AMERICAN MUSEUM Novitates PUBLISHED BY THE AMERICAN MUSEUM OF NATURAL HISTORY CENTRAL PARK WEST AT 79TH STREET, NEW YORK, NY 10024 Number 3372, 16 pp., 5 figures, 2 tables June 21, 2002 Redescription of the Skull of Dacquemys Williams, 1954, a Podocnemidid Side-Necked Turtle from the Late Eocene of Egypt EUGENE S. GAFFNEY,’ DONALD D. DEBLIEUX,”? ELWYN L. SIMONS,? MARCELO R. SANCHEZ-VILLAGRA,* AND PETER A. MEYLAN> ABSTRACT The known material of the side-necked turtle Dacquemys paleomorpha Williams, 1954, consists of the type skull and a new skull from the late Eocene of Egypt. Dacquemys is reaffirmed as a member of the Podocnemididae because of its well-developed cavum ptery- goideus. Within the Podocnemididae Dacquemys uniquely possesses a fully roofed temporal region produced by a posteriorly extensive parietal and wide supraoccipital, a very wide in- terorbital area, and two accessory maxillary ridges meeting anteriorly to form an enclosed trough. INTRODUCTION species Stereogenys libyca Andrews, 1906. In 1954, Williams recognized that this attri- The type skull of the taxon redescribed — bution was inconsistent with the type skull here was first described by E. Dacqué in material of Stereogenys cromeri Andrews, 1912 and referred by him, doubtfully and for 1901, the type of the genus, and that the reasons that are unclear, to the shell-based Dacqué skull was distinct from all other ' Curator, Division of Paleontology, American Museum of Natural History. e-mail: [email protected] ? Duke University Primate Center, Durham, NC 27705. Present address: Utah Geological Survey, Salt Lake City, UT 84114. 3 James B. Duke Professor of Anthropology, Department of Biological Anthropology and Anatomy, and Primate Center, Duke University, Durham, NC 27705. 4 Universitat Tiibingen, Zoologisches Institut, D-72076, Germany. e-mail: marcelo.@s unai-ntucebhingeenz.d e > Eckerd College, St. Petersburg, FL 33733. e-mail: [email protected] Copyright © American Museum of Natural History 2002 ISSN 0003-0082 2: AMERICAN MUSEUM NOVITATES NO. 3372 known Pleurodira. Williams named the form the classic sense). A useful compilation of Dacquemys paleomorpha and provided a the pelomedusoid literature is found in Broin brief description. Since 1954 there have been (1988) and in Lapparent de Broin (2000). no further contributions to the understanding The only computer-generated cladogram of of this genus. It is the purpose of this paper Podocnemididae available at present is Mey- to describe the type skull, following repre- lan (1996), which shows the relationships of paration and cleaning, and to describe a new these families and the included genera. skull discovered by the Duke University Pri- mate Lab during ongoing fieldwork in Egypt. Institutional Abbreviations Despite many field seasons of paleontologi- cal work in Egypt by a number of institu- AMNH~ American Museum of Natural History tions, only these two skulls of Dacquemys DPC Duke Primate Center, Duke University, have come to light. The new skull preserves North Carolina SMNS Staatliches Museum fiir Naturkunde, a number of areas unavailable in the type, Stuttgart and together they allow a complete recon- struction of the skull. This paper limits de- Anatomical Abbreviations scription to comparisons with the well- known forms, the living podocnemidids Po- bo basioccipital docnemis, Erymnochelys, and Peltocephalus. bs basisphenoid It is worth examining the basis for ex exoccipital Dacqué’s original assignment of the type fr frontal skull to Stereogenys libyca, primarily to see ju jugal why he associated a shell with the skull. In mx maxilla Dacqué (1912: 337) we translate his argu- op opisthotic ment as follows: ‘““The reasons why we as- pa parietal pal palatine sign the skull to Stereogenys libyca Andrews pf prefrontal are the following: 1) because it is particularly pm premaxilla similar to Podocnemis but it cannot belong po postorbital to this species, 2) because [the skull] is rather pr prootic convex so it was not retractable, a condition pt pterygoid which is consistent with the epiplastral tu- qj quadratojugal bercles [gular projections] of Stereogenys li- qu quadrate byca, 3) it is consistent with the size of the SO supraoccipital shell [of Stereogenys libyca], 4) [both shell sq squamosal and skull] are found in the same