FOSSIL TURTLE RESEARCH

FOSSIL TURTLE RESEARCH VOLUME 1 Proceedings of the Symposium on Turtle Origins, Evolution and Systematics August 18 20, 2003, St. Petersburg, Russia Edited by Igor G. Danilov and James F. Parham St. Petersburg, 2006

FOSSIL TURTLE RESEARCH VOLUME 1 Editors: Igor G. Danilov and James F. Parham Proceedings of the Symposium on Turtle Origins, Evolution and Systematics August 18 20, 2003, St. Petersburg, Russia Published in St. Petersburg, March 2006 Papers should be cited as (e.g.): Joyce W. G. and Karl H.-V. (2006), «The world s oldest fossil turtle: fact versus fiction,» in: Danilov I. G. and Parham J. F. (eds.), Fossil Turtle Research, Vol. 1, Russ. J. Herpetol., 13(Suppl.), pp. 104-111. This issue is published with the financial support of Dr. Ren Hirayama, grants of the President of the Russian Federation to the Leading Scientific Schools (Nsh-1647.2003.4 and Nsh-4212.2006.4), grant of the Russian Foundation for Basic Research 04-05-65000-a and with the use of the office and laboratory facilities of the Zoological Institute of the Russian Academy of Sciences. Cover photo: PIN 52-1a, holotype of Yaxartemys longicauda Riabinin, 1948, Upper Jurassic of Kazakhstan, Karatau Ridge, vicinity of Mikhailovka village Photograph: Igor Danilov ISSN 1026-2296

Fossil Turtle Research Vol. 1, 2006, pp. 104 111 THE WORLD S OLDEST FOSSIL TURTLE: FACT VERSUS FICTION Walter G. Joyce 1 and Hans-Volker Karl 2 Priscochelys hegnabrunnensis is a fossil vertebrate previously described from a single fragment from the Middle Triassic (lower Ladinian) of Hegnabrunn, Germany. A morphological review of the type specimen confirms the identity of this taxon as a turtle, as opposed to a placodont, based on the lack of distinct bony sutures and the presence of irregularly sized, conical scutes that exhibit highly distinct sulci. The apparent presence of a larger number of supramarginals relative to Proganochelys quenstedti indicates that this taxon may be the most basal known turtle to date. The habitat preferences of Priscochelys hegnabrunnensis remain uncertain, given that the available material was found mixed with marine and terrestrial faunas. INTRODUCTION This contribution is concerned with a single, fragmentary vertebrate fossil (SMNS 80141) from the Middle Triassic of Germany, that has previously been reported as the oldest fossil turtle occurrence (Wild, 1972). Unfortunately, SMNS 80141 was neither figured nor described by Wild (1972), making it impossible for a wider audience to scrutinize its identity as a turtle. It was nevertheless occasionally cited in the literature as the potentially oldest turtle (e.g., Młynarski, 1976; Rieppel and Reisz, 1999; Lucas et al., 2000) until it was formally described as a new species of fossil turtle, Priscochelys hegnabrunnensis Karl, 2005. Although Karl s (2005) description and diagnosis of Priscochelys hegnabrunnensis are sufficient to establish an available taxon name according to the rules of zoological nomenclature, those that doubt its identity as a basal turtle may consider the description too brief. To alleviate the situation, SMNS 80141 will be redescribed herein and 1 Peabody Museum of Natural History, Yale University, 170 Whitney Avenue, New Haven, Connecticut 06520, USA. E-mail: walter.joyce@yale.edu 2 Department of Geography, Geology and Mineralogy, University of Salzburg, Hellbrunner Straße 34, A-5020 Salzburg, Austria its identity as a basal turtle argued in detail. All information that pertains to SMNS 80141 will be summarized herein. Some may thus be partially redundant with points made adequately by Karl (2005). Terminology The nomenclature to the turtle shell used herein follows that of Zangerl (1969). Many taxon names used throughout the text are clade names and follow the definitions of Joyce et al. (2004). They are placed throughout the text in small capital letters to distinguish them from traditional, ranked-based taxon names. Institutional abbreviations SMNS, Staat liches Museum für Naturkunde Stuttgart, Germany. GEOLOGICAL SETTING AND HISTORY OF FINDING The putative fossil turtle fragment (SMNS 80141) was acquired by the Staatliches Museum für Naturkunde Stuttgart, Germany, in 1988 as part of the private Muschelkalk fossil collection of M. Wild of Kulmbach, Germany. The specimen was found on March 6th 1969 in the subsequently abandoned Schmidt Quarry, located half way along the road that connects the villages of Hegnabrunn and Feuln near the Upper

