO (/; r3 

i i u 



o 

CM 



to 

h 



"cj 
3 

OJ 

G 



"T, is rt 



o 

a. 



o 
Pi U 






o 
h 



be 
O 



CM CM to CM l> t^ 



to 

CM 



* 














ct! 




* 










B 

s 














Ctj 




pq 










^ 




«i 














^ 




ij 






o 

Nl 




^ 




C3 






C 

u 


Si 

pq 


OJ 

3 
O 

c/2 


pq 

Si" 


(5 
6 






5^ 


U 


OJ 

P3 


5 


QJ 

rt 


M 




w 






H-^ 


d" 




3 
o 

S 


tsj 


1— 


rt 

n 


JJ 


3 

rt 




Si 


s 

r3 







ft, 
O 


"rt 
I 


■J 


[2 


% 


r3 


(5 





1973 AMERICAN UPOGEBIA 5 

Table 2. Characters used in the analysis of Upogebia affinis and U. omissa, 
and indicated by niniiber in Fig. 1. Each character is referred to in the text 
and tables by the words in italics. 

Number Character 

1 Number of ocular spines 

2 Rostral ventral spines 

3 Number of spines on epistome 

4 Number of spines behind the cervical groove 

5 Ventral abdominal spines 

6 Serration of uropod distal edges 

7 Number of uropodal spines 

8 Style of dactylar teeth (PI) 

9 Style of teeth on fixed finger (PI) 

10 Carpal exterior lateral spines (PI) 

11 Number of dorsal palni ridges (PI) 

12 Proximal meral spine (P2) 

13 Width of the rostral base 

14 Rostral length 

15 Length of eye stalk 

16 Length of rostral lateral teeth 

17 Length of sixth abdominal segment 

18 Length of telson 

19 Width of telson distal margin 

20 Width of telson proximal margin 

21 Length of fixed finger (PI) 

22 Length of dactylus (PI) 

23 Length of palm (PI) 

24 Width of palm (PI) 

25 Length of mcrus (PI) 

26 Width of merus (PI) 



BREVIORA 



No. 408 














/\ 






-N 


n 


vV-llS 













Figure 1. Diagrammatic presentation of characters analyzed, numbered 
as in Table 2. Underlined numbers are scores for the dentition patterns 
of fixed finger and dactylus. 



testing the position of the lines at this grand median rather than 
differences in slope (E. W. Fager, personal communication). 

When tested for sexual bias in terms of numbers of indi\iduals, 
none of the collections departed significantly from the null hy- 
pothesis of a binomial distribution (p = q=/4) at the 90 
percent level. Assuming that the sex ratio is 1 : 1 in both species, 
these lots are not significantly biased in terms of sex. 

In order to minimize the effect of ontogenetic changes in 
morphology on interspecific comparisons, this study used the 
overall length of the smallest ovigerous female in each sample 
as a criterion for restricting the analysis to adults. This pro- 
cedure assumed that overall length, age, and maturity were 
highly correlated so that the probability of females longer than 



1973 AMERICAN UPOOEBIA 7 

this niininuini bciii"- adult was lan'e. The overall lene^ths of 
males versus females in each sample were such that a Mann- 
Whitney "U" test (Tate and Clelland, 1957: 89-91) revealed 
no significant difference at the 90 percent le\el. Because the 
samples were not biased in terms of sex (see above) nor were 
the sexes different in oxerall size, it seemed reasonable to extend 
the adult overall length minimum to males. Thus, an adult 
upogebiid was defined as an individual that was larger, and by 
inference older, than the smallest sexually mature female present 
in the sample. In two cases this criterion was set aside for prac- 
tical reasons. The available specimens of Upogebia omissa con- 
tained only two o\"igerous females (36, 44 mm), while the 
lengths of all specimens ranged from 1 9 to 44 mm ( Table 1 ) . 
Correa (1968) reports adults ranging from 27 to 47 mm. Those 
indix'iduals smaller than Correa's minimum were considered 
ju\-eniles, as Corre'a's range of adult o\'erall lengths was based 
on 106 o\'igerous females. The Wellfleet, Massachusetts, collec- 
tion contained no ovigerous females. The closest population of 
U . a/finis in overall length is that from Miami and its minimum 
(28 mm) was used. After the removal of subadults in this man- 
ner, subsamples for analysis were taken at random from samples 
of more than ten indi\iduals. 

Results 

Measured characters. To provide a quantitati\'e estimate of 
the variability within a species, Upogebia affinis and U. oyyiissa 
were analyzed by the regression of 14 measured characters on 
overall length. Each character was tested for sexual dimorphism 
by the comparison of 90 percent confidence limits erected about 
regression lines formed for each sex. For U . omissa none of the 
14 characters differed significantly between sexes. In U. affinis 
fixed finger length (21) and palm width ( 24 ) were significantly 
sexually dimorphic (Figs. 2, 3); the remaining characters were 
not. Interspecific comparisons using the regression lines for each 
sex separately re\ealed no significant difference for either sex on 
any character. Regression fines formed from both sexes still 
showed no significant diflference between species on any char- 
acter, excluding characters 21 and 24. 

The characters measured contain infonnation about the shape 
of much of the animal. The results show the two species to be 
largely indistinguishable in gross morphology, making speculation 
about the origin and niche separation of these two partially 



8 



BREVIORA 



No. 408 



3A 



y 



IJ 




30 



40 



50 



GO 



Overall length [min] 

Figure 2. Regression of the fixed finger length onto overall length for 
males (squares) versus females (circles) of Upogebia affinis. The upper 
triangle is the x-axis median for males; through it passes the best-fit 
median, regression line. The envelope of lighter lines are 90% confidence 
limits. The lower triangle marks the female x-axis median point with a 
similar set of lines. 



sympatric species interesting, but to little purpose until their 
natural history is better known. The analysis does point to prob- 
lems latent in the use of measured characters in this genus. One 
must quantify the variability and examine it comparatively be- 
fore any but the most ob\'ious differences in proportion are given 
taxonomic weight. 

The regression analysis confirmed one feature of taxonomic in- 
terest. Upobegia affinis has conspicuous sexually dimorphic 
chelipeds. In the males the cheliped is consistently more robust, 
larger, and better calcified than in the female. In U. omissa, 
while the males tended to be more variable about the regression 
line reflecting the occasional dimorphic indixidual as reported 
by Correa (1968), there was no significant difference between 



1973 



AMERICAN UPOGEBIA 




y 4 



iji 



41 



SB 



Overall Lenath rmml 



Figure 3. Regression of the width of pahii (PI) onto overall length 
for male (squares) versus female (circles) Upogebia affinis. The upper 
triangle is the x-axis median for males; through it passes the best-fit median, 
regression line. The envelope of lighter lines are 90% confidence limits. 
The lower triangle marks the female x-axis median point witli a similar 
set of lines. 



10 BREVIORA No. 408 

sexes. Upogebia affinis is sexualh' dimorphic in palm width and 
fixed finger length; U. ornissa is not. This dichotomy helps to 
distinguish the species. 

Meristic characters. Despite this similarity of shape, there are 
differences between Upogebia affinis and U. omissa. The diag- 
nostic characters of these two species are differences in orna- 
mentation, as are those which distinguish the other species. 
Table 3 summarizes the results of an examination of 1 2 of these 
characters. In it one can see the type of individual and geo- 
graphic variability present in U. affinis and U. omissa, the fea- 
tures which separate them, and some of those they share. 

In delineating Upogebia affinis from U . omissa, clear disjunc- 
tions are most useful. Upogebia omissa has ventral abdominal 
spines and P4 is armed; U. affinis does not ha\e these spines. 
The other characters that show differences between species are 
less distinct. For a given character, each species has a different 
dominant state, though some individuals of each species exhibit 
the character state of the other species. The greater the fre- 
quency of the inappropriate character state, the less useful the 
character, but because of the obvious effect of the interaction 
between those populations contributing the most specimens to 
the relative frequency of a character state within a species, these 
frequencies were not tested statistically. 

De Man (1927) redescribed Upogebia affinis from a few 
Carolina specimens. His detailed description agrees with my 
material. He did not comment on variability beyond two lo- 
calities. Table 3 can be considered to supplement his description. 
In addition, the rostrum is not always longer than wide; the 
telson is rectangular to wider posteriorly; and the upper surface 
of the telson is not always punctate. De Man refers to reports 
of U. affinis from the "coast of Brazil, Mamanguape stone reef, 
Parahyba river . . ." On reexamination these specimens were 
found to be U. omissa (Table 1 ) . 

Correa (1968) described Upogebia omissa in detail, including 
its variability. Beyond those features already discussed, I found 
the following differences. The eyes are slightly shorter than the 
rostrum. I have examined a female that is 44 mm long versus a 
maximum of 35 mm given by Correa, Also, in the table pro- 
vided by Correa for comparison of Upogebia affinis with U. 
omissa, the distinction based on spines on the lower surface of 
the rostrum is not useful, as U. affinis from Venezuela lacks the 
spines. The protopods of the uropods bear two spines in U. 
affinis from Venezuela rather than one. 



1973 AMERICAN UPOGEBIA 11 

^ T. 2 .'' " •- ^ ^ 

uTIsS-^'^': >r= - I ~^+ ^ + ^ ^ ^ - + 

< - - ~. -^ ^ '-S ^ ^ "-^ ^ 






■ — r" f^ Q 



P '~ )^ O ^ |;., 3^(^^^ln■«t<aO-^-j- 



^ -n :i '' -z - o 



is" o « 






• .M *-^ .M "^ W V3 

- Q - = .i2 ■— 

~ = 2 5 u 

1. • .5 = r, 3 



^(M-|_ir)C^re<^ — ^ 



■z: -6 ^ -^ ^ -^ 



y; ^ 



1. rt 

■1 .2 ■= = .S « ^ ^ 

O O rt 

u ui _: • ^ ^^ '^ ^— N ^^ 






C'-I CM -^ 



o - >- ^ 

2 o y «-- i; ci 15, CO* 

•I I i i I I a ^ ^ - ^1 + ^- 1 or 0. iM '^^ + 

5 3 n ^ o" 3 - -I- " '^ '^'* 



^ y n ^ si ii 

S « - 

5^ — fcT £- = 



-£02 = 



r3 



cr 



SP 3 5 



or o 



^ o « 5 < g *^ S. -H- ' -h' ' ^ Of *^ C-; + 

"-■ « S i- o 

t) « ~ "T" S 



w .. rs ■"■ ?? 



>H s •- i: „ w 

E ~ .2 _- ti ~ 



= ■>-■- b ^ o 



2 y= ■=: 

" C « J u 

- ,^ ii -J - o 



ii^ '^ a ^ - '^ ~L^ '^- r'. -^ u: ^ ?\ 



12 BREVIORA No. 408 

The Species of Upogebia 

On the basis of the analysis of Upogebia affinis and U. omissa, 
I selected characters that had little within-species variability 
while setting off at least one described species from the others. 
These characters, coupled with several useful characters from 
the literature, are presented as a diagnostic matrix showing the 
interrelationships among the species (Table 4). Characters 
based on measurements were not included, because the lack of 
specimens of other American species made it impossible to prop- 
erly evaluate their variability, except that the ratio of PI fixed 
finger length to dactylus length, which is routinely given in the 
literature, was included. 

Table 4 reveals a portion of the interrelationship among the 
members of the genus in the Americas, and can be used to 
distinguish among the species. The information presented is 
from several sources: holotypes of Upogebia annae n. sp. and 
U. jatnaicensis n. sp., paratypes of U. omissa and U. rostro- 
spinosa, museum collections of U. affinis. The original descrip- 
tions of U. noronhensis Fausto-Filho 1969 and U. brasiliensis 
Holthuis 1956 were used. For U. spinigera (Smith 1871) the 
original description was supplemented by Holthuis' (1952) re- 
description. Similarly, for U. pugettensis (Dana 1852) de Man 
(1929) and Stevens (1928) were used. 

