ERNEST WILLI,A.MS

BULLEIN OF TlHE AMERICAN.MUSEUM" OF" "NATURAL HISTORY-'. VOLM 95: ARTICLE NEW OK 1950f'~ TESTUDO CUBENSIS AND THE EVOLUTION OF WESTERN HEMISPHERE TORTOISES Page 000_02 is blank TESTUDO CUBENSIS AND THE EVOLUTION OF WESTERN HEMISPHERE TORTOISES

ERNEST WILLIAMS Harvard University

BULLETIN OF THE AMERICAN MUSEUM OF NATURAL HISTORY VOLUME 95 : ARTICLE 1 NEW YORK 1950 BULLETIN OF THE AMERICAN MUSEUM OF NATURAL HISTORY Volume 95, article 1, pages 1-36, text figures 1, 2, plates 1-8, tables 1-3

Issued March 30, 1950

Price: $.75 a copy CONTENTS

INTRODUCTION ...... 7

Historical Background ...... 7

SYSTEMATIC DESCRIPTION ...... 9 DISCUSSION .10 Geologic Occurrence and Condition of the Material ...... 1 Character Analysis...... 11 Interrelationships of Western Hemisphere Testudines and the Phylogeny of tTestudo cubensis ...... 16 Miocene Testudines ...... 17 Oligocene Ancestors ...... 19 Eocene Prototestudines .20 Pliocene and Pleistocene Descendants. .21 The Phyletic Position of tTest-udo cubensis .. . . 23 The Nomenclature of Western Hemisphere Testudines .24~ The Neotropical Tortoises ...... 24~ The Nearctic Tortoises .25 Species and Genera Incertae Sedis .26

CHECI LIST OF NEW WORLD TESTUDINES AND SUPPOSED TESTUDINES 29'

SUMMARY ...... 32

ACKNOWLEDGMENTS.33

BIBLIOGRAPHY ...... 34: 99 4-: O Its i. L e a a tv 4-

0 . D eV a CdO ,p O a cdO 0

0

Co)0

bcn .cn J &C O44._t,

96- *> ;-1

0)

c :a v 8)

co~

c,

Co

3e 0 io Q

0) c0 co Co (,rn c O .c -C^o0-

6 INTRODUCTION IN THE MORE THAN 80 years since the descrip- living P. decussata. From Jamaica a single tion by Leidy of testudinate fossil renains partial shell records the existence of Pseu- from Cuba and Sombrero Island, few addi- demys terrapen contemporaneously with cer- tions have been made to the knowledge of tain Jamaican fossil mammals sometimes this West Indian fauna. During this period regarded as older than most other so-called enough material has accumulated at the Pleistocene West Indian mammals, while American Museum of Natural History to from Puerto Rico a great number of frag- demonstrate that the testudinate fauna of the ments give evidence of a large and thick- West Indian Pleistocene was much richer shelled emydine not at present generically than has ever been suspected. From Cuba determinable. Finally from the caves of Mona there is now available, in part from the same Island elements of a turtle have been col- deposits that provided the ground sloths lected which in limbs and cervical vertebrae named by Matthew (1931), not only consider- closely resemble Testudo but from the palate able material of Leidy's tTestudo cubensis and certain features of the shell must belong but a number of partial shells and fragments to an undescribed genus of uncertairn relation- of a tPseudemys evidently related to the still ships. HISTORICAL BACKGROUND Vertebrate fossils were first found in Cuba partly studied by him and partly entrusted in April, 1860, in a thermal spring at the to the American Museum of Natural History Bafnos de Ciego Montero by a student at the for more thorough research in conjunction University of Havana, Jose de Figueroa. He with further exploitation of the localities. brought them to Felipe Poey, who presented In 1911 Barnum Brown in association with them on the fifteenth of September, 1861, de la Torre completed collections at the before the Academy of Sciences at Havana. locality in the Sierra de Jatibonico and also Poey sent several of the fossils to Leidy who obtained a large sample from Ciego Montero. in 1868 described tTestudo cubensis from a In 1918 Brown completed the excavation of single partial pleural' plate found at Ciego the Ciego Montero spring. At both localities Montero. Carlos de la Torre, who succeeded material was gathered not only of t T. cubensis Poey in the chair of zoology at Havana, but also of tPseudemys cf. P. decussata. continued collecting at the Ciego Montero Three other localities in Cuba have yielded spring. He also devoted much time to a new testudinate remains. A fragment of a Pseu- locality in the Sierra de Jatibonico, a fissure demys plastron was found by H. E. Anthony discovered by a refugee Cuban patriot, in 1917 in a cave at Daiquiri, Oriente Prov- Ramon Gonzalez, during the Cuban fight for ince, and fragments of a small specimen of the liberation from Spain. In 1910 de la Torre same genus were collected by Barnum Brown reported some of his findings to the Inter- in the Cueva de los Machos near Cienfuegos, national Geological Congress at Stockholm. Santa Clara Province. In a tar pit near Hato The material collected by de la Torre was Nuevo, Matanzas Province, Roy E. Dicker- son and P. J. Bermudez in 1933 found, be- 1 The terminology here employed endeavors to be sides some bones of domestic animals, ground consistent in distinguishing horny scutes from bony sloth claws, and small rodent bones, a few plates. The following are the corresponding terms for the sets of elements in the carapace: plates and a femur of tT. cubensis. The lat- SCUTES PLATES ter fragments from a source apparently vertebral neural deserving of further investigation were do- costal pleural nated to the American Museum of Natural marginal peripheral nuchal nuchal History by Dickerson. supracaudal pygal Leidy, in the same brief note in which he Plastral nomenclature is already consistent and described tT. cubensis, erected the name unambiguous. tEmys sombrerensis for a partial plastron 7 BULLETIN AMERICAN MUSEIUM OF NATURAL HISTORY VOL. 95 found during excavations for fertilizer on at the moment of publication of Schmidt's tiny Sombrero Island. Leidy was uncertain paper, Anthony was in the process of explor- whether the form was an Emys or a Testudo. ing the caves of Mona. The only fossil verte- The type is now lost, and the description is brate he found, aside from a very few bones not determinable. The specimen was not of a small mammal (tlsolobodon) probably Emys as the genus is now understood. That of human importation, was a new and peculiar it was Pseudemys is not probable either; genus of tortoise. the described depth of the xiphiplastral notch Two years previously from Puerto Rico suggests Testudo as more likely. itself (the exact locality and circumstances In 1920 a third island was added to the list of the find uncertain) still another form was of those on which fossil turtles were to be added to the growing collection at the Ameri- found. In that year H. E. Anthony procured can Museum of Natural History. This form, a shell of Pseudemys terrapen from the breccia the gift of Sefnor Rabell Cabrera, was repre- of the same cave in Jamaica in which he sented by numerous fragments of plastron, found tClidomys, tSpirodontomys, tSpeoxe- pelvis, and of the buttress region of the cara- nus, and tAlterodon (Anthony, 1920a). pace, but was unfortunately so incomplete The abundant mammalian remains in West as to afford no generic characters. Indian caves reported principally by Anthony Of this material only tTestudo cubensis stirred considerable interest at this time. In Leidy is described in the present paper, 1926 K. P. Schmidt, discussing the modern special emphasis being placed on the relation- herpetofauna (in which no turtles are in- ships of that form and on its position in the cluded) of the small island of Mona off phylogeny of Western Hemisphere tortoises. Puerto Rico, suggested that the numerous It is intended to discuss the other forms in limestone caves on that island would prob- succeeding papers. ably afford much of interest. As it happened, SYSTEMATIC DESCRIPTION ORDER TESTUDINATA ADDITIONAL MATERIAL: Fragments of cara- FAMILY TESTUDINIDAE pace, plastron, and limb bones from the SUBFAMILY TESTUDININAE Casimba locality in the collection of the American Museum of Natural History; frag- DEFINITION: Testudinids with never more ments (Moreno collection) from the Casimba than two phalanges per digit on fore or hind locality in the collections of the Museum of foot. Comparative Zo6logy, Harvard College; a GENUS TESTUDO LINNi, 1758 femur and plate fragments from the tar GENOTYPE: Testudo graeca Linn6, 1758. pit in Matanzas Province in the collections NEW RESTRICTED DEFINITION: A genus of of the American Museum of Natural History. Testudininae world-wide in distribution, DIAGNOSIS: A species of 7iestudo distin- with the alveolar surface of the maxillae guished by having the borders of the verte- broad and usually with one to three rather bral and costal scutes raised into sharp- distinct ridges, never with a ridge at the crested ridges on the plates and by having symphysis of the alveolar surfaces of the the humerus markedly compressed in the premaxillae, palatines and vomers forming a plane of the head and with a deep medial deep trough, cervical vertebrae rather slender, pit for the latissimus dorsi. carapace never hinged, centrum of first dorsal ADDITIONAL CHARACTERS: Size comparable vertebra not elongate. to that of Gal6apagos species. Shell rather elongate, thin and weak in central portions tTestudo cubensis Leidy, 1868 of carapace and plastron. Areas underlying Testudo cubensis LEIDY, 1868, Proc. Acad. Nat. the marginal, gular, and anal scutes more or Sci. Philadelphia, vol. 20, p. 179. less thickened. Neural and pleural plates TYPE: Part of a first right pleural plate, probably partly discontinuous with vacuities Academy of Natural Sciences of Philadelphia at the sutural junctions. Nuchal scute ab- No. 8923. sent, the first marginal scutes in broad con- HORIZON: Pleistocene? (see below). tact. Marginal scutes often, and plastral TYPE LOCALITY: The Chapepote spring at scutes mostly, with impressed borders. Pos- Bafios de Ciego Montero, Santa Clara Prov- terior free margin of the carapace strongly ince, Cuba. recurved. Anterior rim moderately recurved. ADDITIONAL LoCALITIES: Casimba de Jati- Free margins markedly thin, dentate. Anteri- bonico, Santa Clara Province, Cuba; a tar or lobe of plastron flat and long. Epiplastral pit near Hato Nuevo, Matanzas Province, lip either projecting and quadrate or low and Cuba. rounded. Gular scutes paired, never encroach- TOPOTYPES: The American Museum of ing on the entoplastron. Pectoral scutes, so Natural History Nos. 6201, left posterior far as known, not narrowed. Posterior lobe quadrant of plastron; 6202, anterior lobe of of plastron long, with somewhat convex plastron; 6204, part of nuchal plate; 6209, lateral margins. Femoral-anal boundary first pleural; 6211, sixth and seventh pleurals; transverse. Anal scutes small. Anal median 6227, pygal; and numerous other fragments sulcus short. Xiphiplastral notch very shallow of carapace, plastron, and limb bones. or wholly absent.

9 DISCUSSION GEOLOGIC OCCURRENCE AND CONDITION OF THE MATERIAL THE TYPE PLEURAL PLATE of tT. cubensis was listed are represented by dissociated bones. obtained at the Chapepote thermal spring, Barnum Brown (1919) remarks that in his one of three such springs located in a bend two expeditions the only bones found to- of the Analla River at the Bafios de Ciego gether were a few crocodile vertebrae (1911) Montero, Santa Clara Province, 30 miles and a few fMegalocnus vertebrae (1918). northwest of Cienfuegos, Cuba. The deposits The same species occurred from top to bottom were black mud at the bottom of a pool 10 of the deposits, except that the earliest feet deep which required much pumping to layers contained the amphibian and most of clear. The fossiliferous layers were believed the small bird, snake, and rodent bones. by Barnum Brown to result from periodic The age of the deposits is difficult to esti- overflows of the Analla River. Heavy rains mate. The ground sloths are usually listed still bring shells and river turtles into the as Pleistocene in age (Simpson, 1945). tNeso- spring (Brown, 1913). The accumulations of phontes is regarded as "Pleistocene or Re- organic material included bones, river shells, cent." tGeocapromys columbianus, mentioned crustaceans, trees, pine cones, nuts, and by Matthew as occurring rarely in these de- leaves. posits, is, according to Brown, at least as old The vertebrate fauna found is listed in as the sloth bones. The fossil crocodile de table 1 (taken with only minor changes from scribed by Leidy (1868) as tCrocodilus pri'sti a manuscript of Matthew). The forms so nus from a single vertebra from these de TABLE 1 PLEISTOCENE AND SUB-RECENT CUBAN FAUNAS Ciego Montero CasimbaCsmaCaves(Various) EDENTATA Megalonychidae Megalocnus rodens Abundant Abundant Species 2a ? Scarce Rare Species 3 - Rare Mesocnus browini Scarce Scarce torrei Rare Common Miocnus antillensis Scarce Microcnus gliriformes Rare Rare RODENTIA Octodontidae Capromys sp. Geocapromys columbianus Rare Rare Abundant Boromys sp. Abundant INSECTIVORA Nesophontidae Nesophontes micrus Rare Common TESTUDINATA Testudo cubenssis Abundant Abundant Pseudemys cf. P. decussata Abundant Abundant Present CROCODILIA Crocodylus rhombifer Abundant Abundant AVES, SAURIA, SALIENTIA