stratigraphic level’’. It does not seem that any shells were SYSTEMATICS associated in site with the original skull. He was probably wrong about gular projections ORDER TESTUDINES LINNAEUS, 1758 preventing head retraction; living tortoises MEGAORDER PLEURODIRA COPE, 1864 (FIDE have projections and retract with no problem. GAFFNEY AND MEYLAN, 1988) Therefore, as Williams (1954) implied, there is no shell associated with the type skull of HYPERFAMILY PELOMEDUSOIDES COPE, 1868a Dacquemys paleomorpha. FAMILY PODOCNEMIDIDAE We use the terminology for pleurodires COPE, 1868b previously developed in Broin (1988), An- Dacquemys Williams, 1954 tunes and Broin (1988), and more recently in Meylan (1996), Lapparent de Broin and Wer- TYPE SPECIES: Dacquemys paleomorpha ner (1998), Tong et al. (1998), and Gaffney Williams, 1954. et al. (200la, 2001b, 2001c). This usage DISTRIBUTION: Late Eocene of Egypt. places the families Bothremydidae, Podoc- DIAGNOSIS: A genus of podocnemidid nemididae, and Pelomedusidae (restricted to pleurodire known only from the skull; char- Pelusios and Pelomedusa) in the Pelomedu- acterized by the unique possession of a fully soides (which equals the Pelomedusidae in roofed temporal region with a posteriorly ex- 2002 GAFFNEY ET AL.: DACQUEMYS 3 tensive parietal and wide supraoccipital com- the Staatliches Musuem ftir Naturkunde, pletely covering the otic chamber in dorsal Stuttgart, was presumably part of a collection view; broad parietal-squamosal contact not obtained from German geologist and private seen in other podocnemidids; prefrontal ex- collector Richard Markgraf who lived in tends to anterior margin of apertura narium Egypt and made extensive collections in the externa as in Peltocephalus but in contrast to Fayum region during the early 20th century. Podocnemis, interorbital distance very wide During this early period of collecting in the with orbits facing laterally in contrast to liv- Fayum, locality information was typically ing Podocnemididae; maxillae meet broadly vague and lacked attention to precise strati- behind premaxillae; premaxillae recessed graphic level (Simons, 1968; Simons and and not visible laterally in contrast to all oth- Rasmussen, 1990). The locality given in er Pelomedusoides; vomer absent; anterior Dacqué (1912) first in reference to the shell part of triturating surface more extensive and then the skull is “‘Unteroligocane Flu- than in living Podocnemididae but secondary viomarinestuffe n6érdlich der Birket Qarun; palate as seen in Shweboemys group absent; nordlich non Tamieh; ostlich vom Schwein- two accessory maxillary ridges meet anteri- forthplateau. Schadal: aus derselben stufe bei orly to form enclosed trough unique to this Diméh. Alles Fayum”’ (Dacqué 1912: 310), genus; antrum postoticum small as in Podoc- which we translate as “‘Early Oligocene flu- nemis; precolumellar fossa shallow as in Po- viomarine sediment north of Birket Qarun; docnemis expansa; pterygoid-jugal contact north of Tamieh; east of the Schweinfurth absent. Plateau. Skull from the same level near Di- méh. All specimens being from the Fayum Dacquemys paleomorpha Williams, 1954 area’’. The beginning of this entry is similar to that typically given for fossils found in the TYPE SPECIMEN: SMNS 12645, a nearly Jebel Qatrani Formation during this period, complete skull without lower jaws (figs. 2 which are “‘north of Birket Qarun”’ or ‘“‘flu- and 3), lacking parts of the left side and the viomarine series north of Birket Qarun”’ (Si- central basicranial region (Dacqué, 1912: pl. mons, 1968). Birket Qarun is the name of the II, figs. 6-8; Gaffney, 1979: fig. 128; Wil- large lake in the north of the Fayum depres- liams, 1954: pl. 1). Measurements in table 1. sion. The reference to early Oligocene is also TYPE LOCALITY AND HORIZON: Dacqué consistent with the Jebel Qatrani Formation, (1912) gave the locality at first in reference which was thought to be late Eocene by the to a shell: “‘Unteroligocéne Fluviomarines- earliest workers (1.e., Beadnell, 1905; An- tuffe nordlich der Birket Qerun; nordlich non drews, 1906) but by the time of Dacqué’s Tamieh; ostlich vom Schweinforthplateau. study was interpreted to be early Oligocene Schadel: aus derselben stufe bei Diméh. Al- (Stromer, 1907; Osborn, 1908), with the un- les Fayum”’ (Dacqué, 