Upogebia longipollex was described ver\' incompletely and 
without figures by T. H. Streets (1871) from a Panamanian 
collection of J. McNeil. Lockington (1878) states that the 
material ". . . probably came from the Pacific coast of the 
isthmus." De Man (1928) speculated that U. longipollex might 
be a junior synonym to U. spinigera (Smith) if differences in 
spination of the pereopods were the result of differences of the 
ages of the specimens described. Holthuis (1952) synonymized 
U. longipollex with U. spinigera without comment. 

The results of this study indicate that leg spination, particu- 
larly the P2 meral spine (\'entral proximal spine of merus of 
pereopod 2), is diagnostic at the specific level. Streets, describ- 
ing spination, states, ". . . third article [carpus] . . . armed 
with spine above at distal extremity; remaining pairs [of legs] 
unarmed." U . spinigera has a P2 meral spine as well as spines 
on P3 and P4. It cannot be the same species as U. longipollex. 

Upobegia longipollex has been included in Table 4 as a good 
species and adjacent to U. pugettensis to which it seems to be 
most similar. It appears likely that after an adequate variational 



1973 AMERICAN UPOGEBIA 13 

c ^ ^ Ji -Ho 

. >— ) 3 c « t; <^ 



•a 



a 2 <« 



•^ S 



O t) 



< 

O 



o 



<M 






rt ~. .y 



*-^ S <^ c S . « <^ c^ 



-5 ' u, ^ ^ 






^ (N If) C^ o 

pg" g ? rt 

iir^i „ I s 1-:.:-+ ■...I-++ 1 1-+ 

O O rt - ^ ° 

•?!, c h 2 ^ 

■^ « = rt u ^ 

4j . c C 3 

tjD e P cj ^ y -a 

o >~ o - >, ■-:: "^ I .--s CM ^ -—-«.. I' C4 

§-.!=< s^^ ^ 5 "? -^ " I -. ^^. §<^ + + I I ^- - 

♦^ -^ Lj .Li _I t"^ I ^ ^^ ^^ , 3 ^^ _t 



1^ ^ ^ y < 

"06 „ .. . ^- 



r 



::i^-. :. ^ 7^ \^ >r "^ 






^ ^: 2:i «J ^ U 

«; S: ~ Si <; . E 

rj t^^ ™ ^ w 

V— ;i r* <- 't:i ^ 

= 5 ^ S = &H 

"^ ic i_ -*^ 

• P rt C 

jj o o a. be 

o til n 

.1 5 I 5 '^ 

"O O O i« c 



» I „ .^. ,^ ^ 



^: .n _H 2 =^ I I I I - + 



<^ 10 CM rv. O I I I I r^ <%. 



« '^ • « 9" - 

-,». *^ <M ^ 3 

«j <^ u Si S" 







14 BREVIORA No. 408 

study of U. pugettensis (1852) has been performed, U. longi- 
pollex (1871) will be synonymized under that species, but until 
the Pacific American upogebiids become better known, U. longi- 
pollex should be maintained. 

American upogebiids seem to show examples of speciation by 
geographic isolation. In Table 4 are two cases where a pair of 
morphologically very similar species are separated by Central 
America. Upogebia rostrospinosa from El Salvador is most 
closely related to U. omissa from Panama and Brazil; U. spini- 
gera from the Gulf of Panama to U . noronhensis from northern 
Brazil. Upogebia rugosa from the Gulf of California and U. 
operculata from Barbados are a third example. An interpreta- 
tion of this evidence is that in each of these cases a single 
species existed in tropical waters before the close of the Central 
American seaway. After this event the Atlantic and Pacific 
populations no longer shared a common gene pool and evolved 
separately. Under somewhat different selective pressures each 
population became differentiated while retaining a basically simi- 
lar morphology within a species-pair. Thus the model of specia- 
tion by geographic isolation appears to explain the occurrence 
of species-pairs in Upogebia. 

Conclusions 

A study of the variability within two species of Upogebia 
has shown measured characters to be of uncertain taxonomic 
value. Relatively invariant and taxonomically useful characters 
have been used to help demonstrate the distinctness of ten known 
species. 

It is extremely likely that further collecting will lead to the 
discovery of additional species. Their description as well as a 
fuller understanding of the biology and distribution of each 
species will have to be achieved before one can come to a bio- 
logically real understanding of speciation in this group. 

ACKNOV^LEDGMENTS 

I would like to thank Herbert W. Levi, under whose super- 
vision the project was carried out, and all those who aided me 
in the course of the study. L. B. Holthuis and T. Biffar helped 
formulate the problem. The following people ha\e read and 
criticized the manuscript during its several phases: H. W. Levi, 
T. Biffar, K. Boss, E. E. Williams, S. J. Gould, R. R. Hessler, 



1973 AMERICAN UPOOEBIA 15 

\V. Newman, J. Sepkoski, P. Juniars, and M. BurkenroacL I'he 
Museum of Comparative Zoology pro\"iclcd the facilities for the 
research. 

Portions of this paper were submitted to the Department of 
Biology, Har\ard Unixersity, as partial fulfillment of the require- 
ments for the deo^ree of Bachelor of Arts. 



'O' 



References 

BOTT, R. 1955. Dekapoden (Ci'ustacea) aus El Salvador. Sencken- 
bergiana Bio., 36: 47-72. 

CoRREA, M. M. Gomes. 1968. Sobre as especies de "Upogebia" 
Leach do litoi-al Brasieiro, com descricao de uma especie novo 
(Decapoda, Callianassidae). Revista Brasileira do Biolog-ia, 28: 
97-109. 

Dana, J. D. 1852. Conspectus of the Crustacea of the Exploring 
Expedition under Capt. C. Wilkes, U.S.N. Proc. Acad. Nat. Sci. 
Philadelphia, 6: 10-28. 

Fausto-Filho, J. 1969. Upogebia noronhensis, nova especie de 
Crustaceo do Brasil (Crustacea, Decapoda, Callianassidae). Arq. 
Cien. Mar., 9(1) : 1-7. 

Haswell, W. a. 1881. Description of some new species of Austra- 
lian Decapoda. Proc. Linn. Soc. New South Wales, 6: 750-763. 

Hessler, R. R. 1970. The Desmosomatidae (Isopoda, Asellota) of 
the Gay Head-Burmuda Transect. Bull. Scripps Inst. Oceanog- 
raphy, 15: 1-185. 

Hoi.THUis, L. B. 1952. Report of the Lund University Chile Ex- 
pedition 1948—49. On two species of Crustacea Decapoda Macnira 
from the N.W. coast of South America. Acta Univ. Lund. N.F. 
Avd. II, 47(9): 1-11. 

. 1956. Three species of Crustacea Decapoda Ma- 

crura from Southern Brazil, including a new species of Upogebia. 
Zool. Meded., 34: 173-181. 

LOCKINGTOX, N. W. 1878. Remarks upon the Thalassinidea and 
Astacidea of the Pacific coast of North America. Ann. Mag. Nat. 
Hist., Ser. 5, 2: 209-304. 

Man, J. G. de. 1927. A contribution to the knowledge of twenty- 
one species of the genus Upogebia Leach. Capita Zool., 2(5): 
1-58. 

— ■ . 1928. The Thalassinidae and Callianassidae col- 
lected by the Siboga Expedition with some remarks on the Lao- 
mediidae. Siboga Expedition Reports 39a*, part 7: 1-187. 

. 1929. Papers from Di-. Th. Mortensen's Pacific 



Expedition L. On a small collection of Decapoda . . . Dansk 
Naturhistorisk Forening, Copenhagen. Videns. Meddel., 87: 
105-134. 



16 BREVIORA No. 408 

Pe.\rse, a. S. 1945. Ecology of Upogebia affinis (Say). Ecology, 
26(3): 303-305. 

Say, T. 1818. An Account of the Crustacea of the United States. 
J. Acad. Nat. Sci. Philadelphia, 1(2): 241. 

Smith, S. F. 1871. List of the Crustacea collected by J. A. McNeil 
in Central America. Rep. Peabody Acad. Sci. for 1869: 87-98. 

Stevens, B. A. 1928. Callianassidae from the west coast of North 
America. Pub. Puget Sound Biol. .Sta., 6: 315-369. 

Streets, T. H. 1871. A catalogue of Crustacea from the Isthmus 
of Panama. Pi'oc. Acad. Nat. Sci. Philadelphia, 23: 238-243. 

Tate, M. W., and R. C. Clelland. 1957. Nonparametric and Short- 
cut Statistics. Danville, Ilinois: Interstate. 171 pp. 

Appendix 

Upogebia jamaicensis n. sp. 

Figure 4 

Male holotype from Montego Bay, Jamaica. USNM #41748. 
The species is named for the type locality. Overall length is 
50 mm. 

Diagnosis. Upogebia with 4-5 ocular spines; 8-10 spines 
behind cer\'ical groove laterally; rostral xentral surf?iCe unarmed; 
P2 with strong proximal \entral spine on merus; 1 epistomal 
spine. For relationships to American species see Table 4. 

Upogebia jamaicensis is most closely related morphologically 
to U. spitiifrons (Haswell, 1881) from Austraha. Much of the 
cephalon spination is similar in the two species. Following 
de Man (1927), U. spinifrons has the rostral ventral surface 
armed, the dorsolateral extensions of the carapace without tuber- 
cles, and with 2 epistomal spines. Upogebia jamaicensis has the 
rostral \cntral surface unarmed, tuberculate dorsolateral exten- 
sions, and 1 epistomal spine. There are differences in number 
of PI palm ridges and P2 meral spines. No other described 
species has 4-5 ocular spines. 

Description. A slash separating 2 measurements indicates the 
ratio of the first to the second. For a discussion of the use 
of ratios in taxonomy, see Hessler (1970: 7). length: 40- 
50 mm. cephalothorax : Rostral basal width/rostral length is 
0.6-0.8. Length rostral lateral teeth/rostral length is 0.3-0.5. 
Dorsolateral extensions of carapace with 10-12 spines (be- 
coming spinules posteriorly). Eye length/rostral length is 0.5. 
abdomen: Segmentation typical of genus. Sixth segment width/ 
length is 1.4. Telson width/length is 1.0-1.2. Telson with 



1973 AMERICAN UPOGEBIA 17 

proximal transverse carina, median groove, and wrinkled sur- 
face. PEREOPODs: PI : I'ixcd finger length/ dactylus length is 
0.3-0.7. Dact\lus with larcre tooth latcrallv and lesser tubercles 
distallv; large distal tooth on cutting edge with lesser teeth prox- 
imallv. Fixed finger with 4-6 denticles on cutting edge. Palm 
width/length is 0.4-0.6. Palm o\oid in cross section. Palm 
spination: dorsalh 2 ridges, outer of spines, inner of spinules. 
Exteriolateral surface with 6 spines; interiorly 1 distal spine. 
Carpal spines: 1 large \entrally, 1-2 exteriorly. Dorsally 

1 major spine distally with row of 4-6 behind it; 2 exterior, 
1-2 interior to it. Meral width/length is 0.3-0.5. Meral 
spines: 1 distodorsal spine, 6-7 spines on \-entral margin. 
P2: Carpus with distal spine dorsally and ventrally. Merus 
with distodorsal and proximo\-entral spines. P3 : Merus with 

2 distodorsal spines: ventral margin with 4-6 spines, many 
tubercles. Ischium with 1 spine. P4: Merus of holotype with 
spine on ventral margin, absent in paratypes. pleopods : 
Endite of 2-5 enlarged, squarish, uropods : 1 spine on interior 
protopod, tubercle on exterior protopod; distal edges denticulate. 

In female, width rostral base/rostral length is greater, PI 
dactylus shorter, cheliped less robust. 

Range. Jamaica. Four specimens examined. Features of 
types are: holotype without left P4, right P2, P3. Paratype, 
female, USNM #138897, same locality, left of carapace dam- 
aged. Paratype, female, USNM #138896, same locality, right 
PI missing. 

Upogebia annae n. sp. 

Figure 5 

Female holotvpe: R/\' OREGON, sta. 5421, Bahama Isl., 
lat. 20°54'N,'long. 73°36'\V, 125 fathoms (229 m). USNM 
#138892. The species is named for my wife, Anne. 0\erall 
length is 25 mm. 