& Names are given in Matthew's MS, but they cannot be repeated here since they would be nomina nuda. b Common in uppermost levels only, not in older deposits. 10 1950 WILLIAMS: TESTUDO CUBENSIS 11 posits is now regarded (on Barbour's author- of a skull nor any vertebrae were recovered. ity; Matthew, 1918, p. 662) as identical with Only a few limb bones, broken and water still living Crocodylus rhombifer; the largest worn, are known. fossil specimens, however, much exceeded The deposits at the fissure or "casimba" in any modern specimens in size. the Sierra de Jatibonico in the eastern part Probably significant is the evidence of the of Santa Clara Province are in many ways fossil flora. Pine cones' were found abundant- very similar to those of Ciego Montero. Here ly in the spring deposits, although at present the fossils were found in a fissure, about the nearest pines are known from Pinar del 15 feet deep and 300 feet long, opening into Rio and Oriente Provinces, more than 150 a small cave. The fragments tend to be of miles to the west and east, respectively. This smaller individuals than those of Ciego is certainly indicative of climatic change. Montero (see below). Pines may have been characteristic of the A few fragments of tT. cubensis, including colder periods of the Pleistocene. Brown a perfect femur, parts of a peripheral and of regarded the deposits as early Pleistocene, a nuchal, have been obtained from a tar and they are usually so regarded in standard pit near Hato Nuevo, Matanzas Province. geological works (Schuchert, 1935; Wood et As in the case of the Banos de Ciego Montero, al., 1941). bones of some recent (domestic) animals It is true that Wetmore's (1928) discovery have been found in addition to ground sloth, of Gallus among the bird remains at Ciego tortoise, and crocodile. Fossil invertebrates Montero indicates that at least the upper- taken from the tar pit were identified by most levels of the deposit are quite recent. Carlos de la Torre and are believed to indi- This, however, is not remarkable, since the cate Pleistocene age for the bulk of the fauna. Chapepote sprinlg has remained open to Taken together, the collections from Ciego contamination of this sort. The bones of Montero, Casimba, and Hato Nuevo include Gallus also are, as Wetmore remarks, quite fragments of individuals covering a consider- different in color from the older remains. able size range, but no single individual is The nature of the deposit has had impor- even approximately complete. Such material tant consequences for the condition of the is, of course, not entirely satisfactory, but fossils. The elements of tT. cubensss are as it has several advantages over any single completely dissociated as those of the other perfect shell. Both age and individual varia- animals. The largest parts found together are tion are indicated. A population is repre- the anterior lobe of one plastron and the left sented, and those characters of the species posterior quadrant of another. All the plates that can be determined are the more useful of the carapace were found wholly isolated because they are more securely based. or, at most, parts of few together. No portion CHARACTER ANALYSIS Leidy's (1868) description of this species, ration 16 lines. The articular border with the based upon a single, partial, and apparently first vertebral plate is 30 lines; that with the water-worn pleural plate, was as follows: second vertebral plate 14 lines. The articular "Testudo cubensis n. s. border from the first vertebral plate to the "Indicated by a portion of what I suppose lateral groove defining the first vertebral scute to be the first costal plate of the right side. is convex forward and inward and 14 lines in It is marked by the lateral borders of the first a direct line. A strong costal process projects and second vertebral scutes and their trans- from the under part of the plate nearly paral- verse separation. At the former borders the lel with its length. The surfaces corresponding plate is 51 lines[21 and along the latter sepa- with the vertebral scutes are somewhat de- pressed, and are generally everywhere nearly 'Mostly referable to the liv'ing species Pinus carr- or without so baea, in the opinion of Brother Lion who generously smooth markings conspicuous consented to examine them. One cone, however, may as to affect the investing scutes. The greatest represent a new species. thickness of the bone is where it articulated 2 line is one-twelfth ofan inch, or a little over 2 mm. with the first and second marginal plates, 12 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 95 measuring from 3A to 4' lines, and where tortoises; he found it in a specimen of T. thinnest it measures only one line. radiata and one of T. abingdoni (=T. elephan- "No living Testudo, I believe, at present topus' of this paper) at the United States inhabits Cuba, and the fossil probably indi- National Museum (U.S.N.M. Nos. 29214 and cates a species contemporary with the Mega- 29269, respectively, both on exhibition). lonyx." The present author has verified the exis- Leidy's type, though not so labeled, is in tence of this character on the pleurals and the possession of the Academy of Natural neurals of fT. laticunea and fT. insolitus and Sciences of Philadelphia, bearing the cata- on the nuchal of tT. quadratus, the types of logue number 8923. It is here figured in plate which are in the American Museum of Natu- 1. ral History, and on the living species cited Fifty years after Leidy's description the by Gilmore in the United States National name was revived in the literature by Mat- Museum. He has seen it also on the gular thew, without, however, any formal redescrip- median sulcus of the type of fT. orthopygia tion and only in preliminary reports (1918, (Lower Pliocene of North America) in the 1919) on the Cuban fossils collected by Bar- American Museum of Natural History. It is num Brown. Matthew's brief mentions were well developed over the whole carapace of based on the much more adequate material a Gopherus polyphemus in the Museum of here again reported on, and he characterized Comparative Zoology. It is rather frequently the species for the first time in a recognizable present in Testudo pardalis, mostly on fashion. He had in preparation a manuscript peripherars and posterior pleurals, and tends describing the new material in detail. This to be generally though feebly present in manuscript has been at hand and of consider- T. denticulata and T. radiata. able service in the course of the present study; The rather surprisingly widespread occur- it has not been followed consistently and both rence of this character, while disposing of it additions and corrections have been made. as a feature isolating tT. cubensis from other The character that Matthew believed iso- tortoises, does not diminish its utility as a lated tTestudo cubensts from all other species species character. It is not so much the pres- of the genus was the fact that the margins ence of raised sulci as their degree of expres- of the horny shields were marked out on the sion that is the peculiar mark of tT. cubensis. bony plates not by the usual simple grooves The raised sulci of other fossil and recent but by sharp raised ridges enclosing deep tortoises are feebly developed and not notice- median furrows (see pl. 4, figs. 2, 4). This is able without close examination of the speci- certainly the most striking feature of the men. The raised sulci of fT. cubensis, on the. species, and in fact the majority of the pleural contrary, are very obvious. The ridging is and neural plates can be recognized as belong- usually high and narrow; it varies, however, ing to this species on this character alone. considerably and may, especially on the This feature, however, is not unique; it peripherals, be broad and shallow, fading is no longer possible to say, as Matthew away into grooves without ridging. In other (1919, p. 170) did, that "this alone would forms, also, the raised sulci tend to be most forbid any close relationship with any of the clearly expressed on the nuchal plate or on species I have compared." Matthew and the peripherals or on the plastron. In tT. Granger (1923) discovered the same char- cubensis the raised sulci are best marked on acter in tTestudo insolitus of the Lower the neural and pleural plates, infrequently Oligocene (Ardyn Obo) of Mongolia. Gilmore present and then, as just noted, less developed (1931), in redescribing tT. insolitus, noticed on the peripherals, absent altogether from that the same peculiar feature had already the nuchal plate, in the two instances in been described in 1908 by 0. P. Hay in tTes- which that is known, and from the plastron. tudo laticunea of the Chadron beds of the Clearly the raised sulci of tT. cubensis show North American Oligocene (and in fT. quad- the fixation and elaboration of a chaxacter ratus of the same age as well as in tT. edae of 1 Testudo elephantopus is used in the sense of Roths- the Lower Miocene). Gilmore further pointed child (1931), Flower (1937), and Mertens (1942) to in- out that this character is known in living clude all Galipagos "species." 1950 WILLIAMS: TESTUDO CUBENSIS 13, frequently, but irregularly and poorly, de- thickened, and the amount and nature of veloped in other tortoises. the thickening in each of these regions vary The other characters of tT. cubensis are notably with the individual or perhaps with less striking, but in combination they make the age and sex of the individual. it a very well-marked species: 4. The shape of the pleural plates in 1. The absence of a nuchal scute, seen in modern Testudo and Gopherus as in certain tT. cubensis, is a character of many advanced emydines, e.g., some species of Pseudemys, species of Testudo. All the three living species is that of alternating wedges. tT. cubensis is assigned to the genus Testudo in the Western quite typical in this respect. A number of Hemisphere (all of them Neotropical) agree the testudines of the North American Terti- in this regard (T. denticulata, T. chilensis, ary lacked this advanced character. T. elephantopus). Likewise tT. praestans and 5. The shape of the neural plates, regarded tT. gringorum of the South American Pliocene as of considerable taxonomic importance by and Miocene, respectively, lack the nuchal Hay and others, is not readily ascertained in (Rovereto, 1914; Simpson, 1942). In strong tT. cubensis. The thinning of the neural plates contrast all the species of the North American at their junction with the pleural plates and Tertiary retained this shield, as also does the with each other, combined with the fractures modern genus Gopherus of North America. inevitable in fragile, dissociated, water- 2. Much emphasis was placed by Matthew carried material, effectively prevents knowl- (1919 and MS) on the thinness of the shell edge of the shape of these bones in the intact and especially on the thinning of the neural carapace. The best-preserved neural plate and pleural plates at their junction with might reasonably be interpreted as quadrate. one another. He believed that irregular vacui- 6. Hay has called attention to the reduc- ties must have existed at these sutural junc- tion of the rib heads in advanced testudines tions. The thinness of the shell seems to (for example, T. denticulata) and to their have been a very constant character of the frequent origin in whole or part from the species and, like the large size, is a point neural plates. In this regard, as Matthew (in of resemblance to Galipagos tortoises. It is MS) noticed, tT. cubensis is quite primitive. not, however, as Matthew himself remarked, The rib heads are strong and arise well comparable in detail to the thinning of the within the borders of the pleural plates. shell in Galapagos tortoises and seems to 7. The shape of the pygal was used by Hay have been independently acquired. Thinning to define several species of the North Ameri- of the shell and consequent fontanelles are can Tertiary. That of tT. cubensis, however, not infrequent in testudines. The extreme is not distinctive and resembles that of many of such a process is reached in Testudo (Mala- other species, fossil and recent. The supra- cochoersus) tornieri (Proctor, 1922), but it is caudal scute was apparently undivided, as often carried very far in Gopherus polyphe- is usual in Testudo except in the subgenera mus, in which only a very fragile and in- Manouria and Malacochoersus and in Testudo complete bony shell is sometimes preserved hermani, where it is divided as in emydines. (Gadow, 1901; Proctor, 1922, specimen at 8. The recurved, thin, and dentate margin the Chicago Natural History Museum, speci- of the anterior and posterior free edges of men on exhibition at the American Museum the carapace of tT. cubensis impressed of Natural History). Giant forms, also, as Matthew as quite distinct. The known frag- can be seen not only on individuals from the ments of these areas in tfT. cubensis certainly Galapagos but on those from the islands of show a condition quite different from the the Indian Ocean, seem to be very susceptible straight, blunt, smooth margin of such a to weakening of the bony plates. Unfor- form as T. denticulata. The condition of tT. tunately the extent to which dietary defici- cubensis is perhaps most closely approached encies play a part in this process is not known by T. chilensis (specimens examined in the (Loveridge, 1947). Museum of Comparative Zoology, see also 3. In tT. cubens's, as in many other tor- the figure in Gray, 1870). The importance of toises, the areas covered by the marginal, this character has probably been much exag- gular, and anal scutes are characteristically gerated. In three specimens of Terrapene 14 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 95 carolina (author's personal collection), in one present material is fragmentary, but enough case the margin is very distinctly dentate remains of the axillary region of several and recurved, in another the margin is re- plastra to show that tT. cubensis has curved but only obscurely dentate, and in remained relatively primitive, with the pec- still another the margin is very little recurved torals only moderately narrowed. but very clearly dentate. Other examples 14. In tT. cubensis the gular scutes are could be cited of the individual variability widely removed from the entoplastron in all of this character. the several specimens of the anterior plastral 9. The elongation of the carapace and region that are available. This character plastron of tT. cubensis is a character in has not previously received much attention; which it strongly resembles tT. denticulata. It it seems, however, to be a good species char- is also a feature in which tT. cubensis differs acter for tT. cubensis. Primitively the gulars from most of the species of the North Ameri- extended onto the entoplastron. Except in can Cenozoic (except tT. emiliae of the Lower tT. emiliae (Lower Miocene of South Dakota) Miocene). they did so in all the species of the North 10. The depth of the xiphiplastral notch American Tertiary and in tT. crassiscutata is a character much used by Hay and others and in other less well known Pleistocene to define species of the North American forms. In tT. praestans and tT. gringorum of Cenozoic. The notch is almost always pres- the South American Tertiary the condition ent but varies in depth. In tT. cubensis it is is the same, and in T. chilensis the gulars minimal to absent. In tTestudo osborniana in the figure of Freiberg (1938, fig. 5) at least (Pawnee Creek Miocene) it is nearly as much touch the entoplastron. In T. elephantopus reduced. In living testudines comparable the gulars similarly at least touch the ento- conditions exist in the genus Kinixys of plastron. However, there is some evidence Africa, for example, and in the giant tortoises that the character is not always constant of both the Old and New Worlds. within species. In four specimens of Testudo 11. The position and direction of the anal denticulata in the American Museum of sulcus are also generally considered to be of Natural History the gulars in one case en- taxonomic importance. In tT. cubensis they croach considerably on the entoplastron, in are very characteristic, being not oblique another case they barely reach it, and in two with the inner angle of junction of the sulci others they are as far from the entoplastron of opposite sides anteriorly directed, as in as in tT. cubensis. In Gopherus agassizi the the tortoises of the North American Tertiary gulars at least quite frequently are out of (tT. osborniana, for example) and most contact with the entoplastron, and in this modern Testudo and Gopherus, but oblique, species this phenomenon seems to be corre- with the inner angle of junction posteriorly lated with a lengthening of the anterior directed. The median sulcus between the plastral lobe. Probably it was also the length- anal scutes is consequently very short in tT. ening of the anterior lobe of the plastron cubensis. that brought about the retraction of the 12. The shape of the entoplastron has gulars from the entoplastron in T. denticulata, occasionally been considered of taxonomic tT. emiliae, and tT. cubensss. value. In the one example in tT. cubensis in 15. Epiplastral bones are the most fre- which it is preserved, it is pentagonal, with quently preserved, precisely identifiable ele- internal bosses, and not much broader than ments of tT. cubensis. Nine epiplastra repre- long. The breadth of the entoplastron is senting three growth stages are present in probably a more useful taxonomic character the material studied. This series is of some than its precise shape, but it also must be importance in the evaluation of a character used with caution. A limited check of various upon which much reliance was placed by emydines and testudines seems to show con- Hay, namely, the projection or lack of pro- siderable indivridual variation. jection of the gular region of the plastron. 