AMERICAN MUSEUM Norntates PUBLISHED by the AMERICAN MUSEUM of NATURAL HISTORY CENTRAL PARK WEST at 79TH STREET, NEW YORK, N.Y

Home , Australochelys, Kayentachelys, Proganochelys

AMERICAN MUSEUM Norntates PUBLISHED BY THE AMERICAN MUSEUM OF NATURAL HISTORY CENTRAL PARK WEST AT 79TH STREET, NEW YORK, N.Y. 10024 Number 3130, 29 pp., 25 figures, 2 tables May 24, 1995

The Morphology and Relationships of Australochelys, an Early Jurassic Turtle from South Africa

EUGENE S. GAFFNEY,1 AND JAMES W. KITCHING2

ABSTRACT Australochelys from the Early Jurassic Elliot (Cryptodira plus Pleurodira) than to Proganoche- Formation of South Africa is the oldest African lys. These characters include a sutured basipter- turtle. Known only from the skull and a shell frag- ygoid articulation and middle ear region partially ment, Australochelys has many primitive chelo- enclosed laterally. Australochelys is hypothesized nian characters in common with the Late Triassic as the sister taxon to the Casichelydia; together Proganochelys, such as a large interpterygoid va- they form the Rhaptochelydia. The relationships cuity and a lacrimal foramen. Australochelys has of Australochelys show that the beginning stages large orbits and a ventral basioccipital tubercle, of the unusual turtle hearing mechanism evolved characters that only occur in Proganochelys. How- before the origin ofthe modem turtle groups. Ak- ever, because Australochelys has a series of ad- inesis preceded or accompanied the enclosure of vanced characters in common with the Casiche- the hypertrophied middle ear region in the early lydia, it is more closely related to the Casichelydia evolution of the hearing mechanism.

INTRODUCTION Australochelys, the oldest turtle from Af- described in a short paper (Gaffney and rica, reveals an important stage in the early Kitching, 1994) that did not include a de- history ofturtles. Australochelys was recently tailed morphological description or extended

1 Curator, Department of Vertebrate Paleontology, American Museum of Natural History. 2 Bernard Price Institute for Palaeontological Research, University ofthe Witwatersrand, Johannesburg 2001, South Africa.

Fig. 1. Australochelys africanus Gaffney and Kitching, 1994, BP/1/4933. Dorsal view of skull. Ster- eophotograph. See fig. 3A for key. character discussion. The purpose ofthis pa- terpreted here, is that it is advanced over per is to provide more descriptive informa- Proganochelys but lacks derived characters tion and character discussions. of either Cryptodira or Pleurodira. In addi- The significance of Australochelys, as in- tion to Gaffney and Kitching (1994), relevant

Fig. 2. Australochelys africanus Gaffney and Kitching, 1994, BP/1/4933. Ventral view of the skull. Stereophotograph. See fig. 3B for key. 1995 GAFFNEY AND KITCHING: MORPHOLOGY OF AUSTRALOCHELYS 3

B labial ridge apertura narium interna lingual ridge

interpterygoid vacuity

basipterygoid process Fig. 3. Australochelys africanus Gaffney and Kitching, 1994, BP/1/4933. A. Dorsal view of skull. B. Ventral view of skull. background papers are: Gaffney (1990) for AcKNowLEDGMENTs Proganochelys, Gaffney et al. (1987) for the We are most grateful to Mr. I. van Rooyen earliest cryptodire, Kayentachelys, and Gaff- of the farm Bormansdrift for his kindness, ney and Meylan (1988) for turtle phylogeny hospitality, and for permitting one ofus (J.K.) in general. The following skull description of access to his land to collect fossils. Dr. M. A. Australochelys assumes some familiarity with Raath helped with information on the Gokwe the skull of Proganochelys (Gaffney, 1990) Formation. Dr. B. S. Rubidge, director ofthe and general turtle skull morphology as pre- Bernard Price Institute, very graciously sented in Gaffney (1979). The descriptive opened the facilities and collections of that section here follows the order in Gaffney institution to one ofus (E.G.). The high qual- (1979). ity of the figures is due to the skills of F. Before the discovery ofAustralochelys, the Ippolito, E. Heck (fig. 25), and C. Tarka (fig. oldest known African turtle was the frag- 19). The senior author is particularly grateful mentary specimen identified by Nesbitt and to A. Schermerhorn for improving the manu- Bond (1972) as a turtle from the Gokwe For- script and assisting in its preparation. mation of Zimbabwe. At that time it was on the goodwill and questionably dated as Late Jurassic. More re- Everyone is dependent cently these and other turtle specimens from generosity of colleagues who willingly give the Gokwe Formation of Zimbabwe have their time in reviewing papers. We are par- been reinterpreted as being more likely Late ticularly fortunate to have benefited from the Cretaceous in age (M. Raath, personal com- efforts offour ofour associates: D. Brinkman, mun. to J. Kitching, August 1993). At pres- P. Meylan, R. Reisz, and 0. Rieppel, who ent, the next oldest known African turtles are have generously helped improve this paper. the Early Cretaceous pleurodires summa- rized in de Broin (1989). This represents a gap of some 60 million years, between Aus- ABBREVILATIONS tralochelys and the Early Cretaceous pleu- BP Bernard Price Institute, Johannes- rodires. burg, South Africa 4 AMERICAN MUSEUM NOVITATES NO. 3130

MB Museum fur Naturkunde, Berlin, is defined to consist of Australochelys plus Germany Casichelydia. SMNS Staatliches Museum fur Naturkunde, Stuttgart, Germany FAMILY AUSTRALOCHELYIDAE Gaffhey and Kitching, 1994 bo basioccipital bs basisphenoid Australochelys Gaffney and Kitching, 1994 ex exoccipital fr frontal TYPE SPECIES: Australochelys africanus ju jugal Gaffhey and Kitching, 1994. mx maxilla KNowN DISRmurIoBN: Early Jurassic, one na nasal locality in Orange Free State, South Africa. op opisthotic ETYMOLOGY: australos, Greek for "south"; pa parietal chelys, Greek for "turtle." pal palatine DIAGNOsIs: As for species. pf prefrontal pm premaxilla Australochelys africanus po postorbital Gaffhey and Kitching, 1994 pr prootic TYPE SPECiMEN: Bernard Price Institute, pt pterygoid University of the Witwatersrand, Johannes- qj quadratojugal burg, South Africa, BP/l/4933; consists of a qu quadrate skull without lowerjaws and a fragment from so supraoccipital the bridge area of the shell. sq squamosal LocALITr: Bormansdrift (133), Clocolan vo vomer District, Orange Free State, South Africa. Map sheet: 1:50,000 2827 CD Mekoatlengsnek 28057'33"S, 27026'05"E (Kitching and Raath, SYSTEMATICS 1984). ORDER TESTUDINES HORIZON: Tritylodon Acme-Zone, (middle) Elliott Formation, Early Jurassic (Olsen DLAGNosIs: As in Gaffney and Meylan and Galton, 1984). (1988). COLLECTOR: James W. Kitching, March 1980. Rhaptochelydia Gaffney and Kitching, 1994 ETYMOLOGY: africanus, from Africa. ETYMOLOGY: rhaptos (not "rhaptors" Gaff- ney and Kitching, 1994, in error), Greek for "sewn" in allusion to the sutured basiptery- DIAGNOSIS goid articulation; chelys, Greek for "turtle." Kitching, 1994) EMENDED DiAGNosIs: Posterior margin of (modified from Gaffney and temporal roof extends over opisthotic; pal- Characters unique to Australochelys among atal teeth few or absent; basipterygoid artic- turtles: Orbit larger in relative size than in ulation closed (sutured); cranioquadrate space any other turtle; lacrimal foramen at least a well-defined foramen or canal, the canalis three times larger than in Proganochelys; ex- cavernosus, not an open space; stapes prob- ternal nares elongate in contrast to all turtles; ably articulating with a tympanic membrane, vomers arched dorsally, narrow posteriorly, not with the quadrate; ventral ridge on op- and very broad anteriorly in a unique con- isthotic; distal end of opisthotic covered lat- figuration. erally by squamosal and/or quadrate; par- Primitive chelonian characters found only occipital process ofopisthotic tightly sutured in Proganochelys and Australochelys within to braincase; crista supraoccipitalis a distinct turtles but also found in primitive tetrapods: vertical sheet of bone. Divided external nares formed by premax- DIsCussIoN: This taxon, Rhaptochelydia, illae; large interpterygoid vacuity as in Pro- 1995 GAFFNEY AND KITCHING: MORPHOLOGY OF A USTRALOCHELYS 5