Diagnosis. Upogebia with 1, 2 ocular spines, no spines be- 
hind cervical groove, PI fingers of claw equal, PI with no 
ridges on dorsal surface of palm. P2, P3, P4 with elongate 
merus. 

Upogebia annae is most closely related to U. brasiUensis. 
Upogebia annae differs in having no dorsal PI palmar ridges 
and no epistomal spines. In U. brasiUensis the merus width / 
length ratio of P2 is 0.33, of P3 is 0.41, and of P4 is 0.30; in 



18 BREVIORA No. 408 

U. annae that of P2 is 0.20, of P3 is 0.21, and of P4 is 0.22. 
See Table 4 for comparison to other American species. 

Description. A slash separating 2 measurements indicates 
the ratio of the first to the second, length: 32-40 mm. 
CEPHALOTHORAx: Rostral basal width/rostral length is 0.8- 
0.9. Length rostrolateral teeth/rostral length is 0.2. Dorso- 
lateral extensions of carapace with 10-12 spines (becoming 
spinules posteriorly). Eye length/ rostral length is 0.1-0.3. 
abdomen: Segmentation typical of genus. Sixth segment 
width/length is 1.0-1.2. felson width/length is 0.9-0.10. 
Telson with proximal transverse carina, median groove, and 
wrinkled surface, pereopods: PI: Fixed finger length/dactylus 
length is 1.0. Dactylar row of tubercles variable. Fixed 
finger with 4-6 denticles on cutting edge. Palm width/length 
is 0.3-0.5. Palm ovoid in cross section. Carpal spines: dis- 
tally, one each ventrally, exteriorly, dorsally; dorsal spine with 
3-5 above it, 2 interiorly, 2 exteriorly. Meral width/length is 
0.3. Merus with distal dorsal spine, 4-6 spines on ventral margin. 
P2: Carpus with dorsal, ventral distal spines. Merus with 
distal dorsal variable tubercles on ventral margin. P3 : Carpus, 
ventral distal spine, variable. Meral ventral margin with 2-3 
spines, many tubercles, pleopods with o\al endite. uropods 
with small spine on protopod. 

In female, uropodal exopod extends beyond telson. Males 
with robust PI, more highly calcified, uropod edges finely 
denticulate. 

Range. Known only from types. 

Features of types. Holotype discolored in branchial area. 
Male paratype, R/V OREGON sta. #5421, USNM #138893, 
same locality as holotvpe, left P3, right P5 damaged. Male 
paratvpe, R/V SILVER BAY sta. #5158, USNM' #138894, 
Bahama Isl. lat. 19°55.5'N, long. 71°07'W, 100 fathoms 
(183 m), right P4, P5, left P3, P5 missing. 

Upogebia rostrospinosa Bott 1955 

Figure 6 

Female holot) pe from Puerto el Triunfo, El Salvador. Sencken- 
burg Museum #2116. 

This redescription is based on an examination of one female 
paratype. Differences from the original description should be 
noted. No evaluation of variability is possible. 



1973 AMERICAN UPOGEBIA 19 

Diagnosis. Upogebia with 1 ocular spine, P2 meral spine, 
spinules on \-entral surface of first abdominal segment, no spines 
on P4, spines on telson transx'crse ridge. 

Upogebia rostrospinosa is most closely related to U. oniissa 
on the basis of the ventral abdominal spines. U. rostrospinosa 
differs in ha\'ing no spines on P4 while U. omissa lacks the 
spines on the telson. See Table 4 for comparison to other 
American species. 

Description. A slash separating 2 measurements indicates the 
ratio of the first to the second. All measurements refer to the 
paratype. length: 28 mm. cephalothorax: Rostral basal 
width/rostral length is 0.6. Length rostrolateral teeth/rostral 
length is 0.3. Dorsolateral extensions of carapace with 9-10 
spines [becoming spinviles posteriorly). Eye length/rostral 
length is 0.75. abdomen: Segmentation typical of genus. Sixth 
segment width/length is 1.3. Telson width/length is 1.2. Telson 
with proximal, 8-spined transverse ridge, pereopods: PI: 
Fixed finger length/dactylus length is 0.7. Palm width/length 
is 0.53. Palm ovoid in cross section, 3 dorsal rows of hairs, 
1 external. Carpal spines: 1 \cntral distal, 1 exterior distal, 
1 dorsal distal with a row of 4 behind it. Meral width/length 
is 0.5. Merus with 1 distal dorsal spine, 3 on \entral margin. 
Ischium with 1 spine. P2 : Carpus with distal spinule dorsally 
and ventrally. Merus with distal dorsal spine; proximal ventral 
spine. P3 : Merus with 3 spines on ventral margin, 4 spinules 
on exterior proximal surface, pleopods with elongate endite. 

No males known. 

Range. Known only from type locality. 

Remarks. Bott (1955) mentions four paratypes (SMF 
#2117). I have examined these and three do not fit the de- 
scription of the holotype. The discrepancies are in characters 
that I have found to be diagnostic, and I belie\e that these 
three specimens cannot ser\'e as representati\es of Upogebia 
rostrospinosa. 



20 



BREVIORA 



No. 408 




Figure 4. Upogcbia januiicensis n. sp. male holotype: A) habitus. B) 
Al, C) pleopod 2, D) sixth abdominal segment and telson, E) pereopods 
1-5, F) female pleopod 1 (paratype) , G) cephalothoiax and right cheliped, 
H) A2. Hairs and setae omitted. Scale lines equal 2 mm. 



1973 



AMERICAN UPOGEBIA 



21 




Figure 5. Upogebia annae n. sp. female holotype: A) habitus, B) A2, 
C) pereopods 1-5, D) female pleopod 1, E) Al, F) sixth abdominal seg- 
ment and telson, G) pleopod 2, H) cephalothorax and right cheliped. 
Hairs and setae omitted. Scale lines equal 2 nun. 



22 



BREVIORA 



No. 408 




Figure 6. Upogebia rostrospinosa Bott female paratype: A) A2, B) 
habitus, C) female first pleopod, D) Al, E) pleopod 2, F) pereopods 1-5, 
G) cephalothorax and right cheliped, H) sixth abdominal segment and 
telson. Hairs and setae omitted. Scale lines equal 2 mm. 



1973 AMERICAN UPOGEBIA 23 

Key to the Aincrican Species of Upogebia 

1. a. Telson distal margin conspicuously wider than proximal margin .. 2 
b. Telson essentially rectangular 3 

2.*a. Rostrum armed dorsally with "two strong spines"; Caribbean .... 

• l\ operciildid 

b. Rostrum "beset with small tubercules and hirsute" dorsally; Pacific, 
Central America U. rugosa 

3. a. First abdominal segment spined ventrallv 4 

b. First abdominal segment unspined \cntrallv 5 

4. a. P4 with spines (see Table 4 caption) ; Atlantic, Panama to Brazil 

U. om issa 

b. P4 without spines; Pacific, Central .\mcrica .... U. rostrospinosa 

5. a. P2 with proximal meral spine 6 

b. P2 without proximal meral spine 9 



6. a. PI propodus with two ridges dorsally 7 

b. PI propodus with three ridges dorsallv 8 



7. a. Having one or two ocular spines; Atlantic, widespread . . L'. affinis 
b. Having four or five ocular spines: Caribbean U. jamaicensis 

8. a. P4 with spines, P5 unspined; Pacific, Central America to Colombia 

U. spinigera 

b. P4 unspined, P5 with spines; Atlantic, Brazil U. noronhensis 

9. a. P3 with spines 10 

b. P3 without spines 11 

10. a. Epistome spined, P2, P3, P4 with elongate merus; Caribbean 

U. aunae 

b. Epistome unspined; Atlantic. Brazil U. braziliciisis 

11. a. PI propodus with two ridges dorsally; Pacific, Alaska to Lower 

California U. pugettensis 

b. PI propodus with no ridges dorsallv: Pacific, Central America .... 
U. lougipollcs 



*The present distinction between these two species is probably semantic; 
they are badly in need of redescription. 



, .;k „■ 



•>* 



B R 



A 



J 



V I 



'^'^■'S. CCMP. ZOOL^ 




useiinti of Comparative Zoology 

us ISSN 0006-9098 



Cambridge, Mass. September 20, 1973 Number 409 

THE CLASSIFICATION OF THE 
COTINGIDAE (AVES) 

D. W. Snow 



Abstract. The treatment to be given to the Cotingidae in Peters' Check- 
list of Birds of the World is discussed. Both the limits of the family and 
the best ariangement of the genera within it are problematical, and it is 
stressed that the system to be adopted must be regarded as provisional. 

Previous attempts to classify the iamily are reviewed. Nine genera are 
discussed, whose allocation to the Cotingidae has been disputed. Four of 
these {Laniisoina, Plwcnicircus, LilMugiis and Riifncola) are considered 
best retained in tlie family on available evidence, and five (AttHa, Casiomis, 
Lanioccra, Rhytifitcrna and Xoiojjsayis) are considered to belong to the 
Tyrannidae. 

Doliornis and Zaratornis are not considered to be sufficiently tlistinct 
from A>ul)clion to be maintained as scpaiatc genera. Likewise, Platypsaris 
is merged with Pach\raviphtis, and Eratnr with Tityra. 

The sequence of genera adopted indicates a general progression from 
unspccialized types to highly specialized, sexually very dimorphic frugivores; 
but it is emphasized that this may not represent an evolutionary progression. 
Within the genera, there is a marked tendency for closely related forms to 
replace one another geographically. Many of these are too distinct to be 
treated as races of one species; moreover, in two cases where there is slight 
geographic o\erlap the forms concerned appareirtly behave as distinct 
species although they are very similar. To take accoimt of this general 
situation, liberal use is made of Mayr's concept of the zoogeographical 
species. On this basis the family is treated as containing 79 taxonomic 
species and 56 zoogeographical species. 

Before his death in 1955, J. T. Zimmer prepared a rough 
manuscript of the Pipridae, Cotingidae, Tyrannidae, and related 
families, to constitute \()lume 8 of Pcter.'i' Check-list of Birds of 
the IVorld. The section on the Cotingidae was reviewed in 1957 
by James Bond, who made minor modifications, and it was this 
revised manuscript that was the basis of the classification adopted 
by Me)er de Schauensee (1966). Volume 8 of the Check-list, 
however, unfortunately remained unpublished. 



2 BREVIORA No. 409 

Having been invited by Melvin A. Traylor, who has ac- 
cepted responsibility for the final editing of this volume, to 
revise Zimmer's manuscript of the Cotingidae in the light of 
recent work, I here discuss the controversial points in the classi- 
fication of this difficult family at the species level, in particular 
the affinities of the genera incertae seclis, the limits of some of 
the species, and the sequence to be adopted in the Check-list. 
Zimmer, as is well known, was a meticulous worker with sub- 
species, and his treatment was reviewed by Bond, as already 
mentioned. It has therefore seemed unnecessary to alter his 
arrangement of subspecies except where modification is called 
for in the light of more recent published reviews. Hence in what 
follows no mention is made of subspecies except where they arc 
in\ol\ed in questions of specific status. 

The Cotingidae is a notoriously heterogeneous family, and 
successive systematists have despaired of producing a satisfactory 
classification of it. About half of the genera are monotypic, and 
several others are represented by a single superspecies. Many 
of these are such distinct forms that it is not possible to say to 
which other genera they are most closely related. Moreover, it 
has been doubted, by one worker or another, whether about a 
third of the genera should belong in the family at all. Recently 
there have been some very useful anatomical studies that have 
thrown new light on relationships within the family, and within 
the whole suboscine complex; but even so, it is not yet possible 
to make anything like a final classification of the family. 

What we have to deal with, it seems, is a collection of very 
diverse evolutionary lines whose common ancestry lies far back 
in the stock of forest-inhabiting, insectivorous and frugivorous 
birds that gave rise to the present-day manakins and tyrant- 
fl)'catchers, as well as the present-day cotingas. There is no 
certainty, and probably never will be, that even the "best" 
cotingid genera may not be more closely related to birds that 
are now placed in one of the other two families than they are 
to some other "good" cotingas. Nevertheless, some arrange- 
ment has to be made, and it should be based on the best avail- 
able evidence, and should be justified. It is the purpose of the 
present paper to justify the treatment to be given to the family 
in Peters' Check-list. 