1912: 310). Williams derlying Qasr el Sagha Formation considered (1954; repeated in Gaffney, 1979) has the lo- to be late Eocene. The reference to the skull cality as “‘Diieh’’, which seems to be a mis- ‘“‘from the same sediments near Diméh”’ is spelling of ““Diméh’’. The reference to the more difficult to interpret. Diméh refers to skull “‘from the same sediment near Diméh’’ the ruins of a Greco-Roman temple and set- presumably means the Qasr el Sagha or Bir- tlement located on the north shore of Birket ket Qarun formations of late Eocene age. Qarun. It is possible that this reference in- REFERRED MATERIAL: DPC 5986, partial dicates information given by the collector to skull lacking anterior part of palate and some more accurately place the locality of this of left side (figs. 4 and 5). Field no. 86—292, specimen, or it may be a general reference to collected by Alex van Nievelt. Measure- the region north of Birket Qarun. The sedi- ments in table 1. ments surrounding Diméh include those of LOCALITY AND HoRIZON: B-4 Uower Jebel the Umm Rigl Member of the Qasr el Sagha Qatrani, %4 mile southeast of A). AMNH Formation (Gingerich, 1992) and of the un- quarry B area, Eocene (Gagnon, 1997: fig. 1; derlying Birket Qarun Formation, both cov- Kappelman et al., 1992). ered in part by Pleistocene and Holocene DISCUSSION: The type specimen, housed in lake deposits from the once more extensive AMERICAN MUSEUM NOVITATES NO. 3372 B Fig. 1. Dacquemys paleomorpha, late Eocene, Fayum, Egypt, restorations based on SMNS 12645 and DPC 5986. A, Dorsal; B, ventral; C, lateral. Dotted lines indicate scale margins. Birket Qarun (Bown and Kraus, 1988). The “east of Schweinfurth Plateau’’ may indicate age and nature of the contact between these these lower sediments. We are not aware of formations in the area around Diméh are the location of this plateau but the Qasr el presently being investigated, but both marine Sagha temple was known as Schweinfurth’s and continental vertebrate remains are found Temple, so it may be that his “‘plateau’’ is in these sediments. This, along with the fau- located near the temple. Schweinfurth is nal continuity that is a general feature of the known to have worked primarily in the area Fayum sequence (Rasmussen et al., 1992), around Birket Qarun. At present, the Qasr el makes it entirely possible that the type spec- Sagha Formation is generally thought to be imen of Dacquemys was collected from ei- late Eocene (Priabonian) in age (Bown and ther the Qasr el Sagha or Birket Qarun for- Kraus, 1988; Gingerich, 1992) with the pos- mations. It is possible that the reference to sibility that the lower part of the Qasr el Sag- 2002 GAFFNEY ET AL.: DACQUEMYS 5 TABLE 1 many of the specimens collected by the Measurements of Dacquemys Skulls American Museum of Natural History, with the help of Richard Markgraf, were obtained SMNS DPC (Bown and Kraus, 1988). These sediments 12645 5986 represent the second of four faunal zones that are recognized in the Jebel Qatrani Forma- A. Midline length as preserved 73.28 79.48 tion (Rasmussen and Simons, 1991; Gagnon, B. Maximum width 67.3 76.84 1997). Quarry B is thought to lie just below C. Width between orbits 28.8 34.0 the EoceneOligocene boundary (Kappleman D1. Width of left orbit 14.2 — et al., 1992; Rasmussen et al., 1992; Simons and Chatrath, 1998), although the lack of D2. Width of right orbit 14,2» 16.0 substantial faunal turnover across this bound- E. Width of external nares 19.0 — ary in the Fayum region led to some debate F. Width of internal nares 13.7 — as to the exact placement of this boundary G. Maximum height at quadrate 38.2 — (Gingerich, 1992; Van Couvering and Harris, BOS): H. Maximum width of skull 33.7 34.0» taken at middle of orbits DESCRIPTION Length, anterior margin of prefrontals 98.8 100.1 — to posterior margin of parietals PREFRONTAL Ji. Height of left orbit 13.5 — Both prefrontals are preserved on both J2. Height of right orbit 12.3 — sides of SMNS 12645 and DPC 5986, al- K. Skull height at occipital condyle 33.5 27.3 though its ventral process is complete only L. Anterior width of triturating surface 15.2 — on the left side of SMNS 12645. The pre- frontal of Dacquemys is similar to that in Po- M. Posterior width of triturating surface 14.9 — docnemis except for its much larger size, be- N . Width of palate across 23.96 29.9 ing longer and wider. Erymnochelys and Pel- foramina palatinum posterius tocephalus have wider prefrontals than does O. Length, front of skull to posterior 53.6 — Podocnemis, but not as wide as in Dacque- edge of condylus articularis mys. The anterior projection of the prefrontal aEstimated. extends to the anterior margin of the apertura bDamaged. narium externa, not posterior to it as in Po- docnemis. ha along with the Birket Qarun Formation FRONTAL are close to the middle Eocenelate Eocene (Bartonian-Priabonian) boundary (Holroyd et Both frontals are preserved in SMNS al., 1996). Because the only other known 12645 and DPC 5986. Their ventral surface specimen of Dacquemys was collected from is visible in DPC 5986. The frontal in Dac- the lower sequence of the Jebel Qatrani For- quemys has the same contacts as in the living mation, and it is difficult to know how much podocnemidids, but it differs in its greater credence should be given to the reference to width. The orbits are visible in dorsal view Diméh, the assignment of this specimen to in Podocnemis, Erymnochelys, and Peltoce- one of these earlier formations must remain Phalus but not in Dacquemys. In ventral tentative until further specimens are recov- view, Dacquemys has a much more extensive ered. roof to the fossa orbitalis than in the living The second specimen of Dacquemys (DPC genera. The sulcus olfactorius and fossa na- 5986) was collected by a Duke University salis are about the same size in the living expedition from AMNH Quarry B (B-4). genera and Dacquemys, but the roof of the Quarry B is located in the middle gravelly fossa orbitalis is more extensive in Dacque- sandstone unit of the lower sequence of the mys, due mostly to the lateral extension of Jebel Qatrani Formation (the lower fossil the frontal, although both the prefrontal and wood zone of early authors) from which postorbital contribute to this. 6 AMERICAN MUSEUM NOVITATES NO. 3372 Fig. 2. Dacquemys paleomorpha, SMNS 12645, holotype, probably Jebel Qatrani Formation, Fa- yum, Egypt. A, Dorsal; B, ventral; C, right lateral; D, anterior; E, left lateral; F, posterior. See figure 3 for key. 2002 GAFFNEY ET AL.: DACQUEMYS 7 Fig. 3. Dacquemys paleomorpha, SMNS 12645. Key to figure 2. Dotted lines indicate scale margins, shaded areas are plaster, and hatched areas are broken edges. PARIETAL contacts the frontal anteriorly, the postorbital anterolaterally, and the quadratojugal later- Most of the right parietal is preserved in SMNS 12645, and all of the right and most ally. In Podocnemis there is a jugal-parietal of the left are present in DPC 5986. As in contact that is absent in Peltocephalus and the living genera, the parietal in Dacquemys Erymnochelys, as well as in Dacquemys. In 8 AMERICAN MUSEUM NOVITATES NOs3372 Fig. 4. Dacquemys paleomorpha, DPC 5986, Jebel Qatrani Formation, Fayum, Egypt. A, Dorsal; B, ventral; C, right lateral. See figure 5 for key. Dacquemys, however, there is a broad pari- Dhalus, which are more roofed than is Po- etal-squamosal contact not seen in other po- docnemis. Dacquemys has a posterolaterally docnemidids. The parietal extends posteri- expanded parietal which results in contact to orly to a greater extent than in any other de- the squamosal. scribed podocnemidid, completely covering The processus inferior parietalis is not well the otic chamber. Posteromedially the parie- preserved in either Dacquemys specimen, but tal contacts an expanded supraoccipital when it is best seen on the right side of DPC 5986. compared with Erymnochelys and Peltoce- The processus contacts the supraoccipital pos- 2002 GAFFNEY ET AL.: DACQUEMYS 9 Ries, Dacquemys paleomorpha, DPC 5986. Key to figure 4. Dotted lines indicate scale margins, hatched areas are broken edges. teriorly and the prootic posteroventrally, and the jugal enters the orbit anteriorly, contacts probably the pterygoid ventrally, all as in the the maxilla anteroventrally, the postorbital recent podocnemidids. The more anterior con- anterodorsally, and the quadratojugal poste- tacts and the formation of the foramen nervi riorly. Ventrally the jugal forms the anterior trigemini are not preserved. part of the shallow cheek emargination. The degree of cheek emargination in Dacquemys JUGAL is far greater than in Erymnochelys and Pel- The jugal is preserved in part in both tocephalus, but less than in Podocnemis and skulls, but neither shows the internal rela- Hamadachelys. It is most comparable to tions of the bone clearly. The lateral plate of Neochelys and Bauruemys. 