ganochelys; unenclosed middle ear region; a DERMAL ROOFING ELEMENTS recessed, funnel-shaped, cavum tympani ab- (figs. 1, 3-6) sent; lacrimal foramen present; recessus scalae tympani and fenestra perilymphatica ab- The nasal area of Australochelys is char- sent; foramen jugulare posterius with at least acterized by a deep snout, large fossa nasalis, a medial edge absent; cultriform process pres- and an elongate apertura narium externa. In ent. contrast, Proganochelys has a relatively small Advanced chelonian characters found only snout, small fossa nasalis, and a less elongate in Australochelys and Casichelydia (Cryp- apertura narium externa. These features in todires plus Pleurodires): Basipterygoid artic- Australochelys are probably autapomorphic, ulation sutured; stapes probably does not ar- although a large fossa nasalis is common in ticulate directly with quadrate but may have other amniotes and does occur in some Cas- attached to a tympanic membrane supported ichelydia. Australochelys has a paired aper- by the acute posterior edge of the quadrate; tura narium externa, with a well-developed distal end ofopisthotic covered by quadrate; internarial bar, as in Proganochelys. The in- canalis cavernosus a well-defined canal; few ternarial bar is broad dorsally, becomes nar- or no palatal teeth; crista supraoccipitalis a rower ventrally, but its ventral limits are not distinct vertical sheet ofbone; temporal roof preserved. The dorsal end ofthe bar is pushed extends posteriorly over opisthotic. under the skull roof along what could be a suture. If this is a suture, it is likely that the internarial bar is formed by the premaxillae, DESCRIPTION as in Proganochelys. It is also possible that a crack on the midline of the internarial bar is PRESERVATION a median suture (fig. 8). The type specimen of Australochelys afri- Each apertura narium externa is roughly canus is preserved in a highly indurated, he- oval, elongate dorsoventrally. The lower lim- matite-cemented, fine-grained sandstone. its are not complete on either side, but the Postmortem cracking and hematite infiltra- general extent of the apertura can be esti- tion have caused minute reorientation ofbone mated by extending the ventral edge of the trabeculae and loss offine structural detail in skull anteriorly. The right side has the better most areas. There has been lateral defor- preserved apertura narium externa. mation ofthe skull in which the anterior part Other than a possible premaxilla-nasal has been pushed to the right, in dorsal view, contact on the midline, there are no visible with respect to the posterior part. The skull sutures delimiting the nasal bone in Austral- roof is the most fractured; the palate and ochelys. The anterior part of the skull roof, braincase retain finer detail. The postmortem formed by the nasal in Proganochelys, is more hematite infiltration has destroyed many ar- convex and curved in Australochelys than in eas with potential sutures. However, it is like- Proganochelys. This is related to the larger ly that the general absence of sutures in Aus- size of the fossa nasalis in Australochelys. tralochelys is not only the result of preser- Proganochelys differs from all Casichelydia vational changes, but also because there are in having a lacrimal bone and a lacrimal fo- a number ofplaces that seem sufficiently pre- ramen. Australochelys also has a lacrimal fo- served to reveal sutures ifthey were present. ramen and may have a lacrimal bone as well. The type skull ofAustralochelys probably had On the right side of Australochelys (fig. 5) most of its sutures fused during life. there are cracks in about the same places as Because most sutures are not visible in those in the lacrimal sutures in Proganoche- Australochelys, we have described skull areas lys. However, none ofthese can be confirmed rather than bones. However, for the sake of as sutures when examined microscopically. practicality in the description, certain areas The lacrimal foramen is oval and unusu- in Australochelys are assumed to consist of ally large in Australochelys, about three times the same bones forming those areas in Pro- larger than in Proganochelys. The internal ganochelys and other turtles. fossa nasalis in Australochelys is filled with 6 AMERICAN MUSEUM NOVITATES NO. 3130

Fig. 4. Australochelys africanus Gaffney and Kitching, 1994, BP/1/4933. Left lateral view of skull. Stereophotograph. See fig. 6A for key. matrix, so the presence or absence of a lac- orbital margin is another opening that also rimal groove cannot be determined. seems to be due to breakage. In the right orbit ofAustralochelys there is The frontal and parietal bones (figs. 1, 3) a small hole dorsomedial to the lacrimal fo- presumably make up the central area of the ramen (figs. 5, 6). This is possibly an addi- skull roof in Australochelys, but no sutures tional lacrimal foramen, as seen in Progan- are visible, not even midline sutures. The ochelys SMNS 15759 (Gaffney, 1990). Be- area of the presumed frontal in Australoche- cause ofits irregular shape and broken edges, lys, by comparison with Proganochelys, has we interpret it as an artifact of preservation. a low convexity on the midline, an area that On the left side (figs. 4-6) anterior to the is flat in Proganochelys. There is also a small-

Fig. 5. Australochelys africanus Gaffney and Kitching, 1994, BP/1/4933. Right lateral view of the skull. Stereophotograph. See fig. 6B for key. 1995 GAFFNEY AND KITCHING: MORPHOLOGY OF A USTRALOCHELYS 7

A B

sq * u *

acute edge of quadrate lacrimal foramen fenestra ovalis lacrimal foramen Fig. 6. Australochelys africanus Gaffney and Kitching, 1994, BP/1/4933. A. Left lateral view ofskull. B. Right lateral view of skull. The structures marked with an asterisk (*) are interpreted as artifacts, not premortem foramina.

er convexity in the region of the presumed Proganochelys this wall is formed by the ven- parietal just behind the first convexity. It is tral process ofthe prefrontal, and that is pre- possible that the second convexity is com- sumed to be the case in Australochelys. The parable to scale area 10 in Proganochelys medial edge of the process in Australochelys (Gaffney, 1990: fig. 17). Just behind the sec- can be seen on the left side and it is larger ond convexity in Australochelys there is a and more extensive than the prefrontal pro- shallow concavity, and this may be compa- cess in Proganochelys. rable to scale area 8 in Proganochelys, also a A distinct foramen orbitonasale is not vis- poorly defined concavity. However, these are ible in Australochelys, although there are the only possible similarities between as- poorly preserved indentations in the correct sumed scale areas in Australochelys and Pro- position in both orbits that could be inter- ganochelys. In any case, Australochelys lacks preted as the foramina. definable scale areas, at least as preserved. The postorbital region ofthe skull roof(figs. Australochelys lacks any indication of the 4-6) is preserved on both sides of Austral- supraorbital bumps, scale areas 1-4, seen in ochelys, but it is more complete on the left Proganochelys. On both sides of Austral- side. It is possible that postorbital-jugal su- ochelys the supraorbital margin is depressed tures are represented by cracks in the orbital from the level of the skull roof. No sutures margins on both sides. They are in positions delimiting possible prefrontal or postorbital slightly higher than in Proganochelys. There bones are visible on the dorsal surface or on are no other possible sutures. The postorbital the ventral surface of the skull roof that is section ofthe skull, between the orbit and the visible within the fossa orbitalis. quadrate, is shorter in Australochelys than in The fossa orbitalis in Australochelys (fig. Proganochelys. 1 1) is unusually large, relatively larger than The temporal emargination in turtles has that in any other turtle, including Progan- had a complex history (see Gaffney, 1979, for ochelys. It is not possible to identify the cir- discussion), and its early history as seen in cumorbital bones, but the skull roof is rela- Proganochelys and Australochelys is no ex- tively thin in the roof of the fossa orbitalis. ception (fig. 10). In Proganochelys the su- The bone in the lower margin of the fossa praoccipital and opisthotic are broadly ex- orbitalis (presumed maxilla) is also thin. The posed in dorsal view, whereas in Austral- orbit seems to be as large as morphologically ochelys the posterior margin ofthe skull roof possible in a skull of this size. extends backward to nearly cover them in Both orbits ofAustralochelys have the an- dorsal view. In lateral views the distal end of terior wall of the fossa orbitalis exposed. In the opisthotic in Australochelys is covered, 8 AMERICAN MUSEUM NOVITATES NO. 3130