THE LIMITS OF THE FAMILY COTINGIDAE 

Sibley (1970) has given a useful historical review of the 
attempts to classify the passerine birds, and it is not necessary 
to go over the same ground. For the present purpose it need 



1973 CLASSIFICATION OF COTINGIDAE (aVES) 3 

only be said that, ever since the foundations of the present 
arrangement of the passerines were laid by Nitsch, Aliiller, 
Garrod, and others, and systematized by Sclater (1888), there 
has been great uncertainty as to where the dividing lines should 
be drawn between the three main suboscine famihes of the 
New World, the Cotingidae, Pipridae and Tyrannidae (with 
which are associated the two very small families, Phytotomidae 
and Oxyruncidae) . Recently, moreover, Olson (1971) has 
widened the contro\ersv bv rex'iving Pvcraft's contention that 
the Old \Vorld familv Eurvlaemidae mav be the closest relatives 
to the Cotingidae. According to this view, the Eurylaemidae 
and Cotingidae are derived from the primitive suboscine stock, 
and the Tyrannidae, by implication, are the result of a more 
recent evolutionary radiation. It seems entirely possible, how- 
ever, that much of the detailed resemblance between the special- 
ized frugivorous members of the Eurylaemidae and Cotingidae, 
which Olson stresses, may be convergent; but whether the 
resemblance is the result of relationship or convergence, it does 
not resohe the difficulty under discussion, that among the less 
specialized members of the Cotingidae (as usually considered) 
there are genera that approach the tyrant-flycatchers and mana- 
kins more or less closely. 

Olson's suggestion raises an important general point, that it 
may be incorrect to regard the specialized frugivorous cotingids 
as the culmination of the adaptive radiation of the family. Pre- 
vious classifications of the cotingas ha\e usuallv started with the 
apparently less specialized types, and have ended with the 
specialized frugivores such as the fruit-crows and bellbirds, and 
this is a convenient sequence which is followed here too (p. 13) ; 
but it must be admitted to be no more than a matter of con- 
venience. The e\oIutionary interaction between specialized fruit- 
eating birds and the fruits to which they are primarily adapted, 
belonging mainly to a small number of plant families (Snow,, 
1971b), has obviously been a very long one; in fact, the evo- 
lutionar)- histor\' of the frugivorous stocks may be older than 
that of the more specialized insectivorous types. 

Previous classifications. Sclater, in the Catalogue of Birds in 
the British Museum, vol. 14 (1888), recognized 31 genera in 
the Cotingidae, dividing them from the Tyrannidae and Pipridae 
on the basis of tarsal scutellation. The relevant part of the key 
is as follows: 

a. Tarsus exaspidean. 

a'. Toes nearly free (as in the Oscines) . 

Bill incurved, hooked 1. Tyrannidae 



4 BREVIORA No. 409 

Bill straight, pointed 2. Oxyrhamphidae [^Oxyruncidae] 

b'. Toes more or less united 3. Pipridae 

b. Tarsus pycnaspidean. 

Bill elongated, compressed, 

not serrated 4. Cotingidae 

Bill short, conical, 

serrated 5. Phytotomidae 

In this classification, all three famiUes were distinguished from 
the Eurylaemidae by the conformation of the manubrium of the 
sternum, which, as Olson (1971) has since shown, may show 
considerable intrageneric variation in the characters that have 
been used to separate families. Although he was aware of the 
fact that the main artery of the thigh is the sciatic in Rupicola 
and the femoral in all the other cotingas studied, Sclater did 
not consider this to be of sufficient importance to justify removing 
Rupicola from the family. He placed it in a subfamily, including 
with it Phoenicircus (whose thigh arteries had not been studied 
and still have not been) for no very convincing reason. 

Ridgway (1907), in prefacing his very detailed key to the 
family Cotingidae, pointed out that the nature of the tarsal 
scutellation was not so invariable within the famiUes as Sclater 
supposed. He followed Sclater's arrangement almost exactly, 
however, the only changes being to replace Laniisoma in the 
family (placed in the Pipridae by Sclater), remove Laniocera 
to the Pipridae, and add to the family eight genera that are 
now generally considered to be tyrannid and a ninth, Xenop- 
saris, which has subsequently been bandied between the two 
families. In spite of his caveat on the variability of tarsal 
scutellation, all these changes from Sclater's treatment were 
made on the basis of this character, except for Rupicola, which 
Ridgway placed in a separate family on the basis of the arteries 
of the thigh. Ridgway also divided Sclater's Tityra into two 
genera, Tityra and Erator, largely on the basis of tarsal scutel- 
lation. 

The next important classification of the family was that of 
Hellmayr (1929). Again, he followed Sclater and Ridgway 
closely as far as most genera were concerned. He agreed with 
Ridgway in including Laniisoma, but difTered from him in also 
keeping Laniocera in the family. He removed the eight genera 
of small flycatcher-like birds that Ridgway had brought into 
the family, as well as Xenopsaris. Like Ridgway, he treated 
Rupicola as a separate family. He did not subdivide the genus 
Tityra. 



1973 CLASSIFICATION OF COTINGIDAE (aVES) 5 

These three classifications agreed in large part, in regard to 
the genera admitted into the family. Ridgway's sequence, how- 
ever, departed radically from that of Sclater. Both sequences 
were simply the result of the keys which they devised, and they 
do not appear to have been intended to reflect any evolutionary 
development, such as from more primitive to more specialized 
types. Thus Ridgway began the family with the peculiar genus 
Phoenicircus and followed it with the vers- different genus 
Laniisoma simply because the first division of his key dealt 
with the union (or non-union) of the outer toe to the middle 
toe. 

It is noteworthy that these three authors all questioned the 
correctness of placing Attila (and Sclater also Casiornis) in the 
Cotingidae rather than in the Tyrannidae. Ridgway expressed 
the same doubts about Rhytipterna and Lipaugus and the fly- 
catcher-like genera, which he separated from the Tyrannidae 
"solely on account of the different character of the tarsal en- 
velope." 

Aleyer de Schauensee f 1966) followed Hellmayr fairly closely. 
He replaced Xenopsaris in the family, however, with a footnote 
indicating that it may be tyrannid, and he added the newly 
disco\'ered genera ^aratornis and Conioptilon, and the recently 
described genus Pseudattila. In his later book (1970) he made 
a more important change. As a result of recent anatomical 
research, he removed five genera and placed them in the Tyran- 
nidae. These were the controversial Attila (with Pseudattila), 
Casiornis, Laniocera, and Rhytipterna. He also modified HeU- 
mayr's order of genera in two ways. He placed Ampelion, 
^aratornis, and Doliornis later in the sequence, associating 
them with the other Andean genera Pipreola and Ampelioides 
rather than with the eastern Brazilian genus Phibalura. Secondly, 
he placed Phoenicircus at the end of the family, after Procnias, 
instead of at the beginning. 

The debated genera. The eight genera that Ridgway alone 
included in the Cotingidae are now generally agreed to belong 
in the Tyrannidae, and they will not be discussed further 
{Microtriccus, Ornithion, Tyrannidus, Idiotriccus, Elainopsis, 
Hylonax, Ramphotrigon, Syristes). Apart from these, there are 
nine genera whose position has been debated. 

Lanhsoma 
Laniisoma was placed in the Pipridae by Sclater (under the 
name Ptilochoris) on the basis of its foot structure and tarsal 



6 BREVIORA No. 409 

scutellation. Ridgway replaced it in the Cotingidae, linking it 
with Phoenicircus, also on the basis of its foot structure. Curi- 
ously enough, among the other characters that he listed for 
these two genera, he gave for Phoenicircus "adult males with 
sixth and seventh primaries shortened, especially the seventh, 
which is strongly bowed . . .," and for Laniiso?na "adult males 
with sixth and seventh primaries normal"; but in fact males of 
Laniisorna have the seventh primary modified, as Sclater noted, 
though to a much less extreme degree than Phoenicircus. Thus 
Laniisorna and Phoenicircus agree in structural characters of 
both wing and foot, though they are very different in plumage. 
Laniisorna has not been dealt with in recent anatomical studies, 
being rare in collections, and nothing is known of its l^ehavior. 
Its distribution is similar to that of Phibalura, with a south- 
eastern Brazilian population and restricted populations along 
the eastern slopes of the Andes. It also resembles Phibalura in 
some points of plumage. 

On present evidence Laniisorna is best kept in the Cotingidae. 
It is in accordance with the Kttle that can be inferred about its 
relationships to place it between Phoenicircus and Phibalura, 
but this creates difficulties, as discussed under Sequence. 

Phoenicircus 

Phoenicircus seems to be a very isolated genus, in spite of 
the characters shared with Laniisorna, as noted above. All 
recent classifications have placed it in the Cotingidae, though 
Hellmayr (1929) remarked (footnote, p. 92) that it is "a genus 
of doubtful affinity, perhaps more nearly related to the Rupico- 
lidae or constituting a family by itself." Sclater also linked it 
with Rupicola, in the subfamily Rupicolinae, but on what real 
evidence is not clear. In his key to the subfamilies the designa- 
tion for the Rupicolidae, "Tarsi very strong; gressorial," refers 
only to Rupicola; in fact only the crest (very different in the 
two genera), the presence of modified primaries in the male 
(but difTerent ones, and many cotingid genera show such modi- 
fications) and the general colour serve as a very tenuous link 
between the two genera. 

The anatomy of Phoenicircus has not been investigated, and 
very little is known of its behaviour except that it appears to have 
some form of communal display ( Olalla, 1 943 ) . The united 
toes, the colour of the plumage, and the communal display all 
suggest that it may be an overgrown manakin. At least, this 



1973 CLASSIFICATION OF COTINGIDAE (aVES) 7 

possibility desen-es as much consideration from future workers 
as the more u^ial idea that it is cotingid; but until more is 
known about the genus it is preferable to retain it in the 
Cotingidae. 

Attila 

Superficiall)-, its narrow hooked beak, flycatcher-like plumage, 
general behaviour (Skutch, 1971 ), and habit of nesting in niches 
in tree trunks all tend to suggest tyrannid affinities for the genus 
Attila. Sclater, however, following Sundevall, placed it in the 
Cotinoidae, forming a subfamily Attilinae with Casiornis, on 
the basis of the tarsal en\elope. .Mthough he expressed doubt 
about the correctness of this, subsequent authors all kept Attila 
as cotingid, though usualh' with reservations, Ridgwa\- pointed 
out that the tarsus is not in fact typical of the cotingas. Meyer 
de Schauensee ( 1 966 ) followed the con\entional treatment in his 
1966 work, but in 1970 transferred Attila to the Tryannidae, 
as a result of ^Varter's studies of the skull. ( Warter's work is not 
yet published, although cited with a publication date of 1966 
in Meyer de Schauensee, 1966: 314, footnote.) The more 
recent studies of the syrinx b\- Ames i 1971 ) confirm the tyrannid 
character of Attila, whose syrinx is closely similar to that of 
Mxiarchus. It mav be noted that these two genera have the 
same kind of nest-site. 

There is little doubt that Attila should be remo\ed from the 
Cotingidae, and placed in the Tyrannidae near to Myiarchus. 
The anatomy of Pseudattila has not been studied, but it seems 
so close to Attila (not having been separated as a genus until 
1936) that it may safelv be placed next to it. 

Casiornis 

Most of what has been said about Attila applies equally to 
Casiornis. Structurally it is close to Attila (in tarsus, beak, and 
syrinx [Ames]), but little seems to be known about it in life 
and its nest has apparently not been found. On present kno\\l- 
edge is seems best to place it close to Attila and remo\e it from 
the Cotingidae. 

Lan^ocera 

Sclater placed Laniocera in the Cotingidae, whWc notin<: that 
in foot structure it resenibles the Pipridae. Ridgway, on tlie 
basis of its foot structure fcxaspidean tarsus and extensi\ely 



8 BREVIORA No. 409 

coherent toes), placed it in the Pipridae, but it was replaced 
in the Cotingidae by Hellmayr in spite of its foot structure 
because of the general siniilarit\- of its appearance to Lipaugus 
and Rhytipterna. Recent studies of its osteology and svrinx, 
howe\-er, suggest strongh- that its correct place is in the T\ ran- 
nidae, presumably close to Attila. The genus is litde known 
in life, and no nest has been found. 