13. The pectoral scute tends in many The smaller epiplastra show considerable specialized testudines to be much narrowed. variation from individual to individual in the tT. osborniana is an extreme example. The degree of projection of the gular region. The 1950 WILLIAMS: TESTUDO CUBENSIS is series exhibits a complete transition from a from animals of the same or nearly the same gular region conforming completely to the size; unfortunately proof that they are from general contour of the anterior plastral lobe one animal is lacking, since a middle portion to a gular region markedly and abruptly of the plastron that might unite them is projecting. missing. Size and age are presumably not note- On the other hand it may be that the worthy factors. A middle-sized pair of epi- differences in the form of the gular region iii plastra from the Casimba locality (M.C.Z. tT. cubensis are not secondary sexual char- No. 1901) show a stage in the differentiation acters. In a series of T. denticulata in the of the gular region only a little more advanced American Museum of Natural History very than the least differentiated of the smaller marked differences in the shape and especially epiplastra. In strong contrast A.M.N.H. No. in the thickness of the gular region are not 6235 from Ciego Montero shows the gular sexual in origin; at least none of the conditions region projecting very prominently, as does is consistently correlated with the marked the largest specimen, A.M.N.H. No. 6202, plastral concavity that has been taken as a also from Ciego Montero. sign of the male. This very striking difference in the degree If sexual difference is not involved, there of gular projection within species is paralleled is another possibility to be considered: it is in modern Gopherus, and in that genus, as conceivable that the materials from Ciego Grant (1936) and Woodbury and Hardy Montero and from Casimba are at least (1948) have noticed, is a secondary sexual subspecifically distinct. All the referred character. In Gopherus agassizii, according material from Casimba is consistently smaller to both Grant and Woodbury and Hardy, than the topotypical Ciego Montero tTw the gular projection is longer, larger, and cubensis. The gular region in none of the more upturned in the males, especially as Casimba specimens projects so much as they get older. The difference may be quite in the two large epiplastra from Ciego extreme, as the present author has determined Montero. on a male Gopherus agassizii personally There is precedent for recognition of a captured at Laguna, Arizona, and on material taxonomic difference between materials from in the American Museum of Natural History. the two localities in Matthew's action in a A series of Gopherus berlandieri examined in parallel situation involving the ground sloth the Kansas Natural History Museum show remains. Matthew (MS) erected a distinct the sexual characters of the gular region in subspecies of tMegalocnus rodens for the that species. Characteristically bifurcate in slightly differing material at the Casimba both sexes in this species, the gular region locality and also erected two full species of in G. berlandieri males is longer, more sharply tMegalocnus for some small jaws from Casim.- projecting, more deeply forked, and very ba. One of Matthew's genera of smaller much thickened. There is some individual ground sloths (tMiocnus) is likewise known variation in all these features, but the con- only from Casimba. trast between the sexes is almost always There is therefore some warrant for believ.- sharp and unmistakable. ing that Casimba and Ciego Montero had It is possible that in tT. cubensis also a different faunas in the Pleistocene. It is strongly projecting gular region is an indica- also, of course, uncertain that the two faunas tion that the specimen was a male. In fact, are strictly contemporaneous. It is possible A.M.N.H. No. 6202, the element exhibiting therefore that distinct subspecies of tT.cuben- the most prominent gular projection, per- sis might characterize the two areas. How- haps should be associated with A.M.N.H. ever, an alternative explanation of the differ- 6201, the left posterior quadrant of a plastron, ence in the fossils at the two localities is which is definitely that of a male as shown by possible: it may be that the conditions of the characteristic strong concavity (the area deposition in the fissure springs at Casimba under the anal scutes is also strongly thick- de Jatibonico have introduced a definite bias ened as in males of Gopherus berlandieri and in the matter of the size of the material G. agassizii). The two elements are certainly swept into the deposit, so that the fossils 16 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 95 are not a truly random sample of the popula- equally quadrangular but much broader than tions in the area. It is this hypothesis that high. They do not differ much from compa- the author provisionally adopts. The differ- rable vertebrae of T. denticulata or T. pardalis, ences in size between the Casimba and the probably more closely resembling the former. Ciego Montero materials of tT. cubensis are A specific character may be the extreme considered to be ontogenetic, and the differ- dorsoventral thickness of the neural laminae ences in gular prominence are interpreted as in certain of the anterior vertebrae. There is secondary sexual difference. no evidence of a heavy, closely united dermal 16. The limbs of tT. cubensis are stout armor covering the tail as described by Hay and heavy, resembling in this regard those (1908, figs. 560, 603) for tT. osborniana and of tT. osborniana or of tT. orthopygia rather tT. orthopygia, nor is there any hint of an than those of tT. laticunea or tT. vaga among abrupt, paddle-shaped termination to the the tortoises of the North American Tertiary. tail with several fused vertebrae as in fT. The two known humeri (both from Ciego orthopygia (Hay, 1908, fig. 601). Montero) are strongly compressed in the plane of the head; the depth of the shaft in TABLE 2 that plane is about one and one-half times SUMMARY OF THE CHARACTERS OF the depth in the plane at a right angle to tTestudo cubensis that of the head. In T. denticulata, T. ele- Nuchal scute absent phantopus, and the liv-ing species of Gopherus, Vertebral and costal sulci on raised ridges as well as in the Old World species examined Marginal and plastral sulci usually in simple (T. pardalis, T. elegans), the ratio of the same grooves two dimensions is very nearly 1:1. The femur Thinning of shell greatest at neural-pleural junc- of tT. cubensis from the tar pit in Matanzas tions and in midplastron Province shows the same sort and degree of Areas of thickening under marginal, anal, and compression as the humeri. The femur of gular scutes osborniana No. Rib heads strong, arising well within pleural bor- the type of tT. (A.M.N.H. ders 5868), as Hay has already stated (1908, p. Shape of carapace markedly elongate 425), exhibits the same condition. Xiphiplastral notch minimal or absent There is another peculiarity of the humerus Free carapace margins recurved, thin, and den- of tT. cubensis; it has a large and deep pit tate on the medial surface for the insertion of the Femoral-anal sulcus transverse or posteriorly di- latissimus dorsi. This pit occurs as an ab- rected at its middle normality in T. denticulata (one specimen in Anal median sulcus very short three in the American Museum of Natural Pectoral scutes expanding gradually at axillae History) and is present in a somewhat less Gular scutes never encroaching on entoplastron still visible form in Humerus short, stout, strongly compressed in distinct but plainly many plane of head, with large and deep pit for Galipagos tortoises (many specimens in the latissimus dorsi American Museum of Natural History). This Femur short, stout, strongly compressed in plane pit has not been seen in other tortoises. of head 17. A few caudal vertebrae of tT. cubensis Caudal vertebrae normally testudine except for are present and are typically testudine, the extreme dorsoventral thickness of neural anterior ones being quadrangluar and tending lamina of some anterior caudals which may to be higher than. broad, the posterior ones exceed dorsoventral thickness of centrum INTERRELATIONSHIPS OF WESTERN HEMISPHERE TESTUDINES AND THE PHYLOGENY OF tTESTUDO CUBENSIS Since the relationships of tTestudo cubensis analysis of the relationships and phylogeny are rather certainly with testudines of the of fTestudo cubensis is a task presenting a Western Hemisphere, we may reasonably number of difficulties and requiring an ex- limit our attention to these forms. However, tended discussion. There are already more even with this admittedly serious limitation, than 70 species of testudines assigned to eight 1950 WILLIAMS: TESTUDO CUBENSIS 17 genera described from the Cenozoic of the either side." This contrasts with Hay's New World. The larger number of these (1908) description of tTestudo osborniana: are known only from single shells or shell "There is no ridge occupying the midline fragments. In only 11 species is the skull along the symphysis of the premaxillae; known. In some of these and in a few others hence this species cannot be regarded as a limbs and limb girdles are known. Further, Gopherus. The masticatory ridges of the right the types of these numerous described species and left sides are separated by a deep longi- are very widely scattered, so that no worker tudinal groove." The conditions here de- since Hay has seen even a majority of them. scribed in two fossil species are identically (In the case of a few South American species those used to distinguish Gopherus and it has not been possible to locate even the Testudo in living forms. original descriptions, and they are known The transformation of one type of skull only from citations by Amegh'ino and other into the other would seem to be relatively authors.) simple. A ridge exists on the vomer in In combination this multiplicity of species Testudo; lateral maxillary ridges are also and the inadequacy or inaccessibility of present. All of these ridges approach the material make it inevitable that an assessment borders of a premaxillary pit. To attain the of the phyletic relationships of Western tG. brevisterna condition all that is necessary Hemisphere testudines will be even more is an extension of the vomerine ridge forward theoretical and tentative than phyletic judg- across the premaxillary pit.' The Gopherus ments usually are. It is felt, however, that a palatal pattern is thus only a slightly special- carefully framed set of hypotheses suggesting ized derivative of the typical Testudo con- relationships that can be tested by those with dition. access to the material will add something The skull difference is in fact trivial, but pertinent and perhaps decisive to the pres- it is of considerable antiquity and cannot be ent mass of non-integrated detail. disregarded. Furthermore when the series of fossil tortoises is examined carefully enough, MIocENE TESTUDINES shell characters are found to be associated It will be best to begin the analysis of with the diagnostic skull difference. The Western Hemisphere testudines in the Mio- forms of the North American Miocene can, cene, since only for that period have any for example, be separated into two series, number of species been described from really one of which may be called the tTestudo os- good material-skulls, limbs, and quite per- borniana series and the other the tGopherus fect shells. brevisterna series. In the first series the three Seventeen Miocene North American forms forms with Testudo-like skulls are readily are named in the literature, all allocated by arrayed, and in the other the single form their describers to the genus Testudo. In known to have a Gopherus palate finds its reality only three forms can be assigned with natural place. Table 3 displays the differences immediate certainty to the genus Testudo, between the two series. while the one other form in which the skull A single form of the sufficiently known tor- is known, fTestudo brevisterna Loomis, 1909, toises of the New World Miocene (tTestudo is clearly better assigned on the basis of char- I The skulls of Gopkerus polyphemus examined by the acteristic palatal ridging to the genus Go- author show a further specialization: the lateral maxil. pherus. lary ridges do not stop short at the premaxillary The character on which this generic de- boundary but continue on that bone to join the median termination is made is worth premaxillary ridge. This character of G. polyphemus was examining. long ago observed by George Baur (MS notes made Loomis (1909, pp. 21-22) describes the con- available by Dr. Doris Cochran). This condition has dition in tGopherus brevisterna thus: "The not been seen in Gopherus agassizii or G. berkandieri, the vault of the palate is very low and has a other living species of the genus, by either Baur or the median ridge running from the basisphenoid author; according to Baur it may sometimes be absent in G. polyphemus. It should be emphasized that the onto the premaxillae.... The median ridge contact of maxillary and premaxillary ridges is never mentioned above as continuing onto the evident in unprepared skulls, in which the maxillae and premaxillae separates two deep pits, one on premaxillae are still covered by a horny sheath. is BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 95 TABLE 3 t Testudo osborniana tGopherus brevisterna tTestudo gringorum Series Series PalatePaat OnOne orrtoltrlmxltwo- lateral maxil- One lateral maxillary Not known lary ridges and a pre- ridge and a median maxillary pit premaxillary ridge External surface of den- With fine vertical ribbing Without vertical ribbing Not known tary Nuchal scute Much longer than wide Nearly as wide as long Absent Shell shape Tending to be high and Tending to be flattened Not high globular