TABLE 1 Comparative Skull Measurements (in cm) Australochelys Proganochelys BPI/1/4933 SMNS 16980 MB.1910.42.2 Premaxilla condyle length 9.1 9.7 13.3 Premaxilla-posterior edge of skull roof 9.0 8.1 l2.0a Width of skull over orbits 5.0 7.0 7.4b Width of skull at mandibular articulation 8.7 8.1 7.6b Width of maxillary triturating surface 1.0 0.6 1.2 Length of triturating surface 3.9a 4.4 7.2 (L) Width of orbit 3.4 (L) 3.2 (R&L) 3.5 (R) 3.8 (R) Distance between posterior margin of orbit and poste- 3.3 (L) 3.9 (R) 5.8 (L) rior edge of quadrate 3.5 (L) 3.8 (L) Distance between anterior margin of orbit and posteri- 1.9(L) 2.0 2.9 or margin of apertura narium extema a One margin measured is a broken edge. b Measurement obviously influenced by postmortem deformation.

whereas it is exposed in Proganochelys. In PALATAL ELEMENTS Proganochelys the temporal roof is complete (figs. 2, 3, 12-14, 21, 22) and unemarginated in comparison with most turtles. However, Australochelys has a more Most ofthe premaxillary region ofthe Aus- extensive skull roof, and we think that this tralochelys skull is missing due to prefossil- is best interpreted as the primitive condition ization breakage (fig. 2). Ifthe internarial bar for Casichelydia, but not at the level of Tes- is formed from the premaxilla, then this would tudines. Because the less extensive temporal be the only part of that bone preserved. roof of Proganochelys is hypothesized as the Most of both maxillae are present in Aus- primitive condition for Testudines, the more tralochelys and a few possible sutures are in- extensive roof of Kayentachelys (Gaffney et terpretable on the ventral surface (figs. 2, 3, al., 1987) suggests that the more extensive 21, 22). In lateral view no sutures are visible, Australochelys condition is primitive for Cas- but the maxilla presumably forms the ventral ichelydia. margin of the skull anteriorly. This margin The posterior margin of the skull roof is is an eroded edge on both sides of the skull, nearly complete on the left side and the me- but it is unlikely that very much bone is miss- dial portion is complete on the right side of ing. None ofthe original margin is preserved. Australochelys. There is a distinct recessed Thus, the actual extent of the bone is un- lower margin or shelf along the posterior known, but based on the angle and thickness margin of the skull roof in Australochelys, of the preserved surfaces, the original labial which is not present in Proganochelys. In Pro- ridge would probably have been similar to ganochelys there is a raised tubercle, scale that in Proganochelys. area 9, along the posterolateral edge of the The triturating surface in Australochelys is temporal margin. This scale area is formed narrow, much as in Proganochelys but wider by the supratemporal bone. In Australochelys posteriorly than in that form (figs. 19-22). there is no sign of a raised scale area along The lingual ridge is distinct but low, as in the temporal margin, which may indicate the Proganochelys, but differs from Proganoche- absence of the supratemporal, as in all other lys in becoming lower posteriorly. turtles. However, the absence of sutures in On the right side there is what appears to Australochelys makes this impossible to con- be the palatine-maxilla suture running an- firm. teromedially from the foramen palatinum 1995 GAFFNEY AND KITCHING: MORPHOLOGY OF A USTRALOCHELYS 9

posterius parallel to the triturating ridges (figs. a tooth-bearing ridge formed by the pterygoid 2, 3). This contact can also be seen on the in Proganochelys (Gaffney, 1990). In Progan- left side, adjacent to the foramen palatinum ochelys the pterygoids extend anteriorly on posterius. Also on the left side, the contact the midline to prevent medial contact of the of the palatine with the pterygoid just lateral palatines. This condition could well be the to the foramen palatinum posterius is visible. case in Australochelys, but there are no su- If these interpretations are correct, the fora- tures to allow identification of the bones in- men palatinum posterius is formed primarily volved. by the palatine and pterygoid as in Progan- Laterally, the palatine can be identified ochelys (figs. 19-22). contacting the maxilla and forming the lateral It is possible that a midline suture is pres- wall ofthe apertura narium interna. This con- ent and that the vomer in Australochelys may tact also consists of a curved, nearly vertical be paired. Even though this cannot be con- sheet of bone rather than the flat plate seen firmed, we describe the vomer as if it were in other turtles. Between the two vertical paired. Anteriorly, the vomer has a free, but sheets there are numerous, more horizontal broken, margin as preserved; its presumed fragments ofbone lying in matrix well dorsal contact, the premaxilla, is missing. There is to the level of the triturating surfaces. These no evidene of teeth on the vomer of Aus- fragments probably represent a thin, curved tralochelys. The vomer in Australochelys is a portion of the palatine that connected the very unusually shaped element that curves more medial and the more lateral sheets (figs. dorsally well above the level ofthe triturating 2, 21). surfaces (figs. 2, 3, 23). In Proganochelys the The apertura narium interna in Austral- vomer is somewhat above the rest ofthe pal- ochelys (figs. 2, 3) is unusual in being very ate, but not to the extent seen in Austral- large and confluent anteriorly at the level of ochelys. The vomer in Proganochelys is a nar- the triturating surfaces. The high, dorsally row, relatively straight-sided element, simi- placed vomers only separate the internal nar- lar to that in other turtles; however, in Aus- es above the level ofthe palate. The V-shaped tralochelys the area presumably formed by sheet of palatine/pterygoid posterior to the the vomer is only narrow posteriorly where vomers ends ventrally at the level ofthe trit- it has a small contact with the presumed pal- urating surfaces and separates the nares pos- atine. The vomer is very wide anteriorly with teriorly. As preserved, the apertura narium an anteroventrolateral process reaching to the intema extends posteriorly lateral to the maxilla. V-shaped bone to the position ofthe foramen Parts of the areas interpreted as palatines palatinum posterius. The apertura narium in- (figs. 2, 3, 21, 22) are preserved on both sides terna ofAustralochelys is much more exten- in Australochelys, although the extent of sive than in Proganochelys or in Kayentache- missing bone is not determinable. The pre- lys and other primitive turtles. There is ev- sumed palatine in Australochelys is unusual idence that bone (probably palatine) formed for any turtle, including Proganochelys. In a dorsal, choanal trough for at least some of most turtles the palatine is a flat, horizontal the opening, as shown in figure 21. element that forms the roof of the palate. In Australochelys, however, the palatine is reduced to a series of more vertical plates or PALATOQUADRATE ELEMENTS struts, and the area usually covered by the (figs. 2, 3-6, 12-18) palatine contains an unusually large apertura Parts ofwhat are presumed to be the quad- narium interna. On the midline, what is pre- rates are preserved on both sides ofAustral- sumed to be palatine is a nearly vertical sheet ochelys (figs. 4-6); the left side is more com- ofbone that meets the other palatine to form plete, but both sides lack the ventral portion a V-shaped structure. This structure is con- and condylus mandibularis. On the left side cave ventrally and widens posteriorly as the some of the dorsal contact of the quadrate interpterygoid vacuity (figs. 12-14). This with the squamosal can be seen. In contrast V-shaped strut seems to be comparable with to Proganochelys, the distal end of the opis- 10 AMERICAN MUSEUM NOVITATES NO. 3130

Fig. 7. Australochelys africanus Gaffney and Kitching, 1994, BP/1/4933. Occipital view of skull. Stereophotgraph. See fig. 9A for key.