Rhytipterna 

There was general agreement in placing Rhytipterna in the 
Cotingidae, close to Lipaugus, until Meyer de Schauensee re- 
mo\-ed it to the Tyrannidae along with the genera Attila and 
associated genera, as the result of recent studies of the osteology 
and syrinx. This anatomical e\'idence seems too strong to be 
set aside. 

The \ery peculiar tarsal scutellation of Rhytipterna separates 
it from all other genera, whichever family it is placed in. 
Nothing seems to be known of the genus in life, but it mav be 
guessed that the spiky tarsal scutes are in some way connected 
with its nest-site or e\'en its roosting behaviour. In the former 
case they might be analogous to the horny processes on the 
plantar surface of the tarsi of some hole-nesting non-passerines 
such as jacamars (in which case there would be a further link 
with Attila, a niche-nester), or in the latter case to the tarsal 
serrations of the tree-roosting Tinamus. Field obser\'ations on 
all the genera that are here thought to be close to Attila are 
badly needed. 

Lipaugus 

There has been consistent agreement among systematists that 
Lipaugus is a cotinga, but on the basis of his syringeal studies 
Ames (1971) has recendy suggested that "Lipaugus may later 
be added to this group" (Attila and related genera, which must 
be transferred to the Tyrannidae). The only specimen avail- 
able for study, however, was damaged, and no firm conclusion 
was possible. 

The two closely related species L. vociferans and L. unirufus 
seem to be typical cotingas, so far as they are known in life. 
Males of L. vociferans display in leks (the advertising display 
being vocal, not visual), and the nest (L. unirufus, Skutch, 1969; 



1973 CI.A.--..SIK1C.ATIOX OF COTINGinAE ''aVES) 9 

L. vocijcrdiis. Willis, personal coninumication) is a minute stick 
platform such as other medium-lars^e cotingas (Xipholena, 
Procnias) build, but no t\rannids so far as known. Moreo\er, 
in spite of the dull o;re\, brown, or greenish plumage of most 
of the species, L. streptophorus has a brilliant magenta collar 
encircling the anterior part of the body, and \'olker (1952) 
has shown that the pigment in these feathers, as in the red 
feathers of other cotingids, changes to orange under mechanical 
pressure. 

In summar\-, the e\idence, though scant}', suggests that 
Lipaugus should be retained in the Cotingidae. The peculiarities 
of the syrinx may well be related to the extraordinan' de- 
\'elopment of \ocal display in the genus, and may thus be a 
specialization rather than an indication of relationship. In this 
connection it would be interesting to examine the syrinx of a 
species that has not been reported to ha\e an unsually loud 
voice, e.g., L. streptophorus. 

It is not surprising that Hellmavr was impressed b\' the 
general similarity of Lipaugus, Laniocera, and Rhytipterjia. In 
addition to their general resemblance they show a remarkable 
parallelism in geographical \-ariation. In each genus there is 
a grey form widespread in South America east of the Andes, 
which is replaced by a chestnut-brown form in Central .\merica 
and South America west of the Andes. 

Xenopsaris 

Sclater placed this genus in the Tyrannidae, as did most 
authorities of his time, although the single species, X . albinucha, 
was originally ascribed to the cotingid genus Pachyramphus. 
Ridgway placed it in the Cotingidae, noting that its tarsus is 
not exaspidean, thus not conforming to his main criterion for 
the Tyrannidae. Hellmayr removed it again to the T\rannidae, 
and Meyer de Schauensee replaced it in the Cotingidae. This 
genus has thus been switched back and forth four times. As 
Meyer de Schauensee notes, "anatomical in\estigation is needed 
to settle the disputed status of this gentis," but it is not abundant 
in collections and its anatomy is still unknown. 

If it is a cotinga, it is unique in its habitat; no other member 
of the family is found primarily in reedbeds and other rixerine 
vegetation, Init many t\rannids are. The e\idence for the t\rnn- 



10 BREVIORA No. 409 

nid affinities of Attila and related genera shows that undue 
reHance should not be placed on tarsal characters alone. The 
evidence is clearly inadequate for a proper decision, but on 
balance it seems best to place Xenopsaris in the Tyrannidae. 

RUPICOLA 

As mentioned above, Sclater kept Rupicola in the Cotingidae 
in spite of the fact that the main artery of the thigh is the 
sciatic (as in the Tyrannidae) and not the femoral (as in the 
other cotingas studied). It is uncertain what weight should be 
gi\'en to this character until the thigh arteries of all the cotinga 
genera have been studied, but it should be noted that there is 
apparently no other character which links Rupicola with the 
tyrant-flycatchers rather than with the cotingas. On the basis 
of the thigh arteries, Ridgway placed Rupicola in a separate 
family, and Hellmayr followed him presumably for the same 
reason. Subsequent authors have not been consistent: for 
instance, Meyer de Schauensee (1966) maintains the family 
Rupicolidae, while Sibley (1970) has replaced Rupicola in 
the Cotingidae. 

Sibley's decision was made before any evidence from egg- 
white proteins was available. Since then, he has analysed the 
proteins from a fresh ^gg that I sent him from southern Guyana 
(Snow, 1971c), and he writes as follows: "Your fine speci- 
men has been 'run' at least three different times with different 
comparisons, both in isoelectric focusing and also in the standard 
type of net charge electrophoresis. In all of these it was com- 
pared with excellent specimens of Pachyramphus, with Phyto- 
to7na, Manacus, at least one or more t^rannid flycatchers, an 
oxenbird, and an antbird. Without any question, in all of these 
comparisons, Rupicola agrees most closely with Pachyramphus. 
Thus I think it is entirely in keeping with the egg white evi- 
dence, which seems to me satisfactory, to include Rupicola in 
the Cotingidae." 

If Rupicola is retained in the Cotingidae, its isolated position 
in the family can be given recognition by placing it at the end 
of the sequence of genera, after the very difficult bellbirds 
(Procnias) . The two genera thus represent two distinct t)pes 
of specialized frugivores, in which the pure fruit diet is associated 
with marked sexual dimorphism and the evolution of elaborate 
courtship display. 



1973 CLASSIFICATION OF COTINGIDAE (aVES) 11 

RECOGNITION OF DEBATED GENERA 

Most cotinfrid a:enera are so distinct that there has been c^eneral 
agreement about their recognition, even though their affinities 
may be in doubt. There are, however, a few genera that have 
been recomized bv some authorities but not bv others, and 
these are discussed below. In a family like this, where rela- 
tionships are so hard to detect and the ratio of genera to species 
is so high, it seems far more satisfactory to use generic names, 
when possible, to indicate affinity rather than to emphasize 
minor difTerences between forms that are certainly quite closely 
related. 

Ampeliox, Doliornis and Zaratornis 

Taczanowski in his original description admitted the nearness 
of Doliornis to A?npeIion. but it w^as maintained as a separate 
genus apparently without dissent until Bond '1956) suggested 
that both it and ^aratornis should be merged with Ampelion. 
Ampelion and Doliornis are both Andean, occurring at high 
altitudes; they share the same plumage colours (greys and 
chestnut-browns), and they agree closely in their concealed 
red-brown nuchal crest. The main structural difTerence is that 
Doliornis has a narrower and less hooked beak than Ampelion. 
There has been no study of the internal anatomy of Doliornis, 
of which only two specimens are known from a single locality 
in Peru. It seems most likely that Doliornis sclateri has been 
derixcd from an isolated fragment of Ampelion stock, and the 
maintenance of the monotypic genus seems unnecessary. 

The best treatment of ^ar atom is is more difficult to decide. 
Bond recommended merging it with A?npelidn because of the 
general similarity of its plumage, especially to that of A. ruf axilla. 
The red-brown nuchal patch is present, though reduced. Its 
relict distribution in the Andes, like that of Doliornis, suggests 
an offshoot from early Ampelion stock. According to J. Farrand, 
Jr. 'personal communication), it is \"ery like Ayyipelion in 
general beha\iour. Farrand writes: "I spent several davs at 
Zarate in the Department of Lima with the late Dr. Maria 
Koepcke in 1964, and saw the species daily. In its general 
behavior it is \'erv like Aynpelion rubrocristata. In a manner 
verv reminiscent of that species it often 'pops up' suddenh' onto 
a dead snag and sits upright, looking about rather ner\ously. 
The flight of ^aratornis is very similar to that of Amfjelion. and 
both species approach a perch flying low and making a final 



12 BREVIORA No. 409 

upward sweep, rather like that of a shrike or kestrel." On the 
other hand, Lowery and O'Neill (1966) state that its skull, 
which they have examined, is "so distinct as to preclude making 
^aratornis congeneric with Ampelion." Nevertheless, they place 
its skull next to that of Ampelion in what they describe as a 
"nicely graded series" of six genera, and the exact differences 
are not specified. Without knowing to what extent skull struc- 
ture can be modified in response to differences in feeding be- 
haviour in otherwise closely related forms, it is not possible to 
weigh the skull differences against the other evidence that 
suggests close relationship to Ampelion. Provisionally, I recom- 
mend merging ^aratornis with Ampelion. 

Pachyramphus and Platypsaris 

The closeness of these two genera has been obvious to every- 
one who has studied them, though nobody has Hked to take the 
step of formally merging them. Ridgway gave as distinguishing 
characters the greater size of Platypsaris, the less extreme special- 
ization of the modified ninth primary in the male, and the more 
cylindrical, less broad, beak. These hardly seem to warrant 
the recognition of a separate genus in view of the diversity of 
size and other characters within the genus Pachyram.phus. 
Moreover, the partly concealed white scapular feathers charac- 
teristic of Platypsaris are also present in Pachyramphus surina- 
mus. Both genera build gobular, bulky nests, with a side 
entrance, but that of Platypsaris is typically suspended from 
the tip of a drooping branch, while that of Pachyramphus is 
typicallv supported in a vertical or horizontal fork. 

The four species of Platypsaris are almost completely allopatric 
(see later, p. 21). They overlap widely with several species 
of Pachyramphus, and their greater size is presumably one of 
the means by which they avoid competing with them. None 
of the distinctions seem great enough for generic separation, 
and thev are best united under Pachyramphus, the older name, 
as Bond (1959) has already recommended. 

Tityra and Erator 

At first glance the three species in this group [cay ana, semi- 
jasciata, and inquisitor) seem very close to one another. 
Ridgway, who laid stress on structural characters that might 
be used to separate genera, phced the last-named species 
in Erator on the basis of its feathered lores and orbital region. 



1973 CLASSIFICATION OF COTINGIDAE (aVES) 13 

taxaspidean (not pycnaspidean) tarsus, and broader and flatter 
bill. Hellmayr and most subsequent authors have preferred to 
recognise a single genus, although Wetmore (1926, 1927) has 
upheld Erator for the reasons gi\en by Ridg\vay. It is almost 
inconceivable that this group of hole-nesting cotingas, whose 
plumage, modified ninth primary, general proportions, behav- 
iour, and calls are so similar, are not closely related and 
monophyletic. The presence or absence of areas of bare skin 
clearly need not be considered of generic importance in the 
Cotingidae (cf. Cephalopterus, Procnias) ; beak shape is pre- 
sumably adapted to feeding habits (though exactly how in 
these species is not known) ; while the difference in tarsal scu- 
tellation must be taken to indicate that \ariation in this 
character may e\-oh'e relatively easily without major modifica- 
tion of other characters. 

THE SEQUENCE OF GENERA 

As is true for all diverse families, no linear order can express 
relationships satisfactorily. This would be true even if the 
relationships were perfectly understood. In the Cotingidae we 
have a large number of isolated genera whose affinities are quite 
uncertain, and it seems best to use a sequence that keeps as 
close as possible to what has been customary while taking 
account of the more well-based conclusions from recent studies. 