The world s oldest fossil turtle 105 Franconian town of Kulmbach, Germany. The Gauss-Krüger Coordinates of the quarry are R4468125/H5550075 (Fig. 1). Sixteen meters of Upper Muschelkalk sediments are exposed in the Schmidt Quarry that can be attributed to the lower spinosus to the upper nodosus ammonoid biozones. The fossil was found embedded in a fist-sized clay-rich concretion that was located in a limestone bench of the central section of the enodis/laevigatus ammonoid biozone. The find can thus be placed chronostratigraphically into the lower Ladinian, Middle Triassic. The fossil fragment was found shattered and minor bits were thus invariably lost. The vast majority of parts, however, could be recovered from the geode, were cleaned with water, and later reassembled with minor restoration (Fig. 2). All details regarding the circumstances of this find and the stratigraphy of the Schmidt quarry are based on personal communication of R. Wild (SMNS) summarized in Karl (2005). During the Middle Triassic, a particularly shallow epicontinental sea covered much of central Europe (Germanic facies) that was incompletely connected towards the south with the open sea (Alpine facies), and thus almost fully surrounded by land. Additional vertebrate fossils find from the Germanic facies of Upper Franconia include numerous shallow marine taxa, including sharks, teleosts, placodonts, nothosaurs, pistosaurs, and protorosaurs. Numerous finds of temnospondyls and lungfish, however, suggest a partial terrestrial Fig. 1. Map of southern Germany showing the location of the Schmidt Quarry near Hegnabrunn, Upper Franconia.

106 Walter G. Joyce and Hans-Volker Karl Fig. 2. Photographs of the A) dorsal and B) ventral side of SMNS 80141, holotype of Priscochelys hegnabrunnensis, from the Middle Triassic (lower Ladinian) near Hegnabrunn, Upper Franconia, Germany. The six scute-like fields on the dorsal surface are numbered with Roman numerals to avoid making homology statements. influence as well. Given the close proximity of land during the Middle Triassic, it appears plausible that terrestrial faunas were occasionally washed into the basin, probably from the Bohemian Massif to the east (Wild, 1972). DESCRIPTION OF SMNS 80141 SMNS 80141 is an oddly shaped bone fragment with a maximum width of 84 mm, a maximum height of 73 mm, and a maximum thickness of 12 mm. Given that the anatomical identification remains uncertain, it is impossible to unambiguously orient the specimen. However, because we ultimately conclude the specimen to be a carapacial fragment of a stem representative of the turtle crown, we refer to the more heavily sculptured side as the dorsal surface and the smooth side as the ventral or visceral surface. For simplicity, the edge with the numerous smaller dorsal scute-like fields is referred to as the posterior. Figure 2 follows this convention as well. The specimen generally appears weathered, primarily along the dorsal surface, which is damaged by numerous pockmarks. Postmortem processes probably created these while the specimen was lying on the seafloor. The ventral side, in contrast, is rather smooth, perhaps because it was protected from scavenging. All edges of SMNS 80141 show signs of mechanical breakage revealing the bony nature of the specimen along the broken edges. Determination of the original boundaries of the edge of the specimen is thus impossible. Based on comparisons with other turtle material, however, we conclude SMNS 80141 to be a carapacial element and thus conclude that all edges are indeed broken (see Discussion below). No sutures or foramina are visible on the entire specimen. In lateral view, the specimen can be subdivided into an anterior and posterior section of subequal size that are angled 20 degrees relative to another. A slight thickening accompanies the area where the two sections meet. The ventral surface of SMNS 80141 is rather smooth and lacks any notable features. The dorsal