thotic in Australochelys is covered in lateral its lower portions are missing, but it is curved view by this area of squamosal and quadrate in lateral view so that it appears to have been contact. The lateral covering ofthe opisthotic C-shaped originally. There is no develop- is an advanced feature ofthe Rhaptochelydia, ment of a recessed, funnel-shaped cavum Australochelys plus Casichelydia. tympani as seen in Casichelydia. In Casiche- The posterior margin of the quadrate in lydia the distal end of the columella auris Proganochelys is a relatively thick, broadly attaches to the tympanic membrane, whereas curved surface (figs. 10, 23). In Australochelys in Proganochelys and primitive tetrapods the the posterior free edge of the quadrate is an stapes articulates with the quadrate. A stapes acute margin, quite different from Progan- is absent in Australochelys, so its distal artic- ochelys (figs. 6A, 1 4A). This edge in Austral- ulation is not known; however, it is unlikely ochelys is preserved only on the left side and that it articulated with the quadrate, as in

Fig. 8. Australochelys africanus Gaffney and Kitching, 1994, BP/l/4933. Anterior view of skull. Stereophotograph. See fig. 9B for key. 1 995 GAFFNEY AND KITCHING: MORPHOLOGY OF A USTRALOCHELYS I1I A B

crista supraoccipitalis

foramen magnum foramen jugulare anterius apertura narium externa

ventral tubercle of basioccipital (broken) Fig. 9. A. Occipital view of skull. B. Anterior view of skull.

Proganochelys. The area on the quadrate the foramen palatinum posterius. The right where the stapes articulates in Proganochelys pterygoid contacts the maxilla anterolaterally is deeply recessed and far from the presumed in front of the foramen palatinum posterius tympanic margin in Australochelys. Thus, it and apparently contacts the jugal lateral to is likely that the distal end of the stapes in the foramen. The foramen palatinum pos- Australochelys attached to the tympanic terius of Australochelys is less than half the membrane. size of that in Proganochelys. Australochelys agrees with Proganochelys Both pterygoids in Australochelys have in- in completely lacking an antrum postoticum complete temporal fossa margins, but the left and a recessed cavum tympani of the sort pterygoid does not seem to be lacking very seen in Casichelydia (fig. 23). It is difficult to much. There is no indication ofa cryptodiran be sure where the tympanic membrane at- processus pterygoideus externus or a pleu- tached in Australochelys because there is no rodiran processus trochlearis pterygoideus. clear attachment edge. However, the pre- The size and shape ofthe broken edge instead sumed quadrate in Australochelys has a re- suggest a downturned margin like that in Pro- cessed marginal area paralleling the C-shaped ganochelys. posterior edge, and this could represent an The anterior or palatine process ofthe pter- incipient cavum tympani, if the tympanic ygoid cannot be delimited by sutures, but the membrane attached along its proximal mar- pterygoid may form part or all ofthe V-shaped gin. sheet lying behind the vomers on the midline Most of both presumed pterygoids (figs. 2, (figs. 2, 3, 8, 12-14). The space enclosed by 3, 12-18, 21-23) are preserved in Austral- this sheet anteriorly, and by the main body ochelys. There is a probable suture with the of the pterygoid posteriorly, is the interpter- maxilla on the left side, but no other ptery- ygoid vacuity, which is found in other turtles goid sutures are visible. The left pterygoid is only in Proganochelys and Kayentachelys and, more complete than the right, but both are now, Australochelys (fig. 10). The interpter- missing some bone. The right pterygoid is ygoid vacuity in Australochelys is about the missing most of the processus pterygoideus same length as in Proganochelys but wider extemus and only an isolated fragment of than in Proganochelys. In Kayentachelys the bone contacts the maxilla and forms part of vacuity is much smaller than in either Aus- 12 AMERICAN MUSEUM NOVITATES NO. 3130

A B ,pm

interpterygoid vacuity

fenestra ovalis ' / foramen jugulare anterius

E F

interpterygoid vacuity

-pt

fenestra ovalis OP Fig. 10. Comparative views of turtle skulls. A, B. Proganochelys quenstedti, Late Triassic, from Gaffney, 1990. C, D. Australochelys africanus, Early Jurassic. E, F. Kayentachelys aprix, Early Jurassic, from Gaffney et al., 1987. A, C, E are dorsal views; B, D, F, ventral views. 1995 GAFFNEY AND KITCHING: MORPHOLOGY OF A USTRALOCHELYS 13

so pa A B sq

aquu foramen jugulare anterius cranio-quadrate space ventral tubercle of basioccipital

canalis cavernosus ventral tubercle of basioccipital (broken) E

pm cavum tympani Fig. 11. Comparative views of turtle skulls. A, B. Proganochelys quenstedti, Late Triassic from Gaffney, 1990. C, D. Australochelys africanus, Early Jurassic. E. Kayentachelys aprix, Early Jurassic, from Gaffney et al., 1987. A, C, and E are lateral views; B and D, occipital views. tralochelys or Proganochelys; this is presum- ridge is longer and extends the length of the ably a casichelydian synapomorphy. interpterygoid vacuity. This more extensive In Australochelys a ridge forms the lateral ridge on the right side bends and parallels the margin of the interpterygoid vacuity that is edge of the interpterygoid vacuity. The ridge continuous with the V-shaped sheet. This is roughly comparable to the central set of ridge is in roughly the same position as the tooth rows in Proganochelys. Close exami- medialmost row of palatal teeth in Progan- nation of this ridge in Australochelys reveals ochelys (fig. 10). It is possible that this ridge small bumps that may be worn denticles, but in Australochelys also had denticles, but none the preservation is too poor to allow a definite are preserved. conclusion. In any case, Australochelys lacks Lateral to the ridge forming the margin of the extensive tooth rows seen in Progan- the interpterygoid vacuity, the pterygoid ochelys, although some teeth may have been (possibly also the palatine) is relatively flat. present along the ridges described above. This On this flat portion there is a low ridge that reduction in tooth rows appears to be a syn- parallels the more medial ridge. On the left apomorphy for the Rhaptochelydia. side of the skull the more anterior part ofthe One of the most important indicators of ridge is preserved, but on the right side the the relationships ofAustralochelys is the con- 14 AMERICAN MUSEUM NOVITATES NO. 3130

Fig. 12. Australochelys africanus Gaffney and Kitching, 1994, BP/l/4933. Ventrolateral oblique view of right side of skull. Stereophotograph. See fig. 14A for key.

dition of the basipterygoid articulation (fig. tunately there is also no clearly identifiable 23). In Proganochelys the contacts of ptery- suture to show the actual position ofthe con- goid and basisphenoid are clearly a movable tact ofpterygoid and basisphenoid. Although articulation as in the primitive condition for there is minute breakage and infiltration amniotes (Gaffney, 1990). In Kayentachelys cracks of hematite that obscure sutures in and other casichelydians the articulation is a Australochelys, we do not think that a well- suture between the two bones (Gaffney et al., defined suture was present premortem. The 1987). In Australochelys the basipterygoid ar- bone trabeculae and lineations can be seen ticulation is fused with no indication of a in a number of areas where the pterygoid- movable joint (figs. 15-18, 21, 23). Unfor- basisphenoid suture should be and there is

Fig. 13. Australochelys africanus Gaffney and Kitching, 1994, BP/l/4933. Ventrolateral oblique view of left side of skull. Stereophotograph. See fig. 14B for key. 1995 GAFFNEY AND KITCHING: MORPHOLOGY OF A USTRALOCHELYS 15

basipterygoid process A acute edge of quadrate

apertura narium interna

interpterygoid vacuity

basipterygoid process E apertura narium interna

foramen magnum

canalis cavernosus middle ear region Fig. 14. Australochelys africanus Gaffney and Kitching, 1994, BP/1/4933. A. Ventrolateral oblique view of right side of skull. B. Ventrolateral oblique view of left side of skull.

no indication of it. However, a likely fused pterygoid and basisphenoid anterior to the sutural area can be hypothesized based on articulation in Proganochelys are separated indistinct changes in trabeculae lineation and by a space, but in Australochelys bone fills surface grooves, and this is in the approxi- the space (figs. 15, 16). This has doubled or mate position of the articulation in Progan- tripled the pterygoid and basisphenoid con- ochelys. tact area in Australochelys compared with The basicranial articulation is more than Proganochelys. This increased contact area just fused in Australochelys compared with curves vertically and medially from the main Proganochelys. The adjacent parts of the body of the pterygoid. 16 AMERICAN MUSEUM NOVITATES NO. 3130

Fig. 15. Australochelys africanus Gaffihey and Kitching, 1994, BP/1/4933. Ventral view of left ear region. Stereophotograph. See fig. 16 for key.