10 AMERICAN MUSEUM NOVITATES NOs3372 The jugal in Dacquemys is small in com- tacts seen in most other podocnemidids, such parison to the much larger jugal of Erym- as Neochelys. The postorbital enters the pos- nochelys and Peltocephalus, in which the ju- terodorsal margin of the orbit as in the other gal extends posteriorly to meet the quadrate, podocnemidids. The medial process of the and the cheek emargination is quite reduced. postorbital is visible in DPC 5986 and con- The medial process of the jugal is present in tacts the frontal dorsomedially, the jugal ven- SMNS 12645, but its contacts are not clearly trolaterally, and the palatine ventromedially. visible. As in the living podocnemidids, the jugal does not appear to form any of the trit- PREMAXILLA urating surfaces. Both premaxillae are preserved in SMNS 12645, but neither is present in DPC 5986. QUADRATOJUGAL The premaxilla of Dacquemys contacts the A nearly complete right quadratojugal is maxilla laterally and the other premaxilla preserved in DPC 5986 and a partial one is medially. There is no vomer contact, as that present on the right side of SMNS 12645. In bone is absent and the maxillae meet on the midline behind the premaxillae as in Podoc- Dacquemys, the quadratojugal contacts the parietal medially, the squamosal posteriorly, nemis. Dacquemys is unique among Pelo- the quadrate laterally, the postorbital antero- medusoides because the premaxillae are re- dorsally, and the jugal anteroventrally. The cessed and not visible laterally. There is a quadratojugal forms the dorsalmost part of shallow, anteroventral trough on the anterior the margin of the cheek emargination. surface of each premaxilla, another feature found only in Dacquemys. As far as can be determined, the foramen praepalatinum is ab- SQUAMOSAL sent. The midline concavity on the triturating The right squamosal is nearly complete in surface in Dacquemys is very deep, deeper DPC 5986, and in SMNS 12645 the bone is than in Peltocephalus, and much deeper than complete on the right side and partial on the in Podocnemis and Erymnochelys. left. In Dacquemys the squamosal has a long medial contact with the parietal that is absent MAXILLA in all other described podocnemidids, and an Both maxillae are preserved in SMNS anterior contact with the quadratojugal that 12645, and both are absent in DPC 5986. is comparable in size to the contact seen in The vertical plate of the maxilla in Dac- Peltocephalus and wider than in Podocnem- quemys is shallower than in Podocnemis and is. Medially the squamosal contacts the op- Bauruemys and is comparable to Peltoce- isthotic and anteromedially the quadrate, as in other podocnemidids. The squamosal Phalus and Erymnochelys. The contacts are forms part of the antrum postoticum (see as in the living podocnemidids. The horizon- tal plate of the maxilla in Dacquemys has the Quadrate). The posterior margin of the squa- mosal in podocnemidids is a flange oriented same contacts as in Podocnemis expansa: premaxilla anteromedially, other maxilla on dorsolaterally to ventromedially and reaching midline, palatine posteromedially, and jugal the opisthotic. In most podocnemidids this flange is a continuous sheet, but in SMNS posterolaterally. The midline maxilla contact is more extensive than in Podocnemis expan- 12645 (it is absent in DPC 5986) the sheet, as seen on the right side, curves anteriorly sa and Podocnemis unifilis, the only other Pelomedusoides that have this character. before reaching the opisthotic. The triturating surface of Dacquemys is characterized by two well-developed acces- POSTORBITAL sory ridges parallel to the lingual and labial The postorbital is present but incomplete ridges. Similar accessory ridges occur in on both sides of SMNS 12645, and is com- some species of Podocnemis (expansa, uni- plete on the right side of DPC 5986. The filis, lewyana, vogli), but only in Dacquemys postorbital of Dacquemys is not reduced as are they curved anteriorly to join each other in Podocnemis and has the shape and con- close to the premaxilla-maxilla border.