Anterior shell margin With flared anterolateral Anterolateral corners Anterolateral corners corners rounded, not flared slightly flared Nuchal identation Marked Usually absent Slight Fourth vertebral scute As long as wide Wider than long IJncertain Entoplastron Wider than long Usually longer than wide Not wide Pectoral scutes In contact with entoplas- Usually well separated Separated from the tron or very close, me- from entoplastron, entoplastron, me- dian sulcus 1/15 to 1/30 median sulcus usually dian sulcus 1/5 of of abdominal median about 1/5, never less abdominal median sulcus than 1/10, of abdomi- sulcus nal median sulcus Neural plates Fully differentiated, sec- Usually less completely, Partly differentiated ond and fourth octag- or not at all, differen- but first and second onal, third and fifth tiated, often the fifth hexagonal quadrilateral hexagonal or all hexag- onal Dermal ossicles on fore Present and extremely de- Absent or not well de- Not known and hind limbs and veloped, of-ten com- veloped tail pacted and suturally united gringorum Simpson, 1942) cannot be included Patagonian and probably the ancestor of the in these two series. The characters of tT. modern South American species, all of gringorum are compared with those of the which have retained the relatively primitive two series in table 3. The absence of the nu- Testudo palate seen in fT. osborniana. chal scute readily separates this form from The Miocene representatives of the tTes- both the tT. osbornmana and fG. brevisterna tudo osborniana series are tT. osborniana, series. Relationship, however, is probably IT. impensa, tT. angusticeps, and fT. closer to the tT. osborniana group. The nar- arenivaga. All these show the listed charac- rowed nuchal of the tT. osborniana group, ters of the series in full development. for example, could more easily lead into a tGopherus brevisterna, described from a condition with the nuchal absent and the complete skeleton, is the most adequately first marginals in broad contact (as in tT. known representative of its group. tG. hol- gringorum) than with the broad nuchal of landi, tG. edae, tG. pansa, and tG. mohavense the tG. brevisterna group. tT. gringorum also are known from good shells. tG. vaga is has traces of some of the shell specializations represented by a good plastron, limb bones of the tT. osborniana group, and, while its and girdles and vertebrae, and fragments skull is regrettably unknown, the species is of the carapace. All of these are much alike BULLETIN AMER. Mus. NAT. HIST. VOL. 95, PLATE I