The quadrate ramus of the pterygoid and two bones form a curved plate as in Progan- the pterygoid ramus of the quadrate cannot ochelys, with no indication ofa swollen cryp- be distinguished from each other because of todiran processus trochlearis oticum. In Aus- the absence of sutures in Australochelys. The tralochelys this quadrate and pterygoid plate

rostrum basisphenoidale interpterygoid vacuity

foramen jugulare anterius

Fig. 16. Australochelys africanus Gaffney and Kitching, 1994, BP/l/4933. Ventrolateral view of left ear region. 1995 GAFFNEY AND KITCHING: MORPHOLOGY OF A USTRALOCHELYS forms the anterior wall of the middle ear re- Australochelys there is a distinct crista su- gion, and it is more enclosed than in Pro- praoccipitals running from the underside of ganochelys. the skull roof to the foramen magnum. It is The quadrate has an, acute, posteriorly de- broken along its posterior edge, but seems veloped flange (figs. 4-6, 12-14) that, along unlikely that much is missing. The crista su- with the ventrally developed opisthotic, en- praoccipitalis in Australochelys (figs. 7, 8), closes the middle ear region to a greater ex- although more extensive than in Progan- tent than in Proganochelys. This is not the ochelys, is not as deep or as well developed cavum acustico-jugulare of Casichelydia, as in Kayentachelys or other Casichelydia. which includes an area walled offposteriorly The base of the crista supraoccipitalis, by the exoccipital; instead, it is a step toward where the cavum cranii expands anteriorly an enclosed middle ear region. to form the braincase proper, is large in Aus- In Proganochelys the lateral head vein tralochelys relative to other turtles. In Pro- passes through a relatively open fissure ganochelys this area is larger than in most formed ventrally by the pterygoid and dor- casichelydians, but not as large as in Aus- sally by the prootic and opisthotic (fig. 23). tralochelys. This area in Australochelys is In casichelydians the lateral head vein is en- probably parietal as well as supraoccipital, closed in a canal, the canalis cavemosus. Al- but there are no sutures. The braincase area though the area in Australochelys is damaged beneath the skull roof is expanded laterally and lacks sutures, there is a well-defined fo- and ends in small, posterolateral processes ramen on each side (figs. 14, 18, 23) that is that curve backward to form a posteriorly clearly a canalis cavemosus of the casiche- opening concavity on each side of the crista lydian type, making this character a rhapto- supraoccipitalis (fig. 7). There is a hint ofsuch chelydian synapomorphy. a condition in Proganochelys but nothing like Breakage and distortion ofthe canalis cav- it in any other turtle. The concavities may be ernosus on each side has made the two open- for the attachment ofcraniocervical muscles. ings different in shape but the position is the The large base ofthe crista supraoccipitalis same. Each foramen has pieces of a broken could be interpreted as a Proganochelys-Aus- wall separating it from another more dorsal tralochelys synapomorphy. However, Pro- foramen so that the result is an irregular oval ganochelys is not significantly larger than in partially separated by broken bone (fig. 18). many casichelydians and it seems best to in- This second foramen may contain the facial terpret the Australochelys condition as an (VII) nerve, which is contained in the cran- autapomorphy. ioquadrate space in Proganochelys. In Aus- Much ofthe basioccipital in Australochelys tralochelys there is so much erosion and dis- is broken offin a nearly flat, oblique section. tortion that it is likely that these two openings Although most ofthe ventral tubercle is miss- have become partially confluent. ing, it is clear that one was present (figs. 15- A clearly defined opening for the stapedial 18) and that it was similar in size and shape artery cannot be found in Australochelys. to the ventral tubercle in Proganochelys (figs. However, there are areas of matrix on both 19, 20). This tubercle does not occur in other sides without clear borders that do seem to turtles, nor does it occur in likely turtle out- penetrate dorsally through the otic chamber groups, such as procolophonids, Owenetta, wall. These areas could be stapedial foramina pareiasaurs, or captorhinids. This character (figs. 17, 18). The absence of sutures and the could be interpreted as a synapomorphy that ambiguity ofthese structures prevents useful would unite Proganochelys and Australoche- comparison, but they lie somewhat more an- lys and is the main contradiction to the hy- teriorly than the canalis in Proganochelys. pothesis of Australochelys plus Casichelydia as a monophyletic group. About two thirds of the condylus occipi- BRAINCASE ELEMENTS talis in Australochelys (figs. 7, 9, 15-18) is (figs. 7, 9, 12-18, 21-23) broken off, but what remains is very similar The crista supraoccipitalis in Proganoche- to that in Proganochelys. The articulation is lys is a mere ridge and virtually absent. In roughly triangular, being slightly wider than 18 AMERICAN MUSEUM NOVITATES NO. 3130

P. ..- .1"'. -t JW I.).% ll. ..If 4--t ,

- 1.

Fig. 17. Australochelys africanus Gaffney and Kitching, 1994, BP/l/4933. Ventrolateral oblique view of left ear region. Stereophotograph. See fig. 18 for key.

high. It is probable that the basioccipital tu- tica are formed, but it does seem to approach bercle was closer to the occipital condyle and that condition. was not completely separated from it in Aus- The foramen jugulare anterius in Austral- tralochelys as it is in Proganochelys. ochelys (figs. 1 1, 15-18) is very similar to that The foramen magnum in Proganochelys is opening in Proganochelys, but it is more wider than high and in Australochelys it is deeply surounded by bone. Nonetheless, a vice versa, possibly made more so by defor- definitive foramen jugulare posterius of the mation. casichelydian type is not present. A clearly defined fenestra ovalis (figs. 17, The processus paroccipitalis of the opis- 18) is present on both sides of the skull in thotic in Australochelys (figs. 19-22) differs Australochelys in the same relative position from that in Proganochelys in having a ven- in Proganochelys. The fenestra ovalis in Aus- tral ridge, developed medially between the tralochelys is oriented more vertically than fenestra ovalis and foramen jugulare anter- in Proganochelys. ius, and extending for the entire length ofthe In Australochelys opisthotic sutures can be processus paraoccipitalis. Distally, the opis- discerned distally only, but the areas presum- thotic is covered, presumably by the squa- ably representing the opisthotics are pre- mosal and quadrate. served on both sides of the skull (figs. 15- No sutures are visible delimiting the ba- 18). In Proganochelys the bar of bone be- sisphenoid in Australochelys, but the pre- tween the fenestra ovalis and the foramen sumed area of the basisphenoid is preserved jugulare anterius, which is the area where the (figs. 15, 16). The ventral surface ofthe bone processus interfenestralis is developed in forms a midline trough, in contrast to the Casichelydia, is very low. In Australochelys midline ridge seen in Proganochelys. In cas- this bar is considerably deeper, separating ichelydians the basisphenoid is generally flat those openings by a ridge ofbone. This is still ventrally. Anteriorly, Australochelys has a far from the definitive processus interfenes- blunt cultriform process in comparison to tralis found in casichelydians where a reces- Proganochelys, but it may have been made sus scalae tympani and fenestra perilympha- more blunt by postmortem damage. None- 1995 GAFFNEY AND KITCHING: MORPHOLOGY OF A USTRALOCHELYS

