AMERICAN MUSEUM Novltates Published by Number 1209 the AMERICAN MUSEUM of NATURAL HISTORY November 13, 1942 New York City

AMERICAN MUSEUM NOVlTATES Published by Number 1209 THE AMERICAN MUSEUM OF NATURAL HISTORY November 13, 1942 New York City

A MIOCENE TORTOISE FROM PATAGONIA' BY GEORGE GAYLORD SIMPSON

Some years ago the Scarritt Patagonian interest in a group already containing so Expedition, under my direction, picked up many species, but, more important, it a fossil tortoise shell in the Chubut Valley, represents a terrestrial Holarctic genus central Patagonia. Such specimens are appearing in South America before the seldom of broad scientific interest, and preparation was postponed in favor of late Tertiary land-bridge existed. It also more probably fruitful work. On being casts light on the interesting question of prepared, however, this specimen did prove the origin of the recent South American to be of unusual significance. The fact tortoises and on the widely discussed that it is a new species has no particular Galapagos tortoises.

TAXONOMY Testudo gringorum, new species border of carapace somewhat flattened but only slightly turned upward. Nuchal absent. Testu- TYPE-.Amer. Mus. No. 3366 (Fossil Reptile dinine alternating-wedge arrangement of costa] Catalogue), plastron and much of carapace. plates moderately developed. Marginals longer Collected by Justino Hernandez, Nov. 21, 1933. in vertical-transverse than in anteroposterior HYPODIGM.-Type only. dimensions. Gulars separate, elongate, ex- HORIZON AND LOCALITY.-High in the pre- tending onto entoplastron posteriorly. Epi- dominantly marine Patagonian section over- plastral lip more than twice as wide as long, lying the fissile or laminated beds of Ango- strongly projecting and bifid anteriorly, sturas, south side of the Chubut Valley, be- emarginate at gular-humeral sulcus, border tween Gaiman and Dolavon, Chubut Territory, evenly convex posterior to this. Humeral- Argentina. pectoral sulcus almost straight and transverse DIAGNOSIS.-As far as preserved, a char- medially, turning abruptly anteroexternally acteristic Testudo, sensu stricto. Type rela- in lateral part, crossing mid-line at posterior tively small for this genus (possibly young). edge of entoplastron. Width of medial part Uncrushed carapace probably not very high. of pectoral scutes about one-fifth that of ab- Sculpture slight, no prominent bosses or mar- dominals. Posterior notch of plastron well ginal projections. Sulci shallow. Anterior developed, about four times as wide as deep.

AGE Unfortunately the horizon of the speci- as to the correct identification of the men cannot be precisely defined in strati- lower beds of the Angosturas section, a graphic terms, but the relative age can be thick series of thin-bedded sands, clays and determined with some probability and tuffs. Above these, however, are predomi- within moderate limits. The Chubut nantly marine beds referable to the Valley section about five kilometers west Patagonian with little doubt. The fossil of this locality has been studied in detail was found near the top of these beds. The (Frenguelli, 1927; Simpson, 1935), but exact limit between true or typical Pata- the beds cannot be traced continuously, gonian and the overlying Super-Pata- and they are markedly different in the two gonian, Upper Patagonian or Aonikense of localities. No agreement has been reached Frenguelli is not absolutely identified here, 1 Publications of the Scarritt Expeditions, No. 32. and the tortoise may have been either in the 2 AMERICAN MUSEUM NOVITATES [No. 1209 highest typical Patagonian or the lowest Marine fossils occurred a few meters both "Aonikense," probably the former. In any below and above the tortoise, but were case, these two marine formations are of so not seen in the same bed, which was a nearly the same age that their separability fairly massive yellow-gray sandstone. The is sometimes questioned. The horizon may predominantly marine series was here be early or middle Miocene in age, more evidently deposited near shore and has likely the former (depending also, of course, recurrent definitely shore-line facies, per- on the disputed correlation of the Pata- haps also some truly fluviatile interca- gonian). It is most unlikely but is barely possible that the horizon is really part of lations. The tortoise must be inter- the later Tertiary sands that also occur preted as a terrestrial fossil, but its in this region, in which case the age could occurrence even in a purely marine bed be Mesopotamian or Huayquerian, latest (nearshore) would not necessarily be anoma- Miocene or earliest Pliocene. lous.