The world s oldest fossil turtle 107 surface, in contrast, is characterized by numerous deep sulci that subdivide the surface into six fields, which will be referred to as fields I through VI herein to avoid making homology statements for the moment (Fig. 2). The sulci generally resemble those of turtles, by being u-shaped in cross section and by gently melting together at intersections, but they are significantly deeper than those of any previously described representative of TESTUDINATA. None of the fields are equivalent in size or in shape. The specimen lacks any true symmetry. Field I is clearly surrounded by sulci along its posteromedial contact with field II, its anteromedial contact with field VI, and along the anterolateral edge of the field. A clear sulcus is missing along the posterolateral edge of the field. Field II is pentagonal with clear sulci along all edges, including those along the posterior margin of the specimen. Field III is notably smaller and rectangular. All four surrounding sulci are clearly visible on the dorsal surface of the specimen. Field IV is pentagonal, similar in shape to field II but was probably significantly larger. Fields I through IV are all somewhat conical in shape and exhibit pronounced central tubercles. The original shape of field V is unclear, given that the lateral margins are represented by breaks, and it is possible that this field actually represents two fields that are separated by a poorly developed sulcus. A clear sulcus is developed along the posteromedial contact with field IV and the anteromedial contact with VI, although the adjacent areas of those field are reconstructed. A central tubercle is not present, or at least not preserved. Field VI is the largest field, representing about fifty percent of the dorsal surface. It contacts all five more posteriorly positioned fields along a bowed sulcus. The surface of this field is rather flat and lacks a central tubercle. DISCUSSION Taxonomic identity Providing an unambiguous identification of SMNS 80141 is difficult for a number of theoretical and practical reasons. First, as is common for paleontological specimens, missing data hampers a precise taxonomic placement. Second, despite a long list of spectacular finds, our understanding of Triassic vertebrates remains limited and unusual new vertebrates are still uncovered from this period at regular intervals. This makes any assignment based on known faunas somewhat speculative. Numerous Triassic vertebrate groups are characterized by the presence of osteoderms. These include temnospondyls, placodonts, turtles, and numerous basal archosaurs such as aetosaurs and phytosaurs. To date, however, only two taxa are known from any time period to exhibit interlocking osteoderms with scutes and sulci: cyamodontoid placodonts and turtles (Carroll, 1988; Rieppel, 2002). Although the possibility remains that SMNS 80141 represents an unknown group with these features, we avoid such ad hoc arguments, focus on the positive evidence that is available, and discuss similarities and differences of SMNS 80141 with representatives of Cyamodontoidea and TESTUDINATA. The bony shell of those placodonts that possess more or less enclosed theca is comprised of bony elements of significant thickness that are overlain by epidermal scutes. Although the actual scutes are not preserved, some taxa exhibit sulci making an excellent analogy with the turtle shell. However, noteworthy differences exist between the shells of placodonts and those of turtles. In general, placodont thecae are comprised of a multitude of small bony elements that form a more or less regular mosaic of regular, hexagonal plates (Westphal, 1975; Rieppel, 2002). In contrast, the shell of turtles is generally comprised of a limited set of larger elements that somewhat correspond in their numbers to that of the underlying endochondral skeleton (Zangerl, 1969). A placodontlike pattern can be observed in representatives of Cenozoic PANDERMOCHELYIDS, but phylogenetic and stratigraphic considerations clearly demonstrated this to be homoplastic (e.g., Nick, 1912; Völker,