Testudo cubensis, type pleural plate, Academy of Natural Sciences of Philadelphia No. 8923. 1. External view. 2. Internal view. X 1 BULLETIN AMER. Mus. NAT. HIST. VOL. 959 PLATE. 2

I

Testudo cubensis, anterior lobe of plastron, A.M.N.H. No. 6202. 1. External view. 2. In- ternal view. X 1/2 BULLETIN AMER. Mus. NAT. HIST. VOL. 95, PLATE 3

Testudo cubensis, left posterior quadrant of plastron, A.M.N.H. No. 6201. 1. External view. 2. Internal view. X 1/2 BULLETIN AMER. Mus. NAT. HIST. VOL. 95, PLATE 4

I

Testudo cubensis. 1. Epiplastra, Museum of Comparative Zo6logy No. 1901. 2. Pleural, A.M.N.H. No. 6212. 3. Nuchal, A.M.N.H. No. 6204. 4. Peripheral, A.M.N.H. No. 6215. 5. Pygal, A.M.N.H. No. 6227. All X 1/2 BULLETIN AMER. Mus. NAT. HIST. VOL. 95, PLATE 5

2

A4'4 i,a

3

Testudo cubensis. 1. Caudal vertebrae (caudal column of T. denticulata at right for comparison). X 1/2. 2. Femur, A.M.N.H. No. 6242. X 1. 3. Humeri. X 1/2 BULLETIN AMER. Mus. NAT. HIST. VOL. 95, PLATE 6

7. .,. S

I;Sr

i. k C..

/ .1

'/'- -~~~~~~~~~~~~~~~~ 4 .K. V-1,

Z >0> 4_- 1 q 4-~~~~* i,_ s ,. i' gI , ..r .j1

I li .,

2

Shell and skull of Testudo (Chelonoidis) denticulata. A.M.N.H. No. 46864. 1. Dorsal view. 2. Ventral view. Not to scale BULLETIN AMER. Mus. NAT. HIST. VoL. 95, PLATE 7

2 Shell and skull of Gopherus polyphemus. Cornell University No. 108. 1. Dorsal view. 2. Ventral view. Not to scale BULLETIN AMER. Mus. NAT. HIST. VOL. 95, PLATE 8

3 4 .; 1. Shell and skull of Testudo (Hesperotestudo) angusticeps (skull of type, A.M.N.H. No. 6295; shell of young specimen, A.M.N.H. No. 6799), dorsal view. 2. The same, ventral view. Both X 1/2. 3. Skull of Testudo (Chelonoidis) denticulata, palatal view. A.M.N.H. No. 46864. X 1. 4. Skull of Gopherus polyphemus, palatal view. Cornell University No. 108. X 1 1950 WILLIAMS: TESTUDO CUBENSIS 19 in essentials and in particular agree in hav- be long with parallel sides, and the skull in ing one or two neural plates octagonal, the one specimen of tG. praeextans in which contrasting in this respect with tG. undabuna, it is known, although typical Copherus in tG. inusitata, tG. emilite, and tG. copei (all its palate, is as a whole more dolichocephalic known only from shells) in which the neurals than tG. brevisterna or any modern Gopherus. are all hexagonal. All the latter species are In both these respects the Oligocene forms closely similar, much more like one another can be regarded as more primitive. than like the subseries typified by tG. Like tTestudo brontops these forms are brevisterna, but they fall within the broad large (420-540 mm. in carapace length), limits of the tG. brevisterna group as defined larger certainly than the average of modern in table 3, and they are very different from Gopherus, but otherwise they are very similar the tT. osborniana series. not merely to the Miocene forms but to the tTestudo gringorum stands alone. It is the living members of the genus. As the illustra- only known Miocene testudine of South tions of the shell of a living form (pl. 7) dem- America and shows no close approach to the onstrate, the characters pictured by Hay characters of any other tortoise of that period. (1908, figs. 509, 510, pl. 67) and Gilmore (1946, pls. 38-43) in forms from the Chad- OLIGOCENE ANCESTORS ronian are reproduced with only minor change The two North American testudine lineages in at least some specimens of living species. can be traced into the Oligocene. The repre- The differences between fG. laticunea and G. sentative in this period of the Miocene polyphemus appear to be less than those be- tTestudo osborniana series is tTestudo bron- tween G. polyphzemus and G. berlandieri, al- tops Marsh, 1890. In most respects, in its though the last two are contemporaneous and strongly flared anterolateral corners to the separated from fG. katicunea by half the span carapace, strong nuchal indentation, narrow of the Cenozoic.' The resemblance is the more nuchal scute, humeropectoral sulcus en- astonishing since the specific status of the croaching slightly on the entoplastron, and living forms (G. polyphemus, G. berlandieri, G. fourth vertebral scute long with parallel agassizii) has sometimes been considered sides, it resembles the Miocene members of doubtful. In such circumstances doubt would its group. In one respect, however, it is more appear to be cast on the validity of the nu- primitive: the pectoral scute is not narrowed, merous currently recognized fossil forms. It is its median sulcus being about one-fourth the feeling of the author, however, that at that of the abdominal scute. Two octagonal present any extensive synonymizing of cur- neurals are present in the one described rently accepted names is unwarranted. specimen of this form. Neither material nor analysis is yet adequate Conspicuously different from tTestudo to determine which are the valid species, brontops are the two known Oligocene repre- nor have the osteological characters of the sentatives of the tGopherus brevisterna group. modern species thus far been worked out These two forms, tGopherus praeextans with sufficient care to afford any yardstick by Lambe, 1913, and tG. laticunea Cope, 1873, which to judge the distinctness of the fossil so similar that Gilmore (1946, pp. 306-307) species or vertical subspecies or individual was for a time inclined to regard them as the variants, whichever they may be.2 two sexes of one species, have round or trun- I Hay (1908, p. 403) expressly commented on the re- cate carapace borders, neither flared nor in- semblance of tG. laticunea to G. polyphem-us. He drew dented; the nuchal scute is broad; the humer- from that resemblance no phyletic or nomenclatorial opectoral sulcus is widely removed from the conclusions. This action is characteristic. Admirable in a descriptive sense as Hay's work was, it must be entoplastron; the pectoral scute is very broad, stated that, while he displayed extraordinary temerity its median sulcus about one-half that of the in the erection of specific names, he displayed ex- abdominal, all characters quite congruent traordinary timidity in inferring relationships. with those of their Miocene relatives. In 2 Lack of adequate osteological series and the fact two that this is a subsidiary and not a primary problem for respects, however, these forms are some- this paper have prevented the author from going into what different from the comparable Miocene this matter more completely. Internal shell and skull forms: the fourth vertebral scute tends to characters can, it would seem from a preliminary survey, 20 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 95 Five shells of fGopherus praeextans and and until the feet are found, we cannot be five of tG. laticunea have been described, sure that they did not, like tStylemys, so long enough to establish that the shapes of the considered a testudine, have the emydine neural plates are variable, but more often phalangeal formula (Case, 1936). As in hexagonal, in both forms. In one specimen of emydines and some living testudines, the tG. praeextans two octagonal neurals are supracaudal scute is divided. present, in another specimen, one octagonal tTestudo uintensis Gilmore, 1915, has more neural is present. In tG. laticunea a single testudine characters, but its relationship to neural is octagonal in two of the five shells. the three testudine lineages of the later tTestudo gringorum has no known ancestors Cenozoic is as problematical as that of tH. in the earlier Tertiary of South America, and corsoni. Neither the nuchal scute nor the it is suspected (Simpson, 1942, 1943) that anterior carapace margin is known. The the genus Testudo was a late invader of the fourth vertebral scute is longer than wide. southern continent, rafting there over a sea The pectoral scute is very broad, its median gap after the breaking of the Panama con- sulcus one-half that of the abdominal. The nection in the early Tertiary. humeropectoral sulcus touches the entoplas- tron. The supracaudal scute is divided. There EOCENE PROTOTESTUDINES are two octagonal neurals, but these are the The three testudine lineages we have been second and sixth rather than the second and describing cannot be followed as such into the fourth as in later North American forms. Eocene. The known Eocene forms are essen- There are only seven neurals. The latter tially prototestudine in stage of differentia- character in this form, known from a single tion, perhaps some of them even pretestudine. shell, quite probably is a matter of individual tHadrianus corsoni (Leidy), 1871, of the variation. Bridger Eocene is primitive enough in its A tenuous argument can be made for closer known characters to have given rise to any relationship to the tT. brontops-tT. osborni- later Western Hemisphere testudine. Its ana line. A resemblance can be found in the neurals are all hexagonal; its pleurals do not contact of the humeropectoral sulcus with show alternating broad and narrow distal the entoplastron; this, however, may be a ends; the nuchal scute is present and neither primitive feature. Further, if it is permissible very broad nor very narrow; the contact of to reason from the single shells of tT. the humeropectoral sulcus with the entoplas- uintensis and tT. brontops that have been so tron and the rather narrow, parallel-sided far described to populations of these species, vertebrals may imply a special relation to the early occurrence of octagonal neurals may the tT. brontops-tT. osborniana series, but possibly be considered a diagnostic feature of even if this is true, it is not a specializa- a tT. uintensis-tT. brontops-tT. osborniana tion so radical that it bars relationship to the line. On the basis of our present knowledge of other lineages as well. Whether this species Oligocene and Miocene forms, the Gopherus is itself the ancestor of the three later testu- phyletic group does seem to have a lower fre- dine lineages or not, it would seem to exem- quency of neural patterns with alternating plify a grade of organization through which octagonal and quadrilateral neurals, but ad- they all have passed. It is still not completely mittedly our present knowledge of both the known; the skull has never been discovered, Gopherus and tT. osborniana sequences is seriously defective. be used to distinguish these species, in addition to the If early differentiation of a complex neural external features listed and evaluated by Bogert and Oliver (1942). Baur (MS) cites a number of osteological pattern is characteristic of a tT. uintensi,s- characters. L. T. Murray (MS) finds also a character tT. brontops-tT. osborniana line, neither the of the soft anatomy to distinguish G. berlandieri from South American forms nor Gopherus is likely the other living species. According to Murray, in G. to have been derived from tT. uintensis. agassizii and G. polypkemus the trachea extends into From minor the body cavity before bifurcating; in G. berlandieri, on morphological approaches to the the other hand, the trachea is very short, and the tT. osborniana series in tT. gringorum it bronchi extend down the side of the neck for a consider- appears possible that the South American able distance before reaching the body cavity. tortoises have some not too remote relation 1950 WILLIAMS: TEST UDO CUBENSIS 21 to that series. Such a relationship would be dentation and gular process parallel sided either at the level of tT. uintensis or at that and deeply bifurcate so typical of the living of tH. corsoni. Since neural plate differenti- form. It is distinguished from G. berlandieri, ation is evident in Eocene tT. uintensis and as Hibbard noticed, by the greater thickness not in Miocene tT. gringorum (again reason- of shell and by the lack of differentiation of ing from single shells to populations) it the pleurals. would seem more probable that the relation- tGopherus riggsi may, however, be a syno- ship is at the earlier (11. corsoni) level. nym or race of tG. turgida (Cope), 1892, Gopherus, with fewer similarities and well- from the Blanco beds of Texas. The type of marked specializations of its own, would seem tG. turgida in the Academy of Natural Sci- to have originated at some even earlier stage. ences of Philadelphia resembles to a remark- PLIOCENE AND PLEISTOCENE able degree certain Meade County, Kansas, DESCENDANTS specimens at the Kansas Natural History Museum referred to tG. riggsi. tG. turgida If the three Miocene testudine series can- and the Meade County specimens differ from not be followed with confidence backward the type and paratype of tG. riggsi from into the Eocene, they can be traced forward Seward County, Kansas, primarily in greater into the Pliocene and Pleistocene. But here thickness of shell and an extraordinary tend- also there is a difficulty, not this time arising ency to gibbosity, especially in certain scutal from the primitiveness of the material but areas such as the areas under the gulars and solely from the fragmentary nature of much anals. In these respects and indeed in all of it. Especially in the Pleistocene there is a points tG. turgida and the Meade County multiplicity of named forms too poorly rep- d"riggsI" strikingly resemble tTestudo kal- resented for even generic allocation. Those ganensis Gilmore, 1931, from the Cenozoic of forms that are known adequately cause no Mongolia (horizon undetermined). This set of such difficulty and are each quite typical of described forms merits renewed investiga- their phyletic group. tion. Pliocene tTestudo orthopygia Cope and tT. tBysmachelys canyonensis Johnston, 1937, gilberti Hay, the first very completely known, known from very complete material but im- the second known from the skull only, con- perfectly described by its author, is undoubt- tinue the tT. osborniana line, manifesting the edly a Gopherus. It has many resemblances characters of typical Miocene forms. on the one hand to tG. vaga and tG. brevisterna tTestudo sellardsi Hay of Florida, as de- and on the other to G. potyphemus. The ex- scribed by Loomis (1927), appears to carry at treme brachycephaly (width across quadrates least one of the tT. osborniana specializations 103 per cent of basal length from snout to one point further. The nuchal scute in this occipital condyle), considered part of the form has become still narrower and smaller generic definition by Johnston, is paralleled and is in fact retracted from the anterior in tG. brevisterna (same width 100 per cent of margin of the shell so that the first marginal basal length, Loomis, 1909) and in G. poly- scutes meet in front of it. 6 phemus. The size is large (805 mm. plastral tTestudo riggsi Hibbard, 1940, of the length), larger than any other form definitely Kansas Pliocene, on the other hand, is not known to be Gopherus, but this is not of merely a Gopherus but, as the author can generic value. The very short cervical verte- testify from personal examination of the type brae and especially the exaggeratedly wid- and additional and comparativle material, ened and lengthened postzygapophyses of the one whose affinities to a living species seem eighth cervical resemble the same elements in determinable. In several characters rather tG. vaga and G. polyphemus. The ridge on the peculiar within the genus it is similar to distal end of the femur separating the articu- Gopherus berlandieri.1 It has the nearly lar surfaces for the tibia and fibula, regarded hemispherical shape, moderate nuchal in- as important by Johnston, has been described ' This resemblance has already been noticed by by Hay (1908, p. 417) in tG. vaga. Malcolm Smith in the Zoological Record for 1944 (sec- Pliocene tTestudo hayi Sellards, in which tion on Amphibia and Reptiles, p. 26). all the neurals are still hexagonal, may be a 22 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 95