DESCRIPTION Part of the carapace had decayed before served only from the third neural forward burial, leaving a hole in the posteromedial and is not entirely clear even where present. and right posterolateral parts. Thus The plastron is sufficiently preserved to weakened, the carapace was considerably reveal all its characters. The morphology crushed by compacting of the sediments. is adequately given in the diagnosis and Nevertheless the costal region of the left illustrations. side can be almost completely made out The length of the plastron is 207 mm. and the marginal region from the small The median width of the shell was prob- nuchal notch as far back as the sixth periph- ably about 18-19 cm., and the length of eral. The region of the midline is pre- the carapace 23-24 cm.

AFFINITIES The known parts do not appear to have is a Chelys, it is quite unrelated to the any characters not found in species cur- present species. If it is a Testudo, it is rently referred to Testudo in a strict sense. distinct from T. gringorum specifically It is, of course, possible that more complete because it is said to have heavy costal material, including the skull, would war- bosses, absent in the new species. Virtu- rant some other generic assignment, but ally no other possibly specific characters this is unlikely since the structures known were mentioned by Wieland. in this specimen are, as a rule, diagnostic. Rovereto (1914) described two relatively In any case the relationship to Testudo good Tertiary Testudo from Argentina. must be close. T. gallardoi is from the typical Araucanian Wieland (1923) described but did not of Catamarca and therefore almost surely illustrate a fragmentary specimen from an later than our specimen. Vertebral and unknown horizon and locality supposedly costal scutes had large bosses, as in Wie- in Patagonia. The specimen has not been land's specimen (which he did not com- available for comparison with T. gringorum. pare with Rovereto's prior description) but Wieland compared it with Testudo and not in ours. The species are surely dis- allied genera and suggested that its char- tinct, but little basis for ampler comparison acters tended to indicate pertinence to that with T. gringorum is available. T. praes- genus, but he referred it questionably to tans Rovereto from Monte Hermoso is the very different genus Chelys, calling it given and the name was said to be "purely arbitrary," Chelys (?) patagonica.1 If that specimen I judge that the definition of a new species was not really intended and that this name has no legal status 1 Since no supposedly distinctive characters were in formal nomenclature. 19421 A MIOCENE TORTOISE FROM PATAGONIA 3

certainly much later than our species, but longer and narrower (transversely) anals. it considerably resembles the latter, and Fossil and living species are evidently nearly the two may well be related-could be allied but not identical. ancestor and descendant. They are never- The various Galapagos tortoises, sup- theless distinct in numerous minor points, posed by Garman (1917) to be derived e.g., the less produced and less bifid epi- from an ancestor related to T. tabulata, plastral lip of T. praestans and its notably may represent more highly evolved deri- more posterior and curved hypoplastron- vatives of the same Neotropical stock. xiphoplastron suture. They are basically like T. tabulata and may In both the latter characters and many be more immediately allied to the latter others our fossil resembles the living T. than to our extinct species, which could argentina and T. tabulata, to which it seems nevertheless be in or near the ancestry of to have special relationship within the all. genus, and to either or both of which it Most of the characters of T. gringorurn could well be ancestral. There is, how- can be found singly among the many ever, no question that the fossil merits Tertiary North American species of Testudo specific distinction. From T. argentina it but not in this combination. As far as I differs in the less wedge-shaped and less have observed, the nuchal is always pres- medially narrowed pectorals, more elon- ent in the North American species but gate marginals, less recurved anterior absent in South American and Galapagos carapace margin, etc. T. gringorum was probably less narrow and deep than T. species, recent and fossil. T. gringorum tabulata, although crushing obscures this could theoretically be derived from some of character, and there are numerous slight the earlier North American forms, but I differences such as the deeper gular notch of see no clear evidence of special affinity the fossil, less wedge-shaped pectorals, with one species rather than another. shorter (anteroposteriorly) femorals and Hadrianus and Stylemys are not so close.