possible area of fused basipterygoid articulations

acute edge of quadrate Fig. 18. Australochelys africanus Gaffhey and Kitching, 1994, BP/1/4933. Ventrolateral oblique view of the left ear region. theless, it is closer to the casichelydian re- ervation present in Australochelys in which duced or absent cultriform process than to fine detail is lost. Any of these cracks could the elongate condition of Proganochelys. On be the carotid foramina. either side of the base of the cultriform process in Australochelys, the presumed basisphenoid curves ventrolaterally to the pter- RELATIONSHIPS ygoid (figs. 15, 16). This area is open and has The characters relevant to determining the no contact in Proganochelys, but in Austral- relationships of Australochelys are listed in ochelys and the Casichelydia it is part of the table 2. They are discussed below in the same sutured, akinetic joint between basisphenoid order as presented in the table. Further in- and pterygoid. formation on each character can be found in Posteriorly in Australochelys the basisphe- the descriptive section, which is also arranged noid has paired basipterygoid processes that in the same order. contact and are indistinguishably fused to the 1. Apertura narium externa united: 0 - pterygoids. Thus, in addition to the fused apertura narium externa divided by premax- basipterygoid articulation, akinesis in Aus- illae; 1 = apertura narium externa united.- tralochelys also involves a broad contact of The paired nasal openings divided by dorsal pterygoid and basisphenoid for most of the premaxillary processes are primitive tetra- length of the basisphenoid. pod features found in Proganochelys. Aus- Despite all efforts, neither carotid foramen tralochelys has divided aperturae, but the bone could be found. A number of cracks and fis- is only presumed to be premaxilla. In Kay- sures filled with hematite were probed, but entachelys the nares are united as in most none seem to have formed margins. How- other turtles. Forms such as Meiolania (Gaff- ever, this is consistent with the type of pres- ney, 1983) and Kallokibotion (Gaffney and 20 AMERICAN MUSEUM NOVITATES NO. 3130

Fig. 19. Proganochelys quenstedti, ventral view of skull, from Gaffhey, 1990.

Meylan, 1992) have divided nares, but not usually small; 1 = orbit very large. -Aus- due to dorsal processes of the premaxillae, tralochelys is unique among turtles in having the primitive condition. This character is in- a huge fossa orbitalis. This is an Austral- terpreted as a casichelydian synapomorphy. ochelys autapomorphy. The orbit is also large 2. Apertura narium externa elongate: 0 = in Proganochelys, but it is within the range apertura narium externa oval or wide; 1 = seen in sea turtles and is therefore not a good apertura narium externa elongate. -Austral- synapomorphy with Australochelys. ochelys is unique in possessing this feature 5. Cranial sutures fused: 0 = sutures not and we interpret it as an autapomorphy. It is fused; 1 = sutures fused. -This character oc- quite likely associated with the generally large curs sporadically in turtles and in other am- size of the fossa nasalis in this form. niotes and is very difficult to homologize us- 3. Lacrimal foramen present: 0 = lacrimal ing morphologic tests. It is probably common foramen present; 1 = lacrimal foramen ab- to adults of both Proganochelys and Austral- sent. -The lacrimal foramen is primitive for ochelys, but its sporadic distribution (e.g., turtles. It is present in Proganochelys as well baenids, Platysternon, Emydura) in other as Australochelys. Australochelys differs from turtles causes us to reject it as a synapmorphy. Proganochelys in having an unusually large 6. Posterior margin of temporal roof ex- foramen. This is a presumed autapomorphy, tends past opisthotic in dorsal view: 0 = tem- as the foramen is generally smaller in more poral roof anterior to opisthotic; 1 = tem- primitive tetrapods. The absence of the lac- poral roof posterior to opisthotic. -Because rimal foramen is interpreted as a casichely- this condition occurs in Australochelys, Kay- dian synapomorphy. entachelys, and Kallokibotion (Gaffney and 4. Orbit very large: 0 = orbit variable but Meylan, 1992), we interpret it as a rhapto- 1995 GAFFNEY AND KITCHING: MORPHOLOGY OF A USTRALOCHELYS 21

foramen palatinum posterius

p (\ processus pterygoideus

foramen caroticum basisphenoidale

foramen jugulare anterius ventral tubercle of basioccipital Fig. 20. Proganochelys quenstedti, ventral view of skull, redrawn from Gaffney, 1990. chelydian synapomorphy (i.e., uniting Aus- we conclude that the use of this feature is tralochelys and Casichelydia). Other turtles unclear. that have posteriorly extensive skull roofs 8. Vomers highly arched dorsally, narrow have constituent bones with different pro- posteriorly, broad anteriorly: 0 = vomers rel- portions, which suggests that the Kayentache- atively horizontal; 1 = vomers highly lys condition is primitive for cryptodires. We arched. -The morphology ofthe anterior end do not know with certainty what bones make ofthe palate in Australochelys is very unusual up the temporal roof in Australochelys. For and is here interpreted as an autapomorphy. the present, it seems best to use this condition 9. Palatal teeth few or absent: 0 = palatal as a rhaptochelydian synapomorphy. teeth present (vomerine teeth present); 1 = 7. Triturating surfaces narrow: 0 = tritu- palatal teeth few or absent (vomerine teeth rating surfaces narrow; 1 = triturating sur- absent). -Although we are not able to resolve faces narrow and wide. -The widths ofjaws the question of palatal teeth in Australoche- in turtles vary a great deal, but narrow jaws lys, many teeth on the palatal surface clearly seem to be the primitive condition at least at were not present, as in Proganochelys. We the level ofTestudines. Because all outgroups interpret this reduction or absence of palatal for turtles have narrow, toothed jaws, it is teeth, particularly the absence ofteeth on the hard to make outgroup comparisons here and vomer, as a rhaptochelydian synapomorphy. 22 AMERICAN MUSEUM NOVITATES NO. 3130

Fig. 21. Australochelys africanus, ventral view of skull.

10. Basipterygoid articulation sutured: 0 = 12. Large interpterygoid vacuity: 0 = in- basipterygoid articulation a movable joint; 1 terpterygoid vacuity large; 1 = interpterygoid = basipterygoid articulation sutured. -This vacuity small or absent. -See character 11. is considered to be the most important fea- 13. Cranioquadrate space a well-defined ture relatingAustralochelys to casichelydians. canal: 0 = cranioquadrate space open; 1 = The loss of kinesis of the palatoquadrate el- cranioquadrate space closed to a canal. -In ements in the turtle braincase was closely cor- Proganochelys and generalized tetrapods the related with a number of reorganizational cranioquadrate space is relatively open. In changes in the turtle skull involving the origin casichelydians it is closed, resulting in a canal of the cryptodires and pleurodires (Gaffney, or canals for the enclosed structures. Aus- 1975, 1979). The morphology of the con- tralochelys is united with casichelydians in tacting elements in turtles differs significantly having the cranioquadrate space reduced to from the morphology in other groups that a foramen (presumably leading into a canal) have developed akinesis (e.g., temnospon- for the lateral head vein, the canalis caver- dyls, pareiasaurs, therapsids, archosaurs). nosus (Gaffney, 1979, 1983). This condition 11. Cultriform process present: 0 = cultri- seems to be correlated with the loss of mo- form process present; 1 = cultriform process bility at the basipterygoid articulation, but absent. -The persistence of a visible cultri- this is not always the case in other akinetic form process and related large interpterygoid taxa. We interpret this character as a rela- vacuity are interpreted as primitive features tively well-tested rhaptochelydian synapo- retained in Australochelys and lost in Casi- morphy. chelydia. We consider the small or closed in- 14. Stapes thin and articulates with a tym- terpterygoid vacuity and the absence ofa cul- panic membrane: 0 = stapes heavy and ar- triform process as casichelydian synapomor- ticulates with a concavity on the quadrate; 1 phies. = stapes thin and articulates with a tympamic 1995 GAFFNEY AND KITCHING: MORPHOLOGY OF AUSTRALOCHELYS 23