Nearctic Testudo Neotropical Testudo GEOLOGIC PERIOD subgenus subgenus Gophierus Hesperotestudo Chelonoidis

RECENT denticulata 'IXpbhOntopus berlIndieri qgassizi RQIyphemus chilensis l \ X~~~~~~~o;sziiolmu~ PLEISTOCENE sellardsi cubensis

PLIOCENEPLIOCENE ortho ygia Qi ns£gL \ /ynss gilbeEti hav t d \ I /

FovLu_MRen_s I I I I inusitotocal, \ m~~~~~~~~~~~~~ohaes MIOCENE I OgUIttcORA~ ~ ~ ~ ~ ~~Ihollandi .1 emilioe edae areniva a undobuna brevisterno

oos OLIGOCENE bron lc

umRintensi,/soiunopr xt

EOCENE

Hadrionus corsoni

FIG. 2. Phyletic diagram of Western Hemisphere tortoises, accurate not to species but only as to group, although special relationship is indicated where there is some evidence for it. Forms of doubtful provenance are not included. The separation of Gopherus into two phyletic groups may or may not be real. continuation of the Miocene tundabuna- vereto is the Pliocene representative of tT. tcopei subseries of Gopherus. This is a very gringorum. The two forms are in fact not large and thick-shelled form; its relationto certainly distinct. By an unfortunate error the thick-shelled giant forms of the Pleisto- Rovereto in his original figure mistook the cene might be suggested. femoral-anal sulcus for the hypoxiphiplas- Pleistocene forms definitely referable to tral suture, and the shortness of the xiphi- Gopherus are known but are either stated to plastral region which was the most important be indistinguishable from a living species feature relied upon by Simpson (1942) in (G. polyphemus Hay, 1917b; G. agasslisz distinguishing this species from tT. gringorum L. Miller, 1942) or are doubtfully distinct is therefore illusory. Because tT. praestans (tG. hexagonata Cope, various fragmentary Rovereto, 1914, considerably antedates tT. species). tGopherus praecedens, if a rather com- gringorum Simpson, 1942, the applicability of plete shell in the United States National Mu- the latter name may be called in question. seum identified by Gilmore be accepted as Provisionally, however, fT. gringorum, by such, appears to differ little from Gopherus far the better known form, may be retained on polyphemus except in larger size. the hypothesis that evolution in neural pat- In South America tTestudo praestans Ro- tern (the neurals are hexagonal in tT. 1950 WILLIAMS: TESTUDO CUBENSIS 23 gringorum, unknown in tT. praestans) oc- Still avoiding all discussion of possible curred in the Miocene-Pliocene interval. European and Asiatic affinities,2 I suggest tT. praestans as the later form may reason- the following picture of the origin and evolu- ably be expected to be closer in its neural tion of tT. cubensis: pattern to the alternating octagonal-quadri- The primitive testudines of the New lateral pattern of its living descendants (T. World were early separated into three series: denticulata, T. chilensis, T. elephantopus). (1) Gopherus; (2) Nearctic; and (3) Neotropi- tT. gringorum, as represented by the type, is cal members of the genus Testudo. The also much smaller than the type of tT. Neotropical tortoises probably differentiated praestans. in Central America at a time when this area Pleistocene South American tortoises are tended to be separate from both the northern undescribed or too imperfectly preserved and southern continents (Schuchert, 1935; to merit discussion. Gervais (1877) based Weeks, 1948). During this period of differen- tT. elata on bits of a giant tortoise from the tiation the nuchal scute was lost. Then, Pleistocene of Brazil. Swinton (1928) has probably over a rather narrow sea gap, this cited fragments of a thick-shelled form from progenitor of the Neotropical tortoises sent a Venezuela. Ameghino (1921, p. 708) men- representative eastward into the West Indies, tions the existence of Pampean Argentine the ancestor of tT. cubensis. Southward via forms. Panama another representative went into THE PHYLETIC POSITION OF South America, and southwestward over a tTestudo cubensis much larger sea gap still another to the The analysis has now proceeded to a point Galapagos Islands. Isolated in Cuba, perhaps where it is possible to make some inferences beginning with the Miocene, tT. cubensis had concerning the position of tT. cubensis in the ample time to develop its special characters. phylogeny of Western Hemisphere testu- 2 It is a source of regret to the author to have neg- dines. lected the relation of the North American testudine All the special resemblances of tT. cubensis fauna to that of Eurasia. To do so, however, would have are to the South American tortoises. Like been a crushing task, requiring a complete review and them it is southern in distribution, lacks the revision of the Eurasian (and African) testudines, fossil and recent. Especially discouraging in such a task is the nuchal scute,l and has the first marginals in absence, for the fossil forms, of any general review broad contact. It also resembles, as Matthew nearly so competent as that of Hay. The work most noticed, the individual South American forms nearly comparable in bulk (Bergounioux, 1935) must be in special ways, T. denticulata in the elonga- received with extreme caution. The identification, for example, in that work of Kinixys in the Miocene of tion of the shell, T. chilensis in the dentate France is unquestionably an error, deriving in part from margins, T. elephantopus in the thinning of a radical misconception of the definition of Kinixys, the shell. Resemblances to North American ascribing to it a hinge in the plastron rather than in the forms are less frequent and apparently con- carapace. The works of Peyer (1942) on European vergent. tTestudo vitodurana and its relatives (affinities to African T. pardalis are apparently very justly sug- 1 The nuchal scute is occasionally absent as a con- gested) and of Glaessner (1933) on tT. antiqu?a are, on genital variation in modern Gopher=s ber&andieri, but the other hand, examples of illuminating studies of small when this occurs, the first marginals are not in broad segments of the problem. contact; instead the first vertebral scute attains the Examples may easily be cited to show that the lack, anterior margin of the carapace. The nuchal scute, unavoidable as it is, may not be trivial. De Stefano when present in G. berkandieri, is much broader than (1902) described a possible tStylemys (tS. botti) from long, so much so that its disappearance in the fashion the Miocene of France, and Riabinin (1927) has just mentioned is only an extreme of this normal charac- founded another species, tS. karakolensis, on a shell ter. The reduction of the nuchal scute in Pleistocene tT. from the late Oligocene or early Miocene of Siberia. sellardsi as described by Loomis (1927) is probably a The case of tTestudo kalganensis Gilmore (Cenozoic of better model of the mode of suppression of the nuchal Mongolia), which remarkably resembles tG. turgida scute in tT. cubensis and the other Neotropical tor- (Cope) and the Meade County specimens referred to toises, but in fT. selkardsi the reduction of the nuchal fG. riggsi (Hibbard), has already been mentioned. scute is combined with highly specialized characters Such instances emphasize the impossibility of artifi- (e.g., the extremely narrowed pectoral scute) not found cially limiting discussion to the forms of any one con- in the Neotropical tortoises. tinent. 24 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 95- THE NOMENCLATURE OF WESTERN HEMISPHERE TESTUDINES THE NEOTROPICAL TORTOISES Lindholm (1929) has already argued for Lindholm (1929) placed the three living subdivision of the genus Testudo. The latter Neotropical species of tortoise in separate is, of course, a relict of the Linnean genus and monotypic subgenera, placing T. denticulata as such is much more heterogeneous than, for in Chelonoidis Fitzinger, 1835, T. chilensis example, are emydine genera. The size range in Pampatestudo Lindholm, 1929, and T. within the genus has been commented upon elephantopus in Elephantopus Gray, 1873. He by Lindholm (1929, p. 284) as unique in the was to a certain extent justified in this proce- animal kindgom. Primarily, however, the dure. These three species are very distinct argument for subgeneric separation must forms. The skull of T'. denticulata, for exam- rest upon convenience and utility. As pro- ple, has certain specializations that permit its posed here, the subgenus is intended to in- immediate recognition. The shell peculiarities clude a phyletic line recognizable as such over of the same species are almost equally well several geologic periods, perhaps restricted to marked. Nevertheless, adoption of Lindholm's a geographic area or major region, and not subgeneric criteria would compel recognition morphologically sufficiently distinct to be of tT. cubensis as still another subgenus, and called a genus. It is believed that such phyletic at least one more subgenus would be needed lines can be distinguished in the tortoises, for the reception of the fossil South American and that their formal recognition will clarify species. Such a multiplication of names, the generic picture. probably all of which on these principles Chelonoidis, as here recognized, includes, would be monotypic, would reduce the sub- in addition to the type species T. denticulata generic category to an absurdity. It is there- (Recent of tropical South America, east of fore suggested that the living Neotropical the Andes') the following: tTestudo gringorum tortoises and their fossil ancestors and tT. Simpson (Patagonian = Miocene of Argen- cubensis be united under the oldest available tina); tT. praestans Rovereto (Monte Her- name, which will then be redefined as follows: moso =Pliocene of Argentina); tT. cubensis Leidy (Ciego Montero = Pleistocene of GENUS TESTUDO LINNt Cuba); T. chilensis Gray (Recent of Ar- SUBGENUS CHELONOIDIS FITZINGER, 1835 gentina and Uruguay); and T. elephantopus TYPE SPECIES: Testudo denticulata Linne. Harlan (Recent of the Galipagos Islands). NEW DEFINITION: A Neotropical division Additional South American forms found of the genus Testudo with the nuchal scute fossil have been named and, if valid, probably lacking, the entoplastron large, and with belong in this subgenus, but their status is little or no trace of radiating pattern on the very doubtful. tT. gallardoi Rovereto, 1914 horny shields. Miocene to Recent, South (Araucanian = Pliocene of Argentina), is America; Pleistocene, Cuba; Recent, Gala- poorly known; its only well-defined charac- pagos Islands; probably existing in Central ter as given by its describer is the gibbosity America in the early Cenozoic. of some of the neurals and pleurals. In view This definition will not seem sufficiently of the marked gibbosity of these elements in restrictive unless it is remembered that in a specimen of T. denticulata examined in the broadening the subgeneric concept to include American Museum of Natural History, it is a varied assemblage of forms, both fossil and not felt that this character by itself estab- recent, the number of characters that are lishes the reality of tT. gallardoi. tT. elata possessed by all members of the subgenus and Gervais, 1877 (Pampean=Pleistocene of not also found, or not found in this combina- Brazil), is a giant form known only from a tion, in other members of the genus is re- fragment of the dentary, ends of two arm duced. The group as here defined is certainly a natural group-both a geographic and a I Also extending a short distance into Central Amer- ica, in Panama, and present, perhaps feral, on several phyletic unit. It is possible, however, to raise West Indian islands. A specimen caught by Hassler in a question of the advisability of formal the interior of Haiti (Grant and De Sola, 1934) was. nomenclatorial recognition. identified as this species. 1950 WILLIAMS: TESTUDO CUBENSIS 25 bones, and a very small piece of the plastron; orthopygia Cope (Lower Pliocene of Kansas); it is not at present determinable. t T. australis tT. gilberti Hay (Lower Pliocene of Kansas); Moreno (Monte Hermoso = Pliocene of Ar- and tT. sellardsi Hay (Pleistocene of Florida) gentina), tT. formosa Moreno (Monte Her- appear to be members of this subgenus. moso = Pliocene of Argentina), and tT. The place of the other group of Nearctic paranensis Scalabrini (Entre Rios= Pliocene tortoises, Gopherus Rafinesque, 1832, whether of Argentina) are cited by Ameghino (1921), as genus or subgenus, is problematical. The but unfortunately the original descriptions palatal character that distinguishes this have not been seen. These may be nomina natural taxonomic unit is a minor one but one nuda.1 nowhere else developed within the order. Vertebral and pelvic characters also seem to THE NEARCTIC TORTOISES distinguish this series of forms. Divergence The North American phyletic group of from Testudo, if in fact any close relationship which tTestudo osborniana is an example is, exists, must, it seems clear, have been very as are the Neotropical tortoises, a geographic early, at a stage that would have been classed and phyletic unit. It deserves recognition as by Hay as "Hadrianus." It was at an evolu- a new subgenus, which may be called: tionary level, therefore, which included some emydine characters, e.g., a divided supra- HESPEROTESTUDO, NEW SUBGENUS caudal scute and perhaps even the emydine TYPE SPECIES: tTestudo osborniana Hay. phalangeal formula. (Compare the discovery DEFINITION: A Nearctic subdivision of the by Case, 1936, of an emydine type of foot in genus Testudo with the nuchal scute narrow, fStylemys.) At present it appears wise to re- the entoplastron wider than long, the pec- tain Gopherus as a genus. toral scutes extremely reduced from the axilla Until further investigation establishes ad- to the midline, and the limbs and tail heavily ditional characters (other than merely primi- armored with dermal ossicles. Eocene to tive ones) the genus may be defined as fol- Pleistocene, North America. lows: This is a highly specialized line, early dif- GENUS GOPHERUS RAFINESQUE, 1832 ferentiated, exclusively North American, and now extinct. The number of valid in- TYPE SPECIES: Testudo polyphemus Bar- cluded species is not certain, but the named tram. forms: tT. uintensis Gilmore (Uinta Eocene DEFINITION: A Nearctic genus of Testu- of Utah); tT. brontops Marsh (Chadronian dininae distinguished by the continuation of Oliogocene of South Dakota); tT. amphi- a palatal vomerine ridge upon the premaxil- thorax Cope (Chadronian Oligocene of Colo- laries at their symphysis; cervical vertebrae rado); tT. ligonia Cope (Chadronian Oligo- short. Oligocene to Recent, North America. cene of Colorado); fT. arenivaga Hay (Arik- Formerly widespread, now much restricted areean Miocene of Nebraska); tT. angusti- in range. ceps Matthew2 (Hemingfordian Miocene of The difficulty of establishing the number Nebraska); fT. osborniana Hay (Barstovian of valid species in this genus has been com- Miocene of Colorado); tT. impensa Hay mented upon above. The following named (Barstov'ian Miocene of Montana); forms are, however, undoubted Gopherus: fT. fG. laticunea Cope (Chadronian Oligocene I The species named by Moreno are cited also in the of Colorado); tG. praeextans Lambe (Chad- Zoological Record for 1889 and in Nicholson and ronian Oligocene of Wyoming); tG. brevi- Lydekker (1889, p. 1111). Originally described by Matthew (1924) as tT. sterna Loomis (Arikareean Miocene of Wyo- orthopygia angusticeps "new mutant." For the sake of ming); fG. undabuna Loomis (Arikareean consistency the name is treated as a full species here. Miocene of Wyoming); fG. emiliae Hay The form is, however, certainly close to tT. orthopygia (Arikareean Miocene of South Dakota); tG. and may on further analysis prove to be best regarded edae Miocene of as a synonym. Hay (Arikareean Nebraska); The shells contemporaneous and sympatric with tT. fG. hollandi Hay (Arikareean Miocene of angusticeps referred by Matthew to torthopygia are here Nebraska); tG. vaga Hay (Barstovian Mio- provisionally regarded as tangusticeps. cene of Montana); tG. inusitata Hay (Bar- 26 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 95 stovian Miocene of Montana); tG. copei skulls of tS. nebrascensis (one definitely Koerner (Barstovian Miocene of Montana); associated with "enough of the shell to estab- tG. pansa Hay (Barstovian Miocene of lish clearly the species") that showed the Colorado); tG. mohavense Merriam (Bar- Gopherus ridge at the premaxillary sym- stovian Miocene of California); tG. turgida physis. Hay's comparison of tStylemys with Cope (Pliocene of Texas); tG. canyonensis Gopkerus was contested by Case (1925) in the Johnston (Pliocene of Texas); tG. campester course of a detailed study of an undoubted Hay (Pliocene of Texas); tG. riggsi Hibbard tStylemys skull. Case found differences in the (Pliocene of Kansas); tG. hayi Sellards otic region and in the extension of the (Pliocene of Florida); tG. hexagonata Cope pterygoids lateral to the palatines, but be- (Pleistocene of Texas); and the living forms cause the lower jaw of his specimen was G. polyphemus (Pleistocene of Florida, Re- tightly set in position he was unable to deter- cent of Florida and adjoiniing states), G. mine the presence or absence of the diag- agassizi (Pleistocene of California, Recent of nostic premaxillary ridge. southern California, portions of Arizona, Hay's comparison of tStylemys and Utah, Nevada, and Sonora), and G. ber- Gopherus was not based on skulls alone but landieri (Recent of Texas and adjoining also on limbs (1908, p. 390) and on shells. In Mexico). a discussion of the two genera in 1905 he was quite explicit: "There are also various re- SPECIES AND GENERA INCERTAE SEDIS semblances between the shells of the living A number of named Western Hemisphere lGopherus] and the extinct forms [tStylemys]; testudines or supposed testudines cannot, or so that a real relationship is suggested" (Hay, can only provisionally, be placed in the three 1905, p. 342). phyletic groups that have been described. It is certain that tS. nebrascensis is not These names will now be considered. directly ancestral to Gopherus, since two species of Gopherus have now been shown to GENERA coexist in the Oligocene with that species, tStylemys Leidy (several named species in but it is possible that tStylemys and Gopherus the North American Oligocene and the John are both members of a distinct group in- Day portion of the Miocene): This well- dependently derived from the Emydinae, and known genus has been considered emydine that they are more closely related to each since Case's (1936) discovery of its emydine other than either is to Testudo (or tHad- phalangeal formula. It shows, however, so rianus). There is, however, no uncontested many parallels to testudine structure that it evidence on this point at present. probably must be conceded some fairly close This interesting point will not be settled relation to the common stem of all testudines. until an associated shell and a skull, both Further, when it is considered that the living completely prepared, of an undoubted genera considered testudine may have arisen tStylemys are available. A number of Oligo- independently from a tHadrianus stage (a cene skulls with the Gopherus premaxillary grade of organization with several emydine ridge are now known (specimens seen in characters), it is necessary to inquire whether Princeton Unieversity, the Kansas Natural the Testudininae or even the genus Testudo History Museum, and the United States as currently defined is not polyphyletic, that National Museum). Some of these may be is, has arisen several times independently tStylemys, but since only the skull of Gil- from forms on the basis of present definition more's tG. praeextans has unquestioned as- emydine. If this possibility is substantiated, sociation with an identified and identifiable the definition of the subfamily must be put shell, the critical point cannot now be demon- on a new basis or, alternatively, the Testu- strated. dininae must be admitted to be impossible tCymatholcus Clark (Eocene of Utah) : This of formal separation from the Emydinae. genus, erected by Clark in 1931, will prob- Attention should be called to Hay's de- ably stand in spite of the apparent vagueness scription of the skull of tStylemys (Hay, of its known generic characters. It is rather 1908, pp. 389-390). Hay believed that he had specialized (and specialized in a quite in- 1950 WILLIAMS: TESTUDO CUBENSIS 27 dividual way) for an Eocene form. If the and tT. snovitana (Cope), described from a presence of the testudine phalangeal formula phalanx, may never be determinable. tT. is essential for allocation to subfamily, this undata Cope, tT. klettiana Cope, tT. annae form is not certainly a testudine, since the Hay, tT. distans Hay, tT. equicomes Hay, limbs are quite unknown. The skull also is tT. luciae, Hay, tT. francisi Hay, and tT. unknown. tHadrianus ? schucherti Hay, 1903 munda Hay, described from almost equally (Eocene of Alabama), would seem from de- uncharacteristic fragments, may possibly be tailed similarities in form and in the sulcal validated or at least placed as to genus by and sutural relationships of the plastral further discoveries of material at the type elements, as these are figured by Hay (1908), locality or horizon. The author has seen the to be a member of the same genus and per- types of all these forms except tT. exornata haps even conspecific with tC. longus Clark, and tT. snoviana, and he cannot deny their 1931. different appearance, but he considers that tBystra Hay (Pliocene ? of Florida): This the evaluation of this subjective impression genus is founded upon a single very small of a difference is impossible with material so (105 mm. in carapace length) shell manifest- incomplete as to be nearly beyond the hope ing a single aberrant character. Despite the of proper comparison. temptation to regard this specimen as an in- tTestudo crassiscutata Leidy, tT. nio- vidual variant, the name must be provi- brarensis Leidy, tT. laticaudata Cope, and sionally retained. The aberrant character, tT. louisekressmani Wark may be determi- the transverse gular-humeral sulcus, is in the nable when more complete material is col- form exhibited here rare in tortoises. (See, lected. tGopherus atacosae, tTestudo ocalana, however, T. schweiggeri in Boulenger, 1889, and tT. incisa, all of Hay, though known pl. 3.) When so rare a character is coupled from fragments probably specifically indeter- with extremely small size in an animal ap- minable, are doubtless justifiably referred to parently adult (Hay, 1916), it becomes the genus Gopherus on the basis of close necessary to withhold judgment until addi- resemblance to modern species. In the case of tional material is found. That this genus is tT. thomsoni Cope and tT. peragrans Hay,' testudine is impossible to establish, since the although the skulls of the type specimens are shell alone is known. known, diagnostic features are missing or Hay's name tBystra, 1916, is preoccupied unexposed. tT. cultratra is known only from by Bystra Cameron, 1902, for a genus of Cope's description, in which unfortunately Hymenoptera. It must therefore be replaced, the characters here believed to be significant and tFloridemys, new name, suggested to the are not mentioned. tT. pertenuis Cope is author by Mr. C. M. Bogert, is therefore pro- poorly described and badly figured, and the posed. greater part of Cope's material seems to have tAchilemys Hay (Bridger Eocene of Wyo- been lost (Hay, 1908). tT. ducatelli Collins ming): This genus is reported on extremely 1 The form described and figured as this species in fragmentary material. Its testudine affinities Schlaikjer (1935) is not properly so assigned. It is are more than doubtful. closer to tGopherus praeextans (Lambe) of the Brul6 of tColossoemys Rodrigues (Pleistocene of Wyoming and comes from the county just south of the Brazil): This name is based on two caudal type locality of the latter species. The questionable form is Lower Miocene in age, however, as the as- vertebrae believed to pertain to a turtle of sociated mammalian fauna clearly shows (Schlaikjer, enormous size. Its significance is very un- 1935, pp. 123-125). A close comparison of al} parts of certain. It may be a tortoise. Rodrigues' the skeleton (fortunately very complete in the Miocene plates have not been seen. form) with the appropriate elements of the United States National Museum specimens from the Brul6 SPECIES described by Gilmore (1946) will be necessary to decide the specific identity of the form. Quite similar cases of The species here listed as of uncertain posi- the extension of the same or a closely related species of tion are so listed for reasons that are very land turtle from one geologic period to another are tStylemys nebrascensis (Brul6 Oligocene), tS. conspecta different in the several cases. tTestudo obtusa (John Day Miocene), tTestudo angusticeps (Sheep Leidy, described from a peripheral, tT. ex- Creek Miocene), and tT. orthopygia (Republican River ornata Lambe, described from a few costals, Pliocene). 28 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 95 and Lynn and tT. farri Hay are probably the latter being more like that of an emydine assignable to Gopherus, but in both cases than like the exaggeratedly projecting epi- some important diagnostic features are not plastra of T. quadrata. known, and one other character is atypical As redefined by Gilmore tT. quadrata is a for the genus. In tT. ducatelli the humnero- doubtfully distinct form the precise affinities pectoral sulcus encroaches on the entoplas- of which can be assessed only when it be- tron; in tT. farri the pectoral scute is nar- comes more completely known. rowed, though not to the extreme of tT. Of named living forms regarded as Ameri- osborniana. can, ohily Testudo schweiggeri Gray, 1831, is The type of tT. quadrata Cope (A.M.N.H. questionable. It was regarded as a relative or No. 1149) shows the gular-humeral sulcus variant of Gopherus (G. polyphemus) by transverse somewhat as in the genus tFlori- Boulenger (1889). The nuchal scute is missing demys (= tBystra Hay). The condition of the in the type and only specimen, and the re- type is probably an individual variation as lationships of the first marginal scutes re- suggested by Gilmore (1946); it is quite semble those of Chelonoidis species rather different from tFloridemys in detail, the than those of such variants of Gopherus as whole contour of the anterior plastral lobe of lack the nuchal. CHECK LIST OF NEW WORLD TESTUDINES AND SUPPOSED TESTUDINES BELOW ARE LISTED the named New World Double asterisks indicate probable nomina testudines and those other named forms that nuda. have been considered at one time or another Geologic age is stated in each case and in testudine. The forms, considerable in num- many instances is revised from that of the ber, that have been named from material original literature in accordance with recent so fragmentary as probably to be indeter- standard work on the nomenclature and minable or of which the types are lost are correlation of the North American con- marked with an asterisk. Some of these tinental Tertiary (Wood et al., 1941) and names may be validated (as was, in effect, Simpson's parallel correlation of the South tT. cubensis) by additional material from American Tertiary (Simpson, 1947). Forms the type locality, but at present they cannot of the same age and phyletic group are per- be defined, synonymized, or allocated.' haps synonyms or races, but determination of 1 Dr. Paulo Vanzolini of the Department of Zoology, this point will require very close study of the State of Sao Paulo, Brazil, has suggested to the author types and referred material. that specific names of this nature be designated species Because most forms were originally de- inq?uirendae. This nomenclatorial category seems to be scribed as Testudo, most imperfectly known unusual or unknown in North American taxonomy. It forms remain in that genus for the purposes would, the present author believes, serve a useful pur- pose. Certainly names of possible but presently in- of this list. Such forms are not to be con- demonstrable validity cannot properly be treated as of sidered as referred to Hesperotestudo (this equivalent value with adequately based, well-known paper). taxonomic categories, nor is it at all reasonable to place such names in synonymy, since this could be done only purgatory in which such names may reside until proper arbitrarily and doubtfully. There is evident need for a disposition can be made of them.