108 Walter G. Joyce and Hans-Volker Karl 1913, Joyce, In Press). Similarly, when present, the scutes of all known placodonts display a highly regular pattern of subequally sized elements that systematically correspond to the underlying bony pattern. Sulci are lightly developed and thus only visible in well-preserved material (Rieppel, 2002). Turtles, in contrast, always display a pattern of highly variably sized scutes and all basal turtle possess easily discernable, deep sulci. To achieve greater strength, the shells of all known basal turtles tend to fuse at maturity, in contrast to placodonts, which display distinct sutures regardless of size (Westphal, 1975; Rieppel, 2002). Finally, the scutes not the osteoderm of all placodonts are flat, whereas the scutes of primitive turtles are somewhat conical. SMNS 80141 lacks bony sutures, exhibits irregularly sized, conical scutes, and highly distinct sulci. Thus, our interpretation of SMNS 80141, and as such Priscochelys hegnabrunnensis, is that it is a basal turtle. This is consistent with the available data and is not contradicted by any available morphological evidence. Anatomical identity With a tentative taxonomic identity in mind, it is possible to speculate what portion of a turtle shell SMNS 80141 may represent. The plastron of all known turtles consists of a series of large, bilaterally symmetric scutes and can thus quickly be excluded from considerations. The carapace of the most basal known turtle, Proganochelys quenstedti, consists of scutes of different sizes that can be organized into four rows: a medial row of larger scutes consists of a nuchal, multiple vertebrals, and a pygal; a row consisting of intermediately sized laterals, a row consisting of numerous smaller sized supramarginals, and finally a lateral row of smaller sized marginals (Gaffney, 1990). More advanced turtles generally retain this pattern, although the row of supramarginals is quickly lost (Hutchison and Bramble, 1981). The scutes observable on SMNS 80141 display a pattern of at least one larger scute contacting a series of at least four smaller scutes. The original size of the final scute (field V of the description above) is unclear, so it may be interpreted as another large scute or another small scute. The possibility that the smaller scutes represent marginals is dismissed, because the hypothesized marginal scutes do not extend to the visceral surface, as is seen in all other known turtles. SMNS 80141 is thus interpreted as possessing at least one lateral scute that contacts a series of at least four supramarginal scutes. A similar arrangement of scutes is otherwise only known from Proganochelys quenstedti, although differences exist in the shape and size of the putative supramarginals. Karl s (2005) interpretation of SMNS 80141 as representing the most posterior lateral and supramarginals is primarily based on the curved suture that exists between the putative lateral and supramarginals, as seen P. quenstedti (Fig. 3). Referable Material The fossil fauna of the Germanic Muschelkalk facies is characterized by an interesting conundrum. Although significant quantities of excellent material has been collected over the course of the last 200 years, cyamodontoid placodonts are virtually known from skulls and teeth only. This contrasts the southern alpine facies, which is dominated by cyamodontoid armor (Rieppel, 2002). The only three known osteoderms, all from the Middle Triassic (lower Ladinian) of Crailsheim, Germany (Fig. 1), have recently been referred to the placodont Cyamodus kuhnschnyderi based on stratigraphic considerations (Nosotti and Pinna, 1996). However, Rieppel (2002) noted a resemblance of the Crailsheim osteoderms with the anterior carapacial region of the Upper Triassic turtle Proganochelys quenstedti, but eventually dismissed this possibility, because the Crailsheim osteoderms differ by being hollowed and thin shelled. Our analysis of SMNS 80141 shows no striking similarity between this specimen, the fourth known Germanic Muschelkalk osteoderm, and the Crailsheim material, primarily because the