apertura narium iriterna \N

interpterygoid vacuity foramen palatinum posterius

pal basipterygoid process

processus pterygoideus externus qu foramen juJgulare anterius ventral tubercle of basioccipital (broken) Fig. 22. Australochelys africanus, ventral view of skull. membrane. -As described in Gaffney (1990), the quadrate has a large, posterior flange that the stapes of Proganochelys is a stout, rela- separates the well-defined cavum tympani tively heavy column that extends from the from the more medial middle ear region and fenestra ovalis to a distinct concavity on the forms the unique, funnel-shaped quadrate of quadrate where the distal end of the stapes living turtles. Proganochelys lacks the flange articulates. In pleurodires and cryptodires the and has only a curved, C-shaped quadrate, stapes is much thinner, particularly distally, not funnel-shaped. Australochelys also lacks where it contacts the tympanic membrane. the large quadrate flange that forms the de- In Kayentachelys the stapes is not preserved, finitive funnel-shaped cavum tympani, but it but the position ofthe fenestra ovalis and the does seem to be advanced over Proganoche- shape ofthe quadrate, which lacks a stapedial lys. The Australochelys quadrate has a ver- articulation concavity, indicate a stapes that tical ridge along its posterior edge that we articulates with the tympanum. Similarly, in interpret as an early stage of development of Australochelys the stapes is not preserved, but the larger quadrate flange seen in casichely- there is no stapedial articulation concavity dians. on the quadrate and the stapes probably ar- 16. Funnel-shaped cavum tympani pres- ticulated with a tympanic membrane. In any ent: 0 = funnel-shaped quadrate absent; 1 = case, the light stapes is a casichelydian syn- funnel-shaped quadrate present. -Progan- apomorphy that is circumstantially, but not ochelys and Australochelys lack the funnel- definitely, present in Australochelys. shaped, recessed cavum tympani formed by 15. Quadrate with acute ridge on posterior a large quadrate flange as seen in casichely- margin: 0 = ridge absent; 1 = ridge or well- dians. This character could be interpreted as developed flange present. -In casichelydians a more advanced state of the previous char- 24 AMERICAN MUSEUM NOVITATES NO. 3130

A foramen palatinum posterius

apertura narium interna

B foramen palatinum posterius

apertura nari interna foramen magnum

canalis cavernosus (= cranio-quadrate space) ? foramen nervi facialis Fig. 23. Comparison ofear regions in posterior oblique views. Anterior is to the left. A. Proganochelys quenstedti, redrawn from Gaffhey, 1990. B. Australochelys africanus. acter, although we prefer to use only two-state and Australochelys, though it is modified in characters. various ways within casichelydians. The ven- 17. Ventral ridge on opisthotic present: 0 tral ridge is another component of an en- = ventral ridge on opisthotic absent; 1 = ven- closed middle ear region, that is absent in tral ridge on opisthotic present. -This is an- Proganochelys but present in Australochelys. other character common to casichelydians 18. Distal end of opisthotic covered by 1995 GAFFNEY AND KITCHING: MORPHOLOGY OF AUSTRALOCHELYS 25

Fig. 24. Australochelys africanus Gaffney and Kitching, 1994, BP/1/4933. Shell fragment, probably from the bridge area, exposed only in internal view. Orientation not determined. squamosal and/or quadrate: 0 = opisthotic jugulare posterius, they still have at least the not covered; 1 = opisthotic covered.-This medial edge formed by a flange from the ex- feature, which is held in common by Aus- occipital. This character is closely related to tralochelys and casichelydians, is related to the following character. the expansion of the cavum tympani in cas- 21. Recessus scalae tympani and fenestra ichelydians. In Australochelys it is only a rel- perilymphatica formed in bone: 0 = not de- atively slight increase ofthe dermal bone cov- fined by bone; 1 = defined by bone.-Pos- ering in the upper part ofthe tympanic region. sibly related to the formation ofthis second- 19. Paroccipital process of opisthotic ary foramen, the foramen jugulare posterius, tightly sutured to quadrate and prootic: 0 = is the peculiarly chelonian fenestra perilym- process separated from prootic and quadrate; phatica and recessus scalae tympani (Gaff- 1 = process tightly sutured to prootic and ney, 1979). All these features occur only in quadrate. -In Proganochelys, as in more gen- casichelydians and are interpreted as syna- eralized tetrapods with the primitive form of pomorphic for that group. cranial kinesis (Gaffney, 1990), the paroccip- 22. Ventral basioccipital tubercle: 0 = tu- ital process lies against the quadrate and bercle absent; 1 = tubercle present. -Both prootic with a relatively open fissure sepa- Proganochelys andAustralochelys have a well- rating much of their length. Australochelys developed tubercle on the ventral side. The and casichelydians have this space complete- tubercle does not occur in any other turtles. ly closed with no possible movement of the It also does not occur in any likely turtle out- paroccipital process. Along with other fea- groups, such as procolophonids, Owenetta, tures consistent with the loss of kinesis in pareiasaurs, or captorhinids. As far as can be Australochelys and Casichelydia, we interpret determined, the morphology of the tubercle this character as a synapomorphy uniting is the same in Australochelys and Progan- these groups. ochelys. Because the hypothesis that Aus- 20. Foramen jugulare posterius formed in tralochelys is more closely related to casiche- bone: 0 = foramen jugulare posterius com- lydians than to Proganochelys is most par- pletely absent; 1 = at least medial edge of simonious under our analysis, the tubercle foramen jugulare posterius formed by exoc- would be interpreted as homoplastic and pre- cipital. -In Proganochelys and Australoche- sumably lost in casichelydians. lys the foramen jugulare anterius is the fo- 23. Crista supraoccipitalis is a distinct ver- ramen seen in occipital view. In casichely- tical sheet of bone: 0 = crista supraoccipitalis dians it is a secondary foramen, the foramen absent or small; 1 = crista supraoccipitalis jugulare posterius, that is visible in occipital distinct vertical sheet. -The crista supraoc- view. Even though some casichelydians, such cipitalis is Proganochelys is a small ridge. In as baenids, may have an incomplete foramen Australochelys it is a distinct, thin sheet of 26 AMERICAN MUSEUM NOVITATES NO. 3130

Casichelydia

Fig. 25. Relationships ofAustralochelys with geologic ranges of taxa indicated. Diagnoses of higher taxa are in the text.

bone extending ventrally from the skull roof cause most ofthe preserved edges are broken. and separating the cavum cranii proper from Nonetheless, this may be interpreted as a the skull roof. In casichelydians the crista su- common feature in Proganochelys and Aus- praoccipitalis is deeper and posteriorly more tralochelys. It is most likely a primitive char- extensive, but is clearly similar to that in Aus- acter at the level of Testudines. tralochelys. We interpret this as a rhaptoche- The above list indicates nine characters in- lydian synapomorphy. terpreted as synapomorphies ofAustraloche- 24. Lateral edge of pterygoid without lat- lys and Casichelydia (6, 9, 10, 13, 15, 17-19, eral processes: 0 = edge of pterygoid down- 23). This contrasts with two characters (4, turned, without lateral processes; 1 = edge of 22; possibly three, if24 is included) that unite pterygoid with lateral process.-In Progan- Australochelys with Proganochelys. There are ochelys and probably Australochelys the lat- no derived characters in Australochelys that eral margin ofthe pterygoid lacks lateral pro- link it with pleurodires or cryptodires sepa- cesses and is downturned or rolled with a low rately. Even without a computer-run parsi- ridge along the margin. The conditions in mony analysis, it is clear that the characters possible outgroups for this character vary support only one tree with Australochelys as considerably, but none has lateral processes. the sister group of Casichelydia and Progan- The Australochelys condition is in doubt be- ochelys as the sister group of Australochelys 19951995GAFFNEY AND KITCHING: MORPHOLOGY OF A USTRALOCHELYS 27