Sclaters sequence, as already mentioned, followed directly 
from his key, and subsequent advances in the systematics of the 
family make it obsolete. Ridgway's sequence also followed 
directly from his key, and although it is to this extent artificial it 
has been generally followed. The reason for this is probably 
that it seems a more or less natural sequence — • except that it 
starts with Phoenicircus, but this genus would be anomalous in 
an\- position except the beginning or the end. x'\fter this awk- 
ward start, the sequence proceeds with five genera ( Laniisoma, 
Phibalura, Heliochera {^Ampelion) , Tijuca, and Ampelion 
{=Carpornis) ) which are undoubtedly cotingid but un- 
specialized — that is to sav, not specialized towards the large, 
frugivorous, highly dimorphic types that seem to be the culmi- 
nation of evolution in the family (but see the caveat on p. 3). 
They are not \ery specialized in beak shape, and have plumages 
in which yellows, greens and greys predominate. Moreo\-er, 
these apparently unspecialized genera are distributed peripherally 
to the main lowland forest region, which is the headquarters of 



14 



BREVIORA 



No. 409 



most of the specialized friigi\ores. It is reasonable to regard 
these genera as closer to the early cotingid stock than the larger, 
brilliantly coloured fruit-crows and other bizarre forms that are 
placed towards the end of the generic sequence. 

Four genera follow {Porphyrolaema, Cotinga, Xipholena and 
Carpodectes), which consist of medium-sized speciahzed frugi- 
\orcs with striking colours (purple, wine-red, blue or white) 
in the male plumage and duller-coloured females. There then 
follow (omitting the genera that have been placed in the 
Tyrannidae by all other authors) three genera of small or very 
small frugivorous forms {Pipreola, I odo pleura, Calyptura), the 
attiline genera and Lipaugus, the tityrine genera, and finally 
the large and diverse fruit-crows with the very specialized Ijell- 
birds {Procnias) placed at the extreme end. 

Ob\'iously the sequence in which the middle groups should 
be placed is to a large extent arbitrary, but this order does 
broadly represent an advancing sequence along the line of 
specialization for which the family is noted. Probablv for this 
reason, Hcllmayr followed it almost exactly (except in so far 
as he exluded or included the debated genera). Meyer de 
Schauensee also followed it, except for one major change: he 
placed Aynpelion and the two rehted genera Doliornis and 
^aratornis (^the latter not described when Hellmayr wrote) some- 
what later in the sequence, before Pipreola and after the 
recent!)- disco\ered Conioptilon. 

I iDcliexe that the basis of Hellmayr's sequence can be retained, 
but that some changes are needed in order that it should reflect 
as far as possible the most likely relationships in the light of 
recent knowledge. The suggested sequence, with annotations 
and justification of changes, follows: — ' 



Plinrnirir( //? 



"I retained at llic head of the list liecausc it is 

}- so isolated but, so far as it shows affinities with 

J other genera, may be linked with Lrmiisoma. 



Lnniisoma 
Pliihalurn 
Tijnra 
Cfi) pniiiis 

Ampclion 

(including Doliornis 
and 7.ar(itornis 



these four genera appear to be products of the 
same evolutionary centre (the SE Brazilian high- 
lands, from which the fust two have spread lo 
the eastern slo])('s of the Andes) 



shares some characters (cspeciallv nuthal crest) 
with Phihiilura 



1973 



Cr.ASSIFICATIOX OF COTIXOIDAE 'aVES) 



15 



Pil)reola 
Ainpelioides 

lodoplewa 
Calx j)t inn 

Lipfiugus 
Cliironlla 



almost aitainlv closclv related: perhaps an 
ollshoot frdin Aiulean stock sliariiig a rominon 
j aiucsti\ Avitli A III Ix'lioii. and more remotely 



\\'itli the four precedinji; SF. Rrai'iliaii genera. 

of uncertain statns: small forest hirds not ob- 
viously related to each other nor to anv other 
genera: svrinx of lodoplcuKi of tvrannid l\pc 
(Ames 1971) 



i pioi)al)l\ belong together: not ohviowslv rihited 
f to other genera 



Paclni'diii pints 
(including Plnlxpsniis 

Tit\ra 



Porphxrolaeiiia 
Cotinga 
Xiphohna 
Carpodectcs 



Couioplilon 



Gyiinwderus 

Haematodeius 

Q_uernla 

Pxroderus 

Cephaloptcrus 

Perissocepludus 

Pracnins 



l)rol)abl\ fairh closeh related (plumage colours, 
modified niiuh [)rimar\ of nuilcs, beak shape) : 
not ob\ioiislv related to other genera 

medimn-si/cd specialized frugisores. sharing 
some ])linnage and structural characters: listed 
together in this sequence hv Ridg\sav and later 
authors. Xipliolcna and Cm podcctcs appc;n- to 
he closelv related, on some characters (Snow, 
H'Tli: but Anu;s lias fcuind marked dillei cnces 
in svringeal structiue and puts Carpodectcs in a 
dilfcrcni group from the other three, with 
Qiiriuhi. I-inther evidence is needed before 
U|)setting the existing order, which seems satis- 
fac tor\' in other respects. 

lesemhles Cai podeclcs in .skull characters, and 
seems to form a link between the above group 
and ('•xiiniodcnis. ^vhich it resembles especially 
in tlie great cle\elopmeiU of jiowder do'ivn 
(Lowery and OXeill, 1960). 

large, niainlv frugi\c)r()us cotingas. including the 
so-called fruit-crows. The secpience is the same 
as that adopted bv Ridgwav and followed by 
;dl lalei ;nithors. except that Gxiiiiiodcnis is 
placed first instead of near the end. This en- 
ables Couioplilon and Cxni'iodcius to be placed 
together, and does not \iolate anv other known 
or picsuiued ;i(rniities. The position of Piocnias 
at the end is in iiccord with its highh distiiut 
svrinx. 



Riipirola 



16 BREVIORA No. 409 

Species Limits 

A rather large proportion of all genera of cotingas consists of 
groups of parapatric forms (mainly or entirely allopatric forms 
that replace one another geographically). The differences be- 
tween the allopatric populations range from very slight to \ery 
marked, thus posing problems of e\ery degree of difficulty in 
the often arbitrary decisions that haxe to be made about specific 
limits. Each case needs to be treated on its own merits, but in 
fact there is too little e\idence in nearly every case to allow 
a reasoned decision to be made. It is noteworthy, howe\'er, 
that in two cases where verv similar forms come together 
{Phoenicircus carnifex and P. nigricollis, Xipholena punicea 
and X. lamelbpennis) they appear to behave as good species, 
admittedly on ver\' slender evidence (Haffer, 1970). The mem- 
bers of these two pairs are more alike than the members of most 
other allopatric pairs or groups of cotingas whose specific status 
is debatable. This should lead to caution in lumping allopatric 
forms into single species when only an arbitrary decision is 
possible. It is also very difficult, once one starts to lump allo- 
patric forms, to know where to stop; and awkward situations 
occur. Logically, if closely related allopatric forms are to be 
treated as conspecific, Xipholena atropurpurea should be treated 
as conspecific with one of the \'er\' closely related species which 
replace it geographically; but with which? As far as one can 
tell, it is somewhat intermediate between punicea and lamelli- 
pennis and there is no good reason to link it with one rather 
than with the other. 

For these reasons, I ha\'e been conservatix'C in my treatment 
of allopatric forms, retaining specific status for them unless there 
seems to be particularly good reason to merge them; and even 
this in\ol\'es subjectixe judgment with which it is easy to dis- 
agree. In order to make the classification gixen in Peters more 
useful for zoogeographic analysis, groupings of zoogeographic 
species and possible alternati\'e taxonomic treatments will be 
indicated in footnotes, where appropriate. 

Phoenicircus. The two very similar species, P. carnifex 
and P. nigricollis, meet in the region of the lower Tapajos and 
perhaps the lower Xingu and Tocantins (HafTer, 1970). It 
seems that the\' must exclude each other from their respecti\'e 
ranges, but information is quite inadequate to decide the point. 
There is no evidence for intergradation between them. On 
present evidence they are best kept as separate species. 



1973 CLASSIFICATION OF COTINOIDAE (AVES) 17 

PiPREOLA RiEFFERii and P. INTERMEDIA. Thesc two specics 
are closely related, differing chiefly in the terminal tail markings 
of intermedia, which are absent in riefferii, and, less strikingh , in 
other plumage characters. P. riefferii, which has a wide range 
in northern parts of the Andes, is replaced b\ intermedia in 
Bolivia and most of Peru. Hellmayr ^1929) treated them as 
specifically distinct, since specimens of both, showing no signs 
of intermediacy, have been collected near to each other in the 
Department of Libertad, intermedin at 2400 m and riefferii 
at 1200 m. Altitudinal data from elsewhere do not show inter- 
media as occurring consistently higher than riefferii, and in 
fact 1200 m is an exceptionally low altitude for either species; 
in the Department of Amazonas in northern Peru riefferii has 
been collected at 2300-2800 m, while further south intermedia 
had been recorded mainly at 2500-3000 m. Further informa- 
tion is needed on the distribution of these two species where 
they approach one another, but Hellmayr's opinion regarding 
their specific status appears sound. 

Pipreola arcuata is almost certainh a deri\ati\e of the riefferii- 
intermedia stock, from which it differs chiefly in its large size 
and barred underparts. It now o\'erlaps extensi\ely with them. 

Pipreola aureopectus, P. lubomirskii, P. jucunda, and 
P. pulchra. The last three of these ob\iously form a natural 
group, as Hellmayr recognized. P. aureopectus is more distinct, 
but pulchra forms a clear link between it and the other two. 
The male of pulchra is almost exactly intermediate in its crown 
colour, and the female is almost exactly like the female of 
aureopectus except for the lack of white apical margins to the 
wing feathers. All four species agree in the unique combination 
of red beak, yellow iris, and oli\e-grey feet. They replace one 
another from north to south in the Andes, in the order listed, 
with no overlap so far as known. It seems clear that thev should 
be treated as conspecific. 

Pipreola frontalis and P. chlorolepidota. The males 
of these two species at first sight appear \ery different, since the 
underparts of chlorolepidota are darkish green apart from the 
throat-patch, and those of frontalis are yellow. But the former 
.species retains a small \ellow area lateral to the throat-patch, 
where the feathers are brightest \ellow in frontalis, and the two 
species agree in the colour of the soft parts fin the adult male, 
white or vellow iris, orange or red beak and feet). Moreo\'er, 
the females arc almost identical in plumage. .\lthouQh they 
appear to be mainly allopatric, so far as can be told from the 



18 BREVIORA No. 409 

limited records, they overlap in southeastern Ecuador and prob- 
ably northeastern Peru, and so cannot be considered conspecific; 
but they ha\e probably speciated comparatively recently. 

lODOPLEURA FUSCA 0)1(1 I. ISABELLAE. ThcSC tWO fomiS 

replace one another, so far as known, but there are wide areas 
where neither has been found. Fusca has been found at f\vc 
localities in Guyana, one in Surinam, and one in eastern \'ene- 
zuela on the upper Caroni River; the nearest records for isabellae 
are from the upper Orinoco region, the Rio Negro and the 
middle Amazon (Obidos). Until more is known of the situation 
in the inter\'ening areas it seems premature to treat them as 
conspecific, as Hellmayr and others have suggested; they cer- 
tainly differ more than do Phocnicircus carnijex and P. nigri- 
collis. ft may be noted that lodofAeura pipra, the only other 
species in the genus, which occurs in southeastern Brazil, is also 
known from two specimens from Guyana. The possibility of 
long-distance migration by these small aerial cotingas of the 
trectops cannot be excluded. 

LiPAUGUS VOCIFERANS, L. UNIRUFUS and L. LANIOIDES. 

These three forms replace one another geographically. Hellmayr 
considered lanioides probably conspecific with vocijerans. but 
did not go so far as to merge them. No systematist has merged 
vocijerans and iinirufiis, although they are almost certainly more 
closelv related than vociferans and lanioides. 