The world s oldest fossil turtle 109 Fig. 3. Tentative interpretation of SMNS 80141, holotype of Priscochelys hegnabrunnensis, as the posterior shell region of a turtle, simplified from Karl (2005). The dark line indicates the preserved portions of SMNS 80141, double lines indicate known sulci, dashed double lines hypothesized sulci, dashed single lines the reconstructed edge of the shell, and dots reconstructed shell portions. Abbreviations: plv = pleural V; 9 through 13 = supramarginals 9 through 13; ma = marginal. latter appears to lack such distinct scute sulci. Although all four known osteoderms are from the lower Ladinian, we are thus reluctant to refer the Crailsheim to either Priscochelys hegnabrunnensis or TESTUDINATA as well. Phylogenetic relationships A rigorous phylogenetic hypothesis cannot be formulated for SMNS 80141. The most apparent problem is the highly fragmentary nature of the specimen. An additional significant problem that arises is the lack of suitable outgroup taxa, as discussed below. Based on the presence of a turtle shell consisting of dermal bone and scutes and the presence of a complete row of supramarginals, SMNS 80141 can be phylogenetically within the clade TESTUDINATA but must be considered basal to Proterochersis robusta. (Joyce, In Press). Only three other turtles are currently hypothesized to be more primitive than Proterochersis robusta: Proganochelys quenstedti from the Late Triassic of Germany, Palaeochersis talampayensis from the Late Triassic of Argentina, and Australochelys africanus from the Late Triassic/Early Jurassic of South Africa. At present, the latter two taxa are known from little or no carapacial material and therefore cannot be compared to SMNS 80141. The question thus remains, if SMNS 80141 is the sister taxon to 1) P. quenstedti, 2) all turtles, or 3) all turtles to the exclusion of P. quenstedti. SMNS 80141 and Proganochelys quenstedti are the only turtles presently known to possess a complete row of supramarginals, but it ultimately unclear if this is the symplesiomorphic condition (thus indicating a basal grade of turtles) or the synapomorphic condition (thus uniting Proganochelys quenstedti and Priscochelys hegnabrunnensis as a clade). If field V is interpreted as a supramarginal, then Priscochelys hegnabrunnensis differs from Proganochelys quenstedti by having five supramarginals contact a lateral scute, instead of just four. Given the general trend in chelonian evolution towards the reduction of scutes in general (Hutchison and Bramble, 1981) and of the supramarginals in particular (e.g., Joyce, In Press), the presence of an additional supramarginal scute may be taken as weak evidence in support of a basal position of Priscochelys hegnabrunnensis relative to Proganochelys quenstedti (Fig 4). Such a hypothesis is also generally consistent with its stratigraphically older age. Nevertheless, additional material of Priscochelys hegnabrunnensis will be necessary to rigorously test these hypotheses. Ecological Settings of Turtle Origins A recent review of basal turtle ecology revealed that the entire phylogenetic stem of turtles is optimized as having been terrestrial in its habitat preferences, and that the origin of turtles must have occurred on land and not in water (Joyce and Gauthier, 2004). Using theoretical reasons, Rieppel and Reisz (1999) previously argued that the origin of turtles must have occurred in water and cited the marine Muschelkalk provenance of SMNS 80141 as positive evidence in support of their hypothesis. Although the vast majority of