NORTH AMERICA EOCENE *Achilemys allabiata (Cope), 1872. Wyoming: Bridger Cymatholcus longus Clark, 1931. Utah: Duchesne schucherti (Hay), 1899. Alabama: Jackson Hadrianus majusculus Hay, 1904. New Mexico: Wasatch corsoni (Leidy), 1871. Wyoming: Bnrdger; Utah: Uinta tumidus Hay, 1904. Utah: Uinta robustus Gilmore, 1915. Utah: Uinta utahensis Gilmore, 1915. Utah: Uinta Testudo ui-ntensis Gilmore, 1915. Utah: Uinta OLIGOCENE Gopherus laticunea (Cope), 1873. Colorado: Horsetail Creek; Wyoming: Brul6; South Dakota: Chadron; Nebraska: Oreodon beds praeextans (Lambe), 1913. Wyoming: Brul6 Stylemys nebrascensis Leidy, 1851. Nebraska: Oreodon beds; Colorado: Oreodon beds; Wyoming: Oreodon beds; North Dakota: Oreodon beds; South Dakota: Oreodon beds Testudo brontops Marsh, 1890. South Dakota: Chadron amphithorax Cope, 1873. Colorado: Horsetail Creek ligonia Cope, 1873. Colorado: Horsetail Creek *cultrata Cope, 1873. Colorado: Horsetail Creek *quadrata Cope, 1884. Colorado: Horsetail Creek *exornata Lambe, 1906. Saskatchewan: Cypress Hills *thomsoni Hay, 1908. South Dakota: Chadron *peragrans Hay, 1906. Montana: McCarty's Mountain 29 30 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 95 MIOCENE Gopherus brevisterna (Loomis), 1909. Wyoming: Harrison undabuna (Loomis), 1909. Wyoming: Harrison emiliae (Hay), 1908. South Dakota: Rosebud edae (Hay), 1907. Nebraska: Harrison hollandi (Hay), 1907. Nebraska: Harrison vaga (Hay), 1908. Montana: Deep River inusitata (Hay), 1907. Montana: Deep River copei (Koerner), 1940. Montana: Deep River pantsa (Hay), 1908. Colorado: Pawnee Creek mohavense (Merriam), 1919. California: Barstow Stylemys capax Hay, 1908. Oregon: John Day conspecta Hay, 1908. Oregon: John Day *oregonensis Leidy, 1871. Oregon: John Day *calaverensis Sinclair, 1903. California: auriferous gravels Testudo arenivaga Hay, 1907. Nebraska: Harrison angusticeps Matthew, 1924. Nebraska: Sheep Creek impensa Hay, 1908. Montana: Madison Valley osborniana Hay, 1905. Colorado: Pawnee Creek *farri Hay, 1908. Montana: Deep River *ducatelli Collins and Lynn, 1936. Maryland: Calvert *undata Cope, 1875. New Mexico: Santa Fe *klettiana Cope, 1874. New Mexico: Santa Fe PLIOCENE Floridemys nanus (Hay), 1916. Florida: phosphate mine, Holder, Citrus County Gopherus turgida (Cope), 1892. Texas: Blanco campester (Hay), 1908. Texas: Blanco canyonensis (Johnston), 1937. Texas: Blanco riggsi (Hibbard), 1944. Kansas: Seward and ?Meade counties hayi (Sellards), 1916. Florida: Nichols phosphate mine Testudo orthopygia (Cope), 1878. Kansas: Republican River gilberti Hay, 1899. Kansas: Republican River louisekressmani Wark, 1929. Florida: Nichols phosphate mine *pertenuis Cope, 1892. Texas: Blanco *snoviana Cope, 1886. Kansas: northwest part of state PLEISTOCENE Gopherus agassizii (Cooper), 1863. California: McKittrick Asphalt polyphemus (Bartram), 1791. Florida: Vero praecedens Hay, 1916. Florida: Vero *ocalana (Hay), 1916. Florida: Ocala *incisa (Hay), 1916. Florida: Ocala *hexagonata (Cope), 1903. Texas: Tule *atacosae Hay, 1902. Texas: Atacosa County Testudo sellardsi Hay, 1916. Florida: Vero crassiscutata Leidy, 1889. Florida: Peace Creek *distans Hay, 1916. Florida: Ocala *luciae Hay, 1916. Florida: Vero *laticaudata Cope, 1903. Texas: Tule *annae Hay, 1923. Texas: Brazos River *francisi Hay, 1923. Texas: Temple *munda Hay, 1920. Tennessee: Hamblen County *equicomes Hay, 1917. Kansas: Meade County *obtusa (Leidy), 1877. South Carolina: Ashley River 1950 WILLIAMS: TESTUDO CUBENSIS 31 SOUTH AMERICA MIOCENE Testudo gringorum Simpson, 1942. Argentina: Patagonian PLIOCENE Testudo praestans Rovereto, 1914. Argentina: Monte Hermoso *gallardoi Rovereto, 1914. Argentina: Araucanian **australis Moreno, 1890? Argentina: Monte Hermoso **formosa Moreno, 1890? Argentina: Monte Hermoso **paranensis Scalabrini, 18-? Argentina: Entre Rios PLEISTOCENE Testudo elata Gervais, 1877. Brazil: Pampean Testudo cubensis Leidy, 1868. Cuba: Ciego Montero SUMMARY 1. tTestudo cubensis Leidy is redefined and Gopherus. tHadrianus ?schucherti Hay is redescribed. placed in the genus tCymatholcus Clark. 2. A grouping and phylogeny of the New 4. tTestudo cu&ensfis is regarded as an early World testudines are attempted. offshoot of a Central American stock that 3. The species tlaticunea Cope, tprae- also gave rise to the three living Neotropical extans Lambe, tbrevisternat Loomis, tvaga tortoises. Hay, temiliae Hay, tpansa Hay, tinusitata 5. Subgeneric separation is suggested for Hay, tedae Hay, thollandi Hay, tundabuna the Neotropical tortoises (including Loomis, tmohavense Merriam, tcopei Koerner, tcubensis) under the name Chelonoidis fturgida Cope, triggsi Hibbard, tcampester Fitzinger, 1835, and for the very distinct Hay, thexagonata Cope, tocalana Hay, tin- group of Nearctic tortoises typified by tT. cisa Hay, and thayi Sellards, formerly osborniana under the name tHesperotestudo, placed in the genus Testudo, are here allo- new subgenus. cated to Gopherus. The species tcanyonensis 6. The name tBystra Hay, 1916, preoc- Johnston, placed by its author in the mono- cupied by a genus of Hymenoptera, is re- typic genus tBysmachelys, is also referred to placed by tFloridemys, new name.

32 ACKNOWLEDGMENTS The author is indebted to Dr. Bobb available the collections of Princeton Uni- Schaeffer and Dr. H. E. Anthony for calling versity. Mr. M. Graham Netting has per- his attention to a topic (the Pleistocene mitted the study of the Recent turtle collec- turtles of the West Indies) which has proved tions at Carnegie Museum. Dr. E. R. Hall, so fruitful. He is grateful also to Dr. Schaeffer Dr. E. H. Taylor, Dr. R. Wilson, and Mr. for reading the manuscript in various stages E. C. Galbreath have extended the author of preparation and for many useful sug- many courtesies during his examination of gestions. To Dr. E. H. Colbert he owes a debt the important collection of living and fossil of gratitude for active interest, helpful turtles at the University of Kansas. Dr. J. T. criticism, and generous assistance. The com- Gregory has opened the collections of the pletion of the study has been made possible Peabody Museum of Yale University to the by the generosity of Mr. C. M. Bogert in author's inspection. Dr. Horace Richards of affording to the author free use of the the Academy of Natural Sciences of Phila- facilities and ready access to the comparative delphia procured for the author the photo- materials available in his department. Dr. G. graphs of the type of tT. cubensis which ap- G. Simpson has given the author the benefit of pear as plate 1. Dr. David Dunkle facilitated his advice and criticism. Dr. Alfred Romer, the study of the fossil turtles at the United Miss Nelda Wright, Mr. Arthur Loveridge, States National Museum, and Dr. Doris and Mr. Benjamin Shreve have graciously Cochran permitted the use of the recent given every possible assistance to the author turtle collections and very generously made in the use of the collections of the Museum available George Baur's MS notes on the of Comparative Zo6logy. Dr. W. J. Hamilton genus Gopherus. Dr. H. K. Svenson and has generously allowed the author to study Brother Leon have assisted by attempting at his.leisure skeletal material from Cornell the identification of the botanical material University. Dr. Glenn L. Jepsen has made associated with tT. cubensis.