L. rorifrrans and L. iinirufus between them occupy almost 
the whole of the lowland tropical forest region of Central and 
South America. L. vociferans occupies the greater part of the 
South American range, but does not occur north of the Orinoco 
(except near its mouth). L. unirujns does not extend east of 
the Alagdalcna \"allev in Colombia. There is thus an area in 
eastern Colombia and western Venezuela, containing some low- 
land forest, where neither occurs. The two species are \ery 
different superficially, since vociferans is all grey and unirufus 
all rufous-brown, but in beha\'iour and ecology the litde e\-i- 
dence there is suggests that they are rather similar, and the 
colour (lifTerence could be based on a small genetic difference. 
For the reasons gi\en in the introduction to this section I prefer, 
howe\"cr, not to treat them as conspecific. 

At first sight, /.. lanioides differs from vociferans less than 
docs unirufus. but the differences arc in fact probably more 
numerous and important. It is larger, with a browner rumn 
and undcrparts, and brownish fnot grey) wings and tail. It 
replaces vociferans in southeastern Brazil south of 20*^ S. The 



1973 CLASSIFICATION OF COTIXCIIDAE (AVES) 19 

separation at this point involves habitat as \vell as range, as 
vocijerans occurs in lowland tropical forest while la?iioides is 
found mainly in subtropical montane forest (but has also been 
recorded near sea-lc\el, perhaps as a result of \ertical migration) . 
Further, although little is known of lanioidcs in life it is hard 
to belie\e that it could utter the kind of piercingh' loud calls, 
in Icks, that are so characteristic of vocijerans where\er it 
occurs. It seems more likely that lanioides and the other mon- 
tane species of Lipaugus [strcptophorus. fuscocinereus, crypto- 
lophw: and subalaris) are comparatixeh' quiet birds, with a 
different social organization from voriferans and iinirufus. 
Hellma\r"s suggestion that lanioides and voeiferans are con- 
specific thus recei\es no support in the light of more recent 
knowledge. 

The genus P.achyramphus. \\'ith 1 1 species usually recog- 
nized (15 if Platypsaris is included . P(i< hxKnnphus is easih the 
largest cotingid genus. It is also unique in the famih" in the 
amount of geographical o\erlap between species; in extensixe 
areas up to five or exen six species may occur close together, 
even if not in exactly the same habitats. Some of the species 
ran be placed in groups that constitute zoogeographical species. 
The arguments f(;r and against recognizing larger specific imit^ 
than are currently recognized are discussed belo\\ . 

Pachyramphus castaxeus and P. cixnamomeus. P. 
castaneus differs from P. ( innamomeus principalh in its smaller 
beak and in the presence of a narrow grey stripe separating 
the chestnut of the crown from the paler brown sides of the 
head and neck. Othei-visc thc\ are closeh similar, and are the 
onh' two species in \vhi( h male and female are alike in plumage. 
MoreoNcr, both are characteristicalK found in lowkmd tropical 
forest, a habitat axoided b\ most of their congeners. P. casla- 
neus also ranges up into the subtropica.l zone. 

Between them, these two species apparentlv occupx' mo-^t of 
the tropical forest area of South and Ontral .\merica. In iioith- 
western \ cnezuela and eastern C!olombia, thi'ir ranges aj~)proach 
closeh : P. easlaneus occurs in the coastal mount, liiiv o{ north- 
ern \ cnezuela, and is rej)laced 1)\ /'. ( itniauion!: us round !/:ke 
Maracaibo and in the mountains <outh of Lake Maiacaibo. 
l-'urther south, the potential habitats of the two spaii-s are 
widelv separated b\ saxanna ( onntr\ west of the ()rin(H(i, e\- 
(.vpl in the west wliere a eoriidor of forest aloni; t!u' ea^tein 
foothills of the .\ndes comiei ts the main foiest area^ (Htupied 
hv the two species. P. c/imaiiioninK h.is ajiparenth' occupied 



20 BREVIORA No. 409 

this corridor from the north, as it has been recorded from near 
\'illa\icencio, while P. castaneus occurs not far to the south, in 
the forest of the Sierra Macarena. 

This geographical replacement suggests that the two species 
are potential competitors. The morphological differences be- 
tween them, though slight, seem too clear-cut to warrant 
treating the two forms as conspecific. 

Pachyramphus rufus and P. spodiurus. Hellmayr pointed 
out the similarity of these two species and suggested that 
spodiurus might well proxe to be the Pacific representative of 
P. rufus. P. spodiurus is larger, with a slenderer bill; the male 
is much darker than rufus, but the female plumage is closely 
similar. The little that has been reported on habitat preference 
suggests that rufus avoids hea\'y forest, preferring wooded sa- 
vanna and open woodland; it nexertheless occurs in areas 
dominated by rain-forest. P. spodiurus is also a bird of open 
woodland, and the absence of records from the very humid 
coastal forest of northwestern Colombia suggests that it is absent 
from this area, which thus effectively isolates it from rufus to 
the north. 

This is a case much like se\'eral others, where two allopatric 
forms are clearly closely related and it must remain a matter 
of judgment whether or not they should be treated as con- 
specific. On balance, and in the absence of any real exidence, 
it seems that consistency demands that rufus and spodiurus 
should recei\e the same treatment as castaneus and cinnamomeus 
and should pro\'isionally retain their specific status. 

Pachyramphus marginatus and P. albogriseus. These 
two species agree in many points of plumage, the most con- 
spicuous difference being in crown pattern: the male of 
alboariseus has extensive black on the crown and nape, and 
the female a brown crown bordered posteriorly with black, 
forming a patch of the same extent as the male's black; whereas 
7nar[>inatus has a small black crown-patch in the male and a 
brown crown with no black border in the female. 

P. marginatus is a bird of lowland tropical forest, east of 
the Andes. P. albogriseus replaces it in limited areas of the 
Andes, west of the Andes, and in Central America, being found 
in a variety of habitats, tropical and sul^tropical. Where the 
two species approach one another, on the eastern slopes of the 
Andes of Peru and Ecuador and in northern Venezuela, P. 
albogriseus occurs only at high altitudes and P. marginatus 
in the lowlands. 



1973 CLASSIFICATION OF COTINOIDAE ( AVES ) 21 

Again, it seems clear that these two are related, but the little 
evidence a\ailable does not justifx' treating them as conspecific. 

Platypsaris. The close affinity of Pachyramphus and 
Plairpsaris, and the reasons for merging the two genera, ha\-e 
alreadv been discussed. Platypsaris consists of fi\e forms, usually 
treated as species, which are almost completely allopatric. Four 
of these forms between them occupy almost the whole of tropical 
and subtropical America except some montane and unwooded 
areas. They clearly constitute a zoogeograpical species, but the 
status of the constituent members is not clear. It is conx'enient, 
howexer, in discussing them to use the four specific names that 
are in general use. (Unfortunately, the specific name rufus is 
in use in both Pachyramphus and Platypsaris, and when the 
genera are merged it will be available only for the species cur- 
renth' known as Pach\ramphus rufus, which has priority.) The 
fifth form, P. niger, occurs onlv in Jamaica and is rather dis- 
tinct. The following discussion concerns only the four mainland 
forms. 

P. ininor, the central form, occupies the whole of the tropical 
forest of the Amazon basin and upper Orinoco system. To the 
south and east, P. rufus occurs in a wide \'aricty of more sea- 
sonal wooded habitats, tropical and subtropical. These two 
differ in a number of plumage characters in the male, and less 
strikinglv in the female; they differ also in size. There appears 
to be no intergradation between them, and moreo\-er their ranges 
ON'erlap in two areas, in northern Para and Maranhao in north- 
ern Brazil and along the base of the Andes in Peru and Boli\ia. 
These are areas where lowland tropical forest comes into con- 
tact with more open vegetation types (N Brazil) or with lower 
montane woodland (base of Andes). Their ranges also approach 
closelv, and may e\'en interdigitate, along the southern fringe 
of the main Amazonian forest in central Brazil. 

To the northwest, P. yninnr is isolated from P. hoyriochrous, 
also a forest bird, bv the sa\annas of \'enezuela and eastern 
Colombia. Neither species is found in the forested coastal 
mountains of northern \xnezuela. These two fonns differ 
slightK but constantl\- in both mal-^ and female plumage, and 
are much alike in size. 

In Central America P. hoDiot /nous and /'. aiihiiai', bolh in- 
habiting forest or more open woodland according to the habitats 
ax'ailable, are separated bv a gap between central Panama and 
central Costa Rica in which neither seems to occur. Tiie j:)oj:)u- 
lations on either side of this gap are \en- alike in male plumage. 



22 BREVIORA No. 409 

and in size. The females are more distinct, but some specimens 
of aglaiae resemble female hoynochrous closely. P. aglaiae is 
geographically and individually a \'ery \ariable species. Webster 
(1963), in a recent review of its variation, considered it con- 
specific with homochrous by modern standards, and the judg- 
ment seems sound. 

The evidence does not, however, warrant any further lumping 
within this group. It is obviously necessary to give specific 
status to minor vis-a-vis rufus. No critical test is possible of the 
status of minor vis-a-vis homochrous, but the constant plumage 
differences in both sexes suggest a similar degree of differentia- 
tion to that between such species as Lipaugus vocijerans and 
L. unirufus, whose ranges are also similar to those of minor and 
homochrous. It has been argued above that in the absence of 
evidence to the contrary it is best to give specific status to such 
fomis. 

The genus Cotinga. Six of the seven species usually recog- 
nized in this genus replace one another geographically: the 
Central American and northwest South American amabilis, 
ridgwayi and nattererii; maynana of the upper Amazon basin; 
cotinga of the lower Amazon-Guiana region; and 7naculata of 
eastern Brazil. (The seventh species, cayana, is more distinct 
morphologically and overlaps extensively with mayna?ia and 
cotinga.) The first three are ver\' similar to one another, and 
considered in isolation from the others might reasonably be 
treated as conspecific. C. cotinga and maculata are also very 
similar to one another, and Hellmayr suggested that they might 
be conspecific. They differ most conspicuously from the Central 
American forms by the deeper (not turquoise) shade of blue 
in the male and the more extensi\'e purple below. The sixth 
fomi, maynana. is at first glance rather distinct from the others. 
Hellma\r made the rather surprising suggestion that it might 
prove to be conspecific with the Central American forms, 
apparently on the basis of its similiar shade of turquoise blue. 

A more detailed analysis shows that these six forms exhibit a 
mosaic of characters that link one with another in many 
different ways f Table 1). The Central American group forms 
a unit, ha\ing in common six of the characters considered, and 
the two eastern forms constitute another unit, with se\en com- 
mon characters. C. maynana is certainly the most distinct form, 
having four peculiar characters, while none of the others has 
any character not found in at least one other form. In wing 
formula, however, maynana is closer to nattererii than the latter 



1973 



CLASSIFICATION OF COTINGIDAE AVES 



23 



ex 



be 
• S 

Hid 





n 




a, 


^^ 


o 






w 


rt 


hJ 


P3 


X! 


< 


(« 


h 


«*-i 




C 




« 




OJ 








rt 



u 
n 


























^ 




























































o 


■a 

c 


o 












-a 


o 


TS 




a 


_3 


g 


^^ 












T3 


CJ 




5 

to 


2 






t/5 














r3 




a. 


O 


>^ 










u 

i2 


> 

CJ 


rt 


3 
o 


p 
p 




OJ 




1J. 


a 


















fc 




—' 




r^ 


^ 


15 




3 


rt 


o 


rt 


— 






















^ 












«-^ 


























rt 












— H 


o 








CJ 




p 












CJ 




TS 






3 


> 

4-1 


^_ 












^J 




c; 






9 


'u 












3 




2 

CJ 

4-J 


p 


o 


_flj 


X 


a 


o 














-^ 




o 


^v' 


o 


CU 


r> 


3 




3 


^ 


C, 


rt 


o 






«-t 
















£ 










rt 


























p 
















oc 








t/i 


^ 








OJ 
'55 








. , 






c 
1 


'o 

3 

cr 

3 


4—1 


■4-1 




.S 3 


CJ 


CJ 


P 
I- 

o 


J 


^5 
-p ~ 

.SP o 


;j 


•= 


trt 


(^ 




C C 


a 


"5 


— 


~ 


t^ 


« p 








c 


O 










•a 










i^ 


2 


^ 


^ 








^ 












"o 

3 


«-) 


^^ 


«J 








•a 

OJ 


"re 


o 




C 


■^ 


rt 


>- 


J= 


^ 


J^ 






£ 


.^ 




> 




U' 


o 


u 


o 


u 




CJ 




u 




K- 


■*«j 


K 




r_v 




ra 


pj 




D 


^ 


^-' 







a 


3 


r^ 


CJ 


« 










C 


o 












C 


S 


3 




— 


"- 


-- 




• — 






















^ 


6 
























CJ 


t« 
































D 


c 


O 










"2 




rt 


•5" 




?% 


W5 


n 


4-1 


"tfl" 








bO 


^^ 


o 


^c 




a 


O 


4-> 




<u 








rs 


*j 


— 


c 


3 


3 


rt 


>^ 


J= 


^ 


ja! 