GEOGRAPHIC IMPLICATIONS Testudinids are somewhat doubtful in that a discontinuous, stepping-stone migra- the late Cretaceous, surely known from the tion route to and from Holarctica arose in Eocene. The genus Testudo itself is not or just before Mesopotamian time (prob- known to have existed until about the end ably latest Miocene) and a continuous of the Eocene. It was widespread in bridge little or not at all before Chapad- Holarctica in Oligocene and later times, malalan (latest Pliocene),. ~irPXatagonian became virtually world-wide and survived Testudo is surely much older than Chapad- in the Recent in disjunctive insular and malalan and probably older than Meso- some continental species, mainly southern. potamian. If Testudo reached South America by an Although terrestrial, Testudo has almost early land-bridge, this must have been certainly crossed considerable widths of Paleocene at latest, but the genus may not salt water, probably accidentally carried by have been in existence at that time. The currents. It floats readily and can survive most reasonable hypothesis involving such long periods of involuntary immersion in an early bridge must, then, be that the the sea. Once within sight of land, it can South American Tertiary Testudo devel- and does purposefully swim to it if tide and oped independent of Holarctic Testudo currents permit. by parallelism from a common ancestor. The most probable explanation of the This hypothesis is so contrary to prob- mid-Tertiary appearance of Testudo in ability and so lacking in evidence that it South America is that it reached there cannot now be seriously supported. without a land-bridge, by a partly marine On the relatively abundant mammalian route. North America is the most likely evidence (see Simpson, 1940), I believe source, although Africa is possible. In 4 AIIERICAN MTUSEUAI NOVITATES [No. 1209

either case, but especially if North America American isthmus and that later, becom- was the source, the (doubtless involuntary) ing more numerous or closer together, migration would be facilitated by islands, facilitated the first invasion of proeyo- such as those that preceded a Central nids.

BEARING ON THE GALAPAGOS PROBLEM The occurrence of well-differentiated American mainland in Miocene, and prob- land tortoises on the Galapagos Islands has ably early Miocene, times. It thus re- given rise to extensive argument and dis- moves the necessity for postulating any cussion which will not be reviewed in this other geographic derivation of the Gala- brief paper. The present fossil does not pagos tortoises. It also enormously in- rigidly prove anything in this respect, but creases the evident timne available for tlhose it does provide some interesting new data forms to have made an accidental crossing and suggestions. It adds to the evidlence from the mainland and gireatly increases the that the distribution of Testudo is not chances of such an event, which is im- dependent on and not in the main a result probable in a short time but is probable if of land connections. It demonstrates the the time is sufficiently long. It also in- presence of a possible ancestral stock for creases the possible period available for the the Galapagos tortoises on the South specific differentiation of the insular forms.

REFERENCES FRENGUELLI, J. SIMPSON, G. G. 1927. El Entreiriense de Golfo Nuevo en el 1935. Early and im-iddle Tertiary geology Chubut. Bol. Acad. Nac. Ci. C6rdoba of the Gaiman region, Chubut, Ar- (Argentina), XXIX, pp. 191-270. gentina. Amrier. Mus. Novitates, No. GARMAN, S. 644. 1917. The Galapagos tortoises. MeIn. Mus. 1940. Review of the mammal-bearirng Ter- Comp. Zool. Harvard, XXX, No. 4. tiary of South America. Proc. Amer. ROVERETO, C. Phil. Soc., LXXXIII, pp. 649-709. 1914. Los estratos araucanos y sus f6siles. WIELAND, G. R. An. Mus. Nac. Hist. Nat. Buenos 1923. A inew Paran:1 pleurodiran. Amner. Aires, XXV, pp. 1-250. Jour. Sci., (5) V, pp. 1-14. 19421 A MIOCENE TORTOISE FROM PATAGONIA 5

- .7 A.M36/. 2 _,' A.-M.-3366

Fig. 1. Testudo gringorum, new species. Type, Amer. Mus. No. 3366 (Fossil Reptile Cata- logue). Dorsal view of carapace, one-half natural size. Distortion has been corrected and missing parts restored (cross-hatched where determined by surrounding bone, blank with broken outline where hypothetical). Sulci between scutes dotted, bone sutures and margins solid lines. Drawn by John C. Germann. Abbreviations (horny scute areas labeled on left side, bony plates on right side): c. p., costal plates, as numbered; c. s., costal scutes, as numbered; n., neural plates, as numbered; spy., supra- pygal plate; v. s., vertebral scutes, as numbered. 6 AMERICAN MUSEUM NOVITATES [No. 1209

A.M.3366 N" -

Fig. 2. Testudo gringorum, new species. Type, Amer. Mus. No. 3366 (Fossil Reptile Cata- logue). Ventral view of plastron, one-half natural size. Technique of illustration as in Fig. 1. Drawn by John C. Germann. Abbreviations: ab., abdominal scute; an., anal scute; ent., entoplastron (plate); epi., epi- plastron (plate); fem., femoral scute; g., gular scute; hum., humeral scute; hyo., hyoplastron (plate); hypo., hypoplastron (plate); pec., pectoral scute; xiph., xiphiplastron (plate).