33 BIBLIOGRAPHY AMEGHINO, F. limbs. Contrib. Mus. Paleont. Univ. 1921. Obras completas y correspondencia Michigan, vol. 5, pp. 69-73. cientifica. La Plata, Primera sinopsis CLARK, J. geologico-paleontologica, vol. 12. 1931. A new turtle from the Duchesne Oligo- ANTHONY, H. E. cene of the Uinta basin, northeastern 1919. Mammals collected in eastern Cuba in Utah. Ann. Carnegie Mus., vol. 21, pp. 1917 with descriptions of two new 131-160. species. Bull. Amer. Mus. Nat. Hist., COLLINS, R. L., AND W. G. LYNN vol. 41, pp. 625-643. 1936. Fossil turtles from Maryland. Proc. 1920a. New mammals from Jamaica. rbid., Amer. Phil. Soc., vol. 76, pp. 157-173. vol. 42, pp. 469-475. DARLINGTON, P. J. 1920b. A zoologist in Jamaica. Nat. Hist., vol. 1938. The origin of the fauna of the Greater 20, pp. 156-168. Antilles with a discussion of dispersal BARBOUR, T. of animals over water and through the 1934. Observations on Antillean tortoises and air. Quart. Rev. Biol., vol. 13, pp. 274- terrapins. Copeia, pp. 111-113. 300. BAUR, G. FLOWER, S. 1892. Bemerkungen iuber verschiedene Arten 1937. Further notes on the duration of life in von Schildkrbten. Zool. Anz., no. 389, animals. III. Reptiles. Proc. Zool. Soc. pp. 155-159. London, pp. 1-39. [MS.] Notes on Gopherus. FREIBERG, M. A. BERGOUNIOUX, F. M. 1938. Catalogo systematico y descriptivo de 1935. Contribution a l'etude paleontologique las tortugas argentinas. Mem. Mus. des cheloniens: cheloniens fossiles du Entre Rios, vol. 9, pp. 1-23. Bassin d' Aquitaine. Mem. Soc. Geol. GADOW, H. France, new ser., vol. 11, pp. 1-215. 1901. Amphibians and reptiles. London. BOGERT, C. M., AND J. A. OLIVER 1945. A preliminary analysis of the herpeto- GERVAIS, P. fauna of Sonora. Bull. Amer. Mus. Nat. 1877. Tortue gigantesque fossil au Bresil. Hist., vol. 83, pp. 303-425. Jour. Zool., vol. 6, pp. 281-285. BOULENGER, G. A. GILMORE, C. W. 1889. Catalogue of the chelonians, rhynco- 1915. Turtles of the Uinta. Mem. Carnegie cephalians, and crocodiles in the British Mus., vol. 7, pp. 101-161. Museum (Natural History). London. 1931. Fossil turtles of Mongolia. Bull. Amer. 1890. Amphibians and reptiles. Zool. Rec., Mus. Nat. Hist., vol. 59, pp. 213-257. vol. 26, pp. 1-23. 1946. The osteology of the fossil turtle Testudo BROWN, B. praeextans Lambe with notes on other 1913. Some Cuban fossils. Amer. Mus. Jour., species of Testudo from the Oligocene vol. 13, pp. 221-228. of Wyoming. Proc. U. S. Natl. Mus., [MS.] Annual report [19191 of the Department vol. 96, pp. 293-310. of Vertebrate Paleontology, the Ameri- GLAESSNER, M. F. can Museum of Natural History. [Un- 1933. Die Tertiarschildkr6ten Nieder5ster- published.] reichs. Neues Jahrb. Min. Geol. Pale- CAMERON, P. ont., Abt. B., vol. 69, pp. 353-387. 1902. Descriptions of new genera and species GRANT, C. of Hymenoptera from the oriental 1936. The southwestern desert tortoise Go- zoological region (Ichneumonidae, Fos- Pherus agassizii. Zoologica, vol. 21, sores, and Anthophila). Ann. Mag. Nat. pp. 225-229. Hist., ser. 7, vol. 9, pp. 145-155. GRANT, C., AND C. R. DE SOLA CASE, E. C. 1934. Antillean tortoises and terrapins: Dis- 1925. A specimen of Stylemys nebracensis tribution, status and habits of Testudo Leidy with the skull preserved. Contrib. and Pseudemys. Copeia, pp. 73-79. Mus. Geol. Univ. Michigan, vol. 2, GRAY, J. E. pp. 87-91. 1870. Notes on three species of tortoises living 1936. A specimen of Stylemys nebracensis in the Society's gardens. Proc. Zool. Leidy showing the bones of the feet and Soc. London, pp. 706-708. 34 1950 WILLIAMS: TESTUDO CUBENSIS 35 1872. Appendix to the catalogue of shield Amer. Jour. Sci., ser. 4, vol. 28, pp. reptiles in the collection of the British 17-26. Museum. Part I. Testudinata (tor- 1927. A giant tortoise from Florida. Ibid., toises). London. ser. 5, vol. 13, pp. 435-439. HAY, 0. P. LOVERIDGE, A. 1905. The fossil turtles of the Bridger Basin. 1947. Bonemaking materialfor turtles. Copeia, Amer. Geol., vol. 35, pp. 327-342. p. 136. 1908. Fossil turtles of North America. Wash- MATTHEW, W. D. ington. 1918. Affinities and origin of the Antillean 1916. Description of some Floridian fossil mammals. Bull. Geol. Soc. Amer., vol. vertebrates belonging mostly to the 29, pp. 657-666. Pleistocene. Rept. Florida Geol. Surv., 1919. Recent discoveries of fossil vertebrates vol. 8, pp. 39-76. in the West Indies and their bearing on 1917a. On a collection of fossil vertebrates the origin of the Antillean fauna. Proc. made by Dr. F. W. Cragin in the Equus Amer. Phil. Soc., vol. 58, pp. 161-181. beds of Kansas. Kansas Univ. Sci. 1924. Third contribution to the Snake Creek Bull., vol. 10, pp. 39-5 1. fauna. Bull. Amer. Mus. Nat. Hist., 1917b. Vertebrates mostly from Stratum 3 at vol. 50, pp. 59-210. Vero, Florida, together with descrip- [MS.] Testudo cubensis Leidy. [Unpublished, tions of new species. Rept. Florida written 192-?] Geol. Surv., vol. 9, pp. 43-68. 1931. Genera and new species of ground sloths 1920. Description of some Pleistocene verte- from the Pleistocene of Cuba, with a brates found in the United States. Proc. prefatory note by W. Granger. Amer. U. S. Natl. Mus., vol. 58, pp. 83-146. Mus. Novitates, no. 511, pp. 1-5. 1923. Characteristics of sundry fossil verte- MATTHEW, W. D., AND W. GRANGER brates. Pan Amer. Geol., vol. 39, pp. 1923. The fauna of the Ardyn Obo formation. 101-120. Amer. Mus. Novitates, no. 98, pp. 1-5. HIBBARD, C. MERRIAM, J. C. 1944. A new land tortoise from the Middle 1919. Tertiary mammalian faunas of the Pliocene of Seward County, Kansas. Mohave Desert. Bull. Dept. Geol. Univ. Kansas Univ. Sci. Bull., vol. 30, pp. California, vol. 11, pp. 438-585. 71-76. MERTENS, R. JOHNSTON, C. S. 1942. Bemerkungen zur Sektion einer Elefan- 1937. Osteology of Bysmachelys canyonensis, ten-Schildkr6te, Zool. Garten, new ser., a new turtle from the Pliocene of vol. 14, pp. 276-277. Texas. MILLER, L. Jour. Geol., vol. 45, pp. 439-447. 1942. A Pleistocene tortoise from the McKit- KOERNER, H. E. trick Asphalt. Trans. San Diego Soc. 1940. The geology and paleontology of the Nat. Hist., vol. 9, pp. 439-442. Fort Logan and Deep River formations NICHOLSON, H. A., AND R. LYDE1KKER of Montana. Amer. Jour. Sci., vol. 238, 1889. Manual of paleontology. London, vol. 2. pp. 837-862. PEYER, B. LAMBE, L. M. 1942. Fossile Riesenschildkr6ten aus der ober- 1913. Description of a new species of Testudo en Susswassermolasse der Umgebung von and of a remarkable specimen of Zurich. Abhandl. Schweizerische Palae- Stylemys nebrascensis from the Oligocene ont. Gesellsch., vol. 63, pp. 1-47. of Wyoming. Ottawa Nat., vol. 27, pp. PROCTOR, JOAN 57-63. 1922. A study of the remarkable tortoise LEIDY, J. Testudo loveridgii Blgr. and the mor- 1868. Notice of some vertebrate remains from phology of the Chelonian carapace. the West Indian islands. Proc. Acad. Proc. Zool. Soc. London, pp, 483-526. Nat. Sci. Philadelphia, pp. 178-180. RIABININ, A. LINDEOLM, W. A. 1927. Un noveau chelonien du Tertiaire in- 1929. Revidiertes Verzeichnis der Gattungen ferieur des environs du Lac Issyk-koul. der rezenten Schildkroten nebst Notizen Bull. Com. Geol. Leningrad, vol. 46, zur Nomenklatur der Arten. Zool. Anz., pp. 193-199. vol. 81, pp. 275-295. RODRIGUES, J. B. LooMIs, F. B. 1892. Les reptiles fossiles de la val6e de 1909. Turtles from the upper Harrison beds. l'Amazone. Vellosia, vol. 2, pp. 41-56. 36 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 95 ROTHSCHILD, L. Latin America. Amer. Jour. Sci., vol. 1931. Exhibition of specimens of the saddle- 241, pp. 413-429. backed races of Galapagos tortoises. 1945. The principles of classification and a Proc. Zool. Soc. London, p. 1208. classification of mammals. Bull. Amer. ROVERETO, C. Mus. Nat. Hist., vol. 85, pp. 1-350. 1914. Los Estratos Araucanos y sus Fosiles. 1947. A continental Tertiary time chart. Ann. Mus. Nac. Hist. Nat. Buenos Jour. Paleont., vol. 21, pp. 490-493. Aires, vol. 25, pp. 1-249. SMITH, MALCOLM SCHLAIKJER, E. M. 1947. Amphibia and reptiles. Zool. Rec., vol. 1935. Contributions to the stratigraphy and 81, pp. 1-27. paleontology of the Goshen Hole area, STEFANO, G. DE Wyoming. IV. New vertebrates and the 1902. Stylemys botti n.f. Riv. Italiana Paleo.nt., stratigraphy of the Oligocene and early vol. 8, pp. 72-76. Miocene. Bull. Mus. Comp. Zool., vol. SWINTON, W. E. 76, pp. 97-189. 1928. Note on the fossil Reptilia collected by SCHMIDT, K. P. Mr. Donald Stewart in Venezuela. 1926. The amphibians and reptiles of Mona Jour. Geol. Island. Field Mus. Nat. Hist., zool. Quart. Soc., vol. 84, p. 583. ser., vol. 12, pp. 147-164. TORRE, C. DE LA SCHLUCHERT, C. 1910. Restoration of Megalocnus rodens and 1935. Historical geology of the Antillean- discovery of a continental Pleistocene Caribbean region or the lands bordering fauna in central Cuba. Compte Rendu the Gulf of Mexico and the Caribbean Congr. Geol. Internatl., XIe Session, Sea. New York. Stockholm, fasc. 2, pp. 1023-1024. SELLARDS, E. H. WARIc, A. F. 1916. A new tortoise and a supplementary 1929. A new giant tortoise from the Pliocene note on the gavial Tomistoma ameri- of Florida. Amer. Jour. Sci., ser. 5, cana. Amer. Jour. Sci., ser. 4, vol. 42, vol. 17, pp. 400-402. pp. 235-240. WEEKS, L. G. SIEBENROCK, F. 1948. Paleogeography of South America. Bull. 1919. Emydura macguarris Gray und der Geol. Soc. Amer., vol. 59, pp. 249-282. systematische Wert des Nuchalschildes WETMORE, A. bei den Schildkr6ten. Zool. Anz., vol. 1928. Bones of birds from the Ciego Montero 50, pp. 273-277. deposits of Cuba. Amer. Mus. Novi- SIMPSON, G. G. tates, no. 301, pp. 1-5. 1933. Glossary and correlation charts of North WOOD, H. E., ET AL. American Tertiary mammal-bearing 1941. Nomenclature and correlation of the formations. Bull. Amer. Mus. Nat. North American continental Tertiary. Hist., vol. 67, pp. 79-121. Bull. Geol. Soc. Amer., vol. 52, pp. 1-48. 1942. A Miocene tortoise from Patagonia. WOODBURY, A. N., AND R. HARDY Amer. Mus. Novitates, no. 1209, pp. 1948. Studies of the desert tortoise, Gopherus 1-6. agassizii. Ecol. Monogr., vol. 18, pp. 1943. Turtles and the origin of the fauna of 145-200.