V 


E 

o 


a 


^ 


> 


f 


3 

4-1 


O 

x: 

4-1 


"o 










CJ 


-3 


■~ 


p 


^ 


— 


a. 


3 


3 




-^ 


r: 


O 


— 


-~ 


^ 




rt 


O 










CJ 

be 


CJ 
P3 


CJ 






^ 


*c 


^ 


■^ 


(A 

o 










> 


■3 




r- 


.5 


3 


n 


>-- 


^ 


^ 


^ 




V 


o 




r^ 


e 


c^ 


p 


u 


u 


u 




i) 


■^ 


p 




•-* 












cd 


ca 




3 








o 








OJ 


rt 










(^ 


c 






a 


-^ 


■" 


/-\ 


r^ 


S 


3 




^ 


~a 






-C 




















>, 


















CO 








<— 












11 






^ 




Si 
CJ 




rt 












e- 






.a 




















3 

a, 






■a 




1-. 
be 




r^ 












M^ 








•a 




r^ 


o 








4) 


o 






« 




^ 




t^ 


1 
















CJ 




ci 














s 


.5 






l^ 

VM 


bo 

c 






U1 

o 


W5 

_CJ 




u 




U-l 


*-i 






o 


T* 


n 




_a, £ 


^ 




Q 




o 


2 








r* 


C3 






rt 




,__ 












3 
O 


CJ 

C 


T3 
O 


2 

> 


— bJD 
.3 J 


D, 




o 

O 




on 


■q 






a 


OJ 


8 











24 BREVIORA No. 409 

is to the other two of the Central American group. Similarly, 
amabilis of the Central American group resembles the two 
eastern forms, cotinga and maculata, in its abbreviated 7th pri- 
mary, while ridgwayi, the remaining member of the Central 
American group, approaches them in the modification of its 
two outer primaries (but not in its 7th primary, which is 
normal) . 

It seems probable that the common ancestor of these six 
forms — perhaps during a period of contraction of the forests 
— was divided into a number of isolated populations, which 
evolved a number of small but consistent differences to which 
no obvious adaptive significance can now be attributed. The 
forest refuges suggested by Haffer (1970) provide obvious 
origins for five of the forms: amabilis in the Caribbean (Costa 
Rica) refuge, natter erii in the Choco refuge, maynana in either 
the Napo or the East Peru refuge, cotinga in the Guiana refuge, 
and macidata in the southeastern Brazilian refuge. The sixth 
form, ridgwayi, presumably e\ohed in a small forest refuge on 
the Pacific coast of Costa Rica and Panama, to which it is still 
confined. The rather exact geographical replacement of one 
form by another, and lack of any evidence of morphological 
intergradation, suggest that they have reached specific status to 
the extent of being reproducti\ely isolated, but that they are 
still not able to penetrate one another's ranges. 

In spite of the unequal differentiation of these forms, and 
the fact that they fall into two groups, with one isolated form, 
it is probably best to treat aU six as species, as has been the 
usual practice. 

The genus Xipholena. The status of the three forms has 
already been discussed, with the reasons for keeping them as 
three separate species. 

The genus Carpodectes. This genus consists of three allo- 
patric forms, of which two (nitidiis and antoniae) are very 
similar to one another while the third (hopkei) is a good deal 
more distinct. There is no doubt that together they constitute 
a zoogeographical species, but it is less clear whether they should 
be treated as one, two or three species. All three courses have 
been adopted or advocated. Slud (1964), however, gives some 
reasons for keeping nitidus and antoniae separate, and if this 
is done hopkei must also be gi\'en soecific rank. Such a treat- 
ment is in accord with the general policy adopted for other 
groups, as discussed above. 

The genus Cephalopterus. In accord with the general 



1973 CLASSIFICATION OF COTINGIDAE (aVES) 25 

policy adopted, as discussed above, the three allopatric forms of 
umbrellabirds should be kept as separate species. The differ- 
ences between them are at least as great as the differences 
between the overlapping forms of Xipholena and Phoenicircus. 

Literature Cited 

Ames, P. L. 1971. 1 he morphology of the syrinx in passerine birds. 

Bull. Peabody Mus. Nat. Hist., 37: 1-194, 21 pis. 
Bond, J. 1956. Additional notes on Peruvian birds II. Proc. .\cad. Nat. 
Sci. Philadelphia, 108: 227-247, 2 pis. 

. 1959. Fourth supplement to the check-list of birds of the West 

Indies (1956) . Acad. Nat. Sci. Philadelphia. 12 pp. 
Haffer, J. 1970. Art-Entstehung bei einigen Waldvogeln Amazoniens. 

J. Orn., Ill: 285-331. 
HELLMA'iR, C. E. 1929. Catalogue of birds of the Americas, Part VI. Field 

Mus. Nat., Hist., Zool. Ser., 13: 1-258. 
LowERY, G. H., AND J. P. O'NEILL. 1966. A new genus and species of 

cotinga from eastern Peru. Auk, 83: 1-9. 
Meyer de Schaiensee, R. 1966. The species of birds of South America. 
Narberth, Pa.: Livingston Pub. Co., Acad, Nat. Sci., xvii + 577 pp. 

. 1970. A guide to the birds of South .Vmerica. 

^V'vuncwood, Pa.: Livingston Publishing Co. xvi -(- 470 pp., 50 pis., 

21 figs. 

Olalla, a. M. 1943. .Algumas observagoes sobre a biologia das avcs e 

mamiferos sul-americanos. Pap. .\vuls. Dep. Zool. Sao Paulo, 3: 229-236. 

Olson, S. L. 1971. Taxonomic comments on tlie Emylaimidac. Ibis, 113: 

507-516. 
RiDGVVAY, R. 1907. The birds of North and Middle America, Part IV. 

Bull. U.S. Nat. Mus.. 50: 1-973, 34 pis. 
ScLATER, P. L. 1888. Catalogue of the birds in the British Museum, 

Vol. 14, xix + 494 pp., 26 pis. 
Sibley, C. G. 1970. A comparative study of the egg-white proteins of pas- 
serine birds. Bull. Peabody Mus. Nat. Hist., 32: 1-131, 38 figs. 
Skutch, a. F. 1969. Life histories of Central .\mcrican birtls HI. Pac. 
Coast Avifauna, 35: 1-580, 28 figs. 

. 1971. Life history of the Bright-rumpcd Attila AttUa 

spadiceus. Ibis, 113: 316-322. 
Slud, p. 1964. The birds of Costa Rica. Bull. Am. Mus. Nat. Hist.. 128: 

1-430, 2 pis., 1 fig. 
Snow, D. \V. 1971a. Display of the Pompadour Cotinga Xipholena puni- 
cea. Ibis, 113: 102-104. 

. 1971b. Evolutionarv aspects of fruit-eating by birds. Ibis, 

113: 194-202. 

1971c. Notes on die biology of (he Cock-of-the-rock (Rupi- 



cnla nipiroln) . ]. Orn., 112: 323-333. 

In press. Disti ibution, ecologv and evolution of the bell- 



birds {Procnias, Cotingidae) . Bull. Br. Mus. (Nat. Hist.) 



26 



BREVIORA 



No. 409 



VoLKER, O. 1952. Der Lipochrome in den Federn der Cotingiden. J. Orn., 

93: 122-129. 
Webster, J. D. 1963. A revision of the Rose-throated Becard. Condor, 

65: 383-399. 
Wetmore, a. 1926. Observations on the birds of Argentina, Paraguay, 

Uruguay, and Chile. Bull. U.S. Nat. Mus., 133: 1-448, 20 pis. 
. 1972. The Birds of the Republic of Panama, Part 3. Smiths. 

Misc. Coll., 150: 1-631. 1 pi., 48 figs. 



Appendix 

Annotated species list of the Cotingidae; treatment to be 

adopted in Peters' Check-list of Birds of the World. 

(Zoogeographical species bracketed) 



[ 



Phoenicircus carnifex 
Phoenicircus nigiicollis 
Laniisoma elegans 
Phibalura flavirostris 
Tijuca atra 
Carpornis cucullatus 
Carpornis melanocephalus 
Ampelion rubrocristata 
Ampelion rufaxilla 
Ampelion sclateri 
Ampelion stresemanni 
Pipreola riefferii 
Pipreola intermedia 
Pipreola arcuata 
Pipreola aureopectus 



Pipreola frontalis 
Pipreola chlorolepidota 
Pipreola formosa 
Pipreola whitelyi 
Ampelioides tschudii 
lodopleura fusca 
*- lodopleura isabellae 
lodopleura pipra 
Calyptura cristata 
Lipaugus subalaris 
Lipaugus cryptolophus 
Lipaugus fuscocinereus 
Lipaugus vociferans 
Lipaugus imirufus 
Lipaugus lanioides 
Lipaugus streptophorus 
Chirocylla uropygialis 
Pachyramphus viridis 
Pachyramphus versicolor 



See discussion, p. 16 



Formerly Doliornis sclateri 
Formerly Zaratornis stresemanni 
See discussion, p. 17 



Includes lubomirskii, jucunda and 

piilcha (p. 17) 

See discussion, p. 17 



See discussion, p. 18 



See discussion, p. 19 



^^2 n 



1973 



CLASSIFICATION OF COTINGIDAE (aVES) 



27 



Pachyramphus spodiurus 
Pachyramphus lufus 
Pachyramphus castaneus 
Pachyramphus cinnamomeus 
Pachyramphus polychopterus 
Pachyramphus marginatus 
..Pachyramphus albogriseus 
Pachyramphus major 
Pachyramphus surinamus 

{Pachyramphus aglaiae 
Pachyramphus minor 
Pachyramphus validus 
Pachyramphus niger 
Tityra cayana 
Tityra semifasciata 
Tityra inquisitor 
Porphyrolaema porphyrolaema 

{Cotinga amabilis 
Cotinga ridg^vayi 
Cotinga nattererii 
Cotinga maynana 
'Cotinga cotinga 
Cotinga maculata 
Cotinga cayana 

{Xipholena punicea 
Xipholena lamellipennis 
Xipholena atropurpurea 
{Carpodectes nitidus 
Carpodectes an teniae 
Carpodectes hopkei 
Conioptilon mcilhennyi 
Gymnoderus foetidus 
Haematoderus militaris 
Querula purpurata 
Pyroderus scutatus 

{Ccphalopterus glabricollis 
Cephalopterus ornatus 
Ccphalopterus penduliger 
Perissocephalus tricolor 
pProcnias tricarunculata 
\ Procnias alba 
I Procnias averano 
I- Procnias nudicollis 
'Rupicola rupicola 
^Rupicola peruviana 

Total: 79 
56 



See discussion, p. 20 
See discussion, p. 20 

See discussion, p. 20 



Includes homochrous; 
see discussion, p. 21 



Possibly the first six species could be 
considered as one zoogeogiaphical 
species (p. 22) . 



See discussion, p. 24 



See discussion, p. 24 



See discussion, p. 25 



All four Procnias species are closely 
related and replace one another al- 
most completely (Snow, in press) 



taxonomic species 
zoogeographical species 



ACME 
BOOK-'!'; :3.. INO. 

100 CAM3RI0GE STr?££T 
CHARUESIOWN. MASS. 



^^ Harvard MCZ Library 



_ 3 2044 066 302 795