TABLE 2 Comparison of Australochelys with Other Turtles Primitive Character tetrapod Proganochelys Australochelys Casichelydia 1. Apertura narium externa united No No No Usually yes 2. Apertura narium extema elongate No No Yes No 3. Lacrimal foramen Present, small Present, small Present, large Absent 4. Orbit size Variable Large Very large Variable 5. Cranial sutures fused Usually no Probably in Yes Usually no adults only 6. Posterior margin of temporal roof No No Yes Primitively yes extends over opisthotic 7. Triturating surfaces narrow Variable Yes Yes Primitively yes 8. Vomers highly arched dorsally, nar- No No Yes No row posteriorly, broad anteriorly 9. Palatal teeth Many Many Few or absent Few or absent 10. Basipterygoid articulation sutured No No Yes Yes 11. Cultriform process Present Present Present Absent 12. Large interpterygoid vacuity Yes Yes Yes No 13. Cranioquadrate space a well-defined No No Yes Yes canal 14. Stapes thin and articulates with tympan- No No Probably yes Yes ic membrane 15. Quadrate with acute ridge on posteri- No No Yes Primitively yes or margin 16. Recessed, funnel-shaped cavum Absent Absent Absent Present tympani 17. Ventral ridge on opisthotic No No Yes Yes 18. Distal end of opisthotic covered by No No Yes Yes squamosal/quadrate 19. Paroccipital process of opisthotic tight- No No Yes Yes ly sutured to braincase 20. Foramen jugulare posterius formed No No No Yes in bone at least medially 21. Recessus scalae tympani and fenestra No No No Yes perilymphatica defined by bone 22. Ventral basioccipital tubercle Absent Present Present Absent 23. Crista supraoccipitalis a distinct verti- No No Yes Yes cal sheet of bone 24. Lateral edge of pterygoid Variable, no lat- Downturned, Probably down- Variable but eral process no lateral turned, no lat- with lateral process eral process process plus Casichelydia. The conflicting characters ofAustralochelys is that it gives a view ofthe uniting Proganochelys and Australochelys turtle skull well after the origin ofturtles but would be interpreted as primitive for turtles before the origin ofthe modem groups, when that were lost once rather than having orig- the themes of akinesis and middle ear mod- inated twice. ification were just beginning. The skulls of Recent turtles are highly Akinesis has evolved independently in a modified over the generalized amniote con- number of amniotes, with mammals being dition. However, such forms as Proganoche- the most prominent among the Recent fauna. lys and Australochelys and other fossils do As discussed elsewhere (Gaffney, 1990), the provide important glimpses ofhow the more primitive chelonian skull was kinetic in the derived conditions evolved. The significance sense ofa movable basipterygoid articulation 28 AMERICAN MUSEUM NOVITATES NO. 3130

as retained from the generalized amniote space is largely filled by bony contacts be- condition, and does not necessarily mean that tween palatoquadrate and braincase. The palatoquadrate/braincase movement actual- middle ear region of Australochelys is more ly took place. Proganochelys has this gener- ossified and enclosed than in Proganochelys. alized condition of kinesis; all other turtles Whether this is a consequence or a precursor are akinetic. Living cryptodires and pleuro- of akinesis is not known, nor is the function dires have extensive bony processes and su- of this ossification and enclosure known. It tured contacts that obliterate the open areas seems likely that the hypertrophy ofthe mid- and sliding contacts of the primitive condi- dle ear region preceded the origin of the jaw tion (Gaffney, 1975). Australochelys is aki- muscle trochleas (Gaffney, 1975) and that ak- netic in having a fused basipterygoid artic- inesis and otic chamber hypertrophy are re- ulation and an expanded basisphenoid-pter- lated to each other. ygoid contact but lacks other features of cas- The Casichelydia was originally erected by ichelydian akinesis. The evolutionary reasons Gaffney (1975) for all turtles except Progan- for the development of akinesis are specu- ochelys because they had a series of derived lative. The condition clearly affects nearly ev- characters in common, including akinesis. ery cranial system, especially the feeding The Casichelydia could be redefined to in- mechanism and the auditory system. Turtles clude Australochelys, but we have chosen to had already evolved a beak before akinesis, retain Casichelydia as cryptodires plus pleu- but the trochlear jaw mechanism did not ap- rodires and redefine it with the derived char- pear until cryptodires and pleurodires were acters in the following diagnosis: differentiated. In Australochelys the feeding mechanism appears to be virtually the same as in Proganochelys. CASICHELYDIA The auditory system, however, is clearly different between Proganochelys and Aus- DiAGNosIs: Apertura narium externa unit- tralochelys, and this change is probably re- ed, not separated by premaxillary processes lated to akinesis. The kinetic skull of Pro- as in Proganochelys; lacrimal bone and fo- ganochelys has an open, slit-shaped crani- ramen absent; cultriform process absent; in- oquadrate space between the braincase and terpterygoid vacuity small or absent; cavum the palatoquadrate. The middle ear region is tympani well developed and recessed into a relatively open and has anterior and medial funnel-shaped bony structure; a definitive fo- walls that could be movable with respect to ramen jugulare posterius formed in bone, at one another. In Australochelys the cranio- least medially, in contrast to Proganochelys quadrate space is replaced by a foramen or and Australochelys; recessus scalae tympani foramina for the included structures and the and fenestra perilymphatica defined by bone.

REFERENCES De Broin, F. 1983. The cranial morphology of the extinct 1988. Les tortues et le Gondwana. Examen homed turtle, Meiolania platyceps, from des rapports entre le fractionnement du the Pleistocene of Lor Howe Island, Gondwana et la dispersion geograp- Australia. Bull. Am. Mus. Nat. Hist. hique des tortues pleurodires a partir du 175(4): 326-479. Cretace. Studia Geol. Salmanticensia 1990. The comparative osteology of the Tri- 2(5): 103-142. assic turtle Proganochelys. Bull. Am. Gaffney, E. S. Mus. Nat. Hist. 194: 1-263. 1975. A phylogeny and classification of the Gaffney, E. S., and J. W. Kitching higher categories of turtles. Bull. Am. 1994. The most ancient African turtle. Nature Mus. Nat. Hist. 155(5):387-436. 369: 55-58. 1979. Comparative cranial morphology ofre- Gaffney, E. S., and P. A. Meylan cent and fossil turtles. Bull. Am. Mus. 1988. A phylogeny ofturtles. In M. J. Benton Nat. Hist. 164(2): 65-376. (ed.), The phylogeny and classification 1995 GAFFNEY AND KITCHING: MORPHOLOGY OF A USTRALOCHELYS 29

of the tetrapods. Vol. 1: Amphibians, Nesbitt, E. A., and G. Bond reptiles, birds. Systematics Association 1972. Fossil chelonians from the Gokwe For- Special Vol. No. 35A: 157-219. Oxford: mation, Rhodesia. Arnoldia (Rhodesia) Clarendon Press. 5(37): 1-7. 1992. The transylvanian turtle, Kallokibotion, Olsen, P. E., and P. M. Galton a primitive cryptodire ofCretaceous age. 1984. A review of the reptile and amphibian Am. Mus. Novitates 3040: 1-37. assemblages from the Stormberg of Gaffney, E. S., J. H. Hutchison, F. A. Jenkins, Jr., southern Africa with special emphasis and L. J. Meeker on the footprints and the age of the 1987. Modern turtle origins: the oldest known Stormberg. Palaeontol. Afiicana 25: 87- cryptodire. Science 237: 289-291. 110. Kitching, J. W., and M. A. Raath Visser, J. N. J. 1984. Fossils from the Elliot and Clarens for- 1984. A review of the Stormberg Group and mations (Karoo Sequence) ofthe north- Drakensberg Volcanics in southern Af- eastern Cape, Orange Free State and Le- rica. Palaeontol. Africana 25: 5-27. sotho, and a suggested biozonation based on tetrapods. Palaeontol. Africana 25: 111-125.

Recent issues of the Novitates may be purchased from the Museum. Lists of back issues of the Novitates, Bulletin, and Anthropological Papers published during the last five years are available free of charge. Address orders to: American Museum of Natural History Library, Department D, Central Park West at 79th St., New York, N.Y. 10024. TEL: (212) 769-5545. FAX: (212) 769- 5009. E-MAIL: [email protected]

L-

® This paper meets the requirements of ANSI/NISO Z39.48-1992 (Permanence of Paper).