110 Walter G. Joyce and Hans-Volker Karl specimen is highly fragmentary, Priscochelys hegnabrunnensis is placed along the phylogenetic stem of turtles and speculated to be positioned basal to Proganochelys quenstedti relative to the crown. This assessment renders Priscochelys hegnabrunnensis the oldest and most basal unambiguous stem representative of the turtle clade. The habitat preferences of Priscochelys hegnabrunnensis are regarded as uncertain, because the type specimen is morphological uninformative and found in association with both marine and terrestrial faunas. Fig. 4. Tentative phylogenetic placement of Priscochelys hegnabrunnensis. Muschelkalk localities indeed contain marine faunas only, the locality of SMNS 80141 is unique in containing unambiguously terrestrial elements as well, particularly temnospondyls and lungfish (see Geological Settings above). Although it is impossible to confidently assess the ecology of Priscochelys hegnabrunnensis based on the morphology of a single fragment and the taphonomic settings in which it was found, it is important to note that this lack of affirmative knowledge regarding its habitat preferences neither supports nor refutes either hypothesis. Until more positive evidence is found, it is most consistent with the available data to conclude that the origin of turtles indeed took place on land (Joyce and Gauthier, 2004). CONCLUSIONS A detailed review of SMNS 80141, the holotype of the putative fossil turtle Priscochelys hegnabrunnensis, confirms its identity as a carapacial fragment of a fossil basal turtle. Although the type Acknowledgements. We thank the SMNS for access to this important specimen and to Igor Danilov and Jim Parham for the opportunity to place this contribution in the conference proceedings to the Symposium on Turtle Origins, Evolution, and Systematics. Randall Irmis, Michael Lee, and Olivier Rieppel provided useful comments that improved the quality of the manuscript. REFERENCES Carroll R. L. (1988), Vertebrate Paleontology and Evolution, Freeman and Company, New York. Joyce W. G. (In Press), «Phylogenetic Relationships of Basal Turtles,» Bull. Peabody Mus. Nat. Hist. Joyce W. G. and Gauthier J. A. (2004), «Palaeoecology of Triassic stem turtles sheds new light on turtle origins,» Proc. R. Soc. Lond. Ser. B, 271, 1-5. Joyce W. G., Parham J. F. and Gauthier J. A. (2004), «Developing a protocol for the conversion of rank-based taxon names to phylogenetically defined clade names, as exemplified by turtles,» J. Palaeontol., 78, 989-1013. Gaffney E. S. (1990), «The comparative osteology of the Triassic turtle Proganochelys,» Bull. Amer. Mus. Nat. Hist., 194, 1-263. Hutchison J. H. and Bramble D. M. (1981), «Homology of the plastral scales of the Kinosternidae and related turtles.» Herpetologica, 37, 73-85. Karl H.-V. (2005), «The homology of supramarginals in turtles (Reptilia: Chelonii),» Stud. Geol. Salmanticensia, 41, 63-75.

The world s oldest fossil turtle 111 Lucas S. G. Heckert A. B. and Hunt A. P. (2000), «Probable turtle from the Upper Triassic of eastcentral New Mexico,» N. Jb. Geol. Paläont., Mh., 2000, 287-300. Młynarski M. (1976). Handbuch der Paläoherpetologie, Part 7, Testudines, Gustav Fischer Verlag, Stuttgart. Nick L. (1912), «Das Kopfskelet von Dermochelys coriacea Linneaus,» Zool. Jb. Abt. Anat. Ont. Tiere, 33, 1-238. Nosotti S. and Pinna G. (1996), «Osteology of the skull of Cyamodus kuhnschnyderi Nosotti and Pinna 1993 (Reptilia, Placodontia),» Pal. Lombarda, NS 6, 1-42. Rieppel O. (2002) «The dermal armor of the cyamodontoid placodonts (Reptilia, Sauropterygia): Morphology and systematic value,» Fieldiana Geol., NS, 46, 1-41. Rieppel O. and Reisz R. R. (1999) «The origin and early evolution of turtles,» Annu. Rev. Ecol. Syst., 30, 1-22. Völker H. (1913), «Über das Stamm-, Gliedmaßen- und Hautskelet von Dermochelys coriacea L.,» Zool. Jb. Abt. Anat. Ont. Tiere, 33, 431-552. Westphal F. (1975), «Bauprinzipien im Panzer der Placodonten (Reptilia triadica),» Palaeontol. Zeitsch., 49, 97-125. Wild R. (1972), «Die Wirbeltierfaunen der fränkischen und südalpinen Mitteltrias,» Z. Deutsch. Geol. Ges., 123, 229-234. Zangerl R. (1969), «The turtle shell,» in: Gans C., Bellairs A. and Parsons T. S. (eds.), Biology of the Reptilia, Volume 1. Academic Press, London, pp. 311-339.