Rhothonemys, Gaffney & Tong & Meylan, 2006

GAFFNEY, EUGENE S, TONG, HAIYAN & MEYLAN, PETER A, 2006, EVOLUTION OF THE SIDE-NECKED TURTLES: THE FAMILIES BOTHREMYDIDAE, EURAXEMYDIDAE, AND ARARIPEMYDIDAE, Bulletin of the American Museum of Natural History 300 (300), pp. 1-698 : 457-570

Structures: The prefrontal in Rhothonemys is relatively large, being larger than the frontal and similar in relative size to the prefrontal in Phosphatochelys . The prefrontal in Rhothonemys protrudes anteromedially, as in Phosphatochelys , producing a blunt point on the midline. In Rhothonemys the degree of protrusion is more extensive than in Phosphatochelys , also correlated with the larger size of the apertura narium externa. The posterior margin of the prefrontal in Rhothonemys is straight, as in Taphrosphys , not indented, as in Phosphatochelys , due to the very small frontal in that taxon. The prefrontal in Rhothonemys forms the anterodorsal margin of the orbit, as in Taphrosphys , but does not extend posteriorly to the degree seen in Phosphatochelys . The interorbital distance in Rhothonemys is much narrower than in Phosphatochelys and Azabbaremys ; it is similar in relative width to Taphrosphys .

The ventral process of the prefrontal in Rhothonemys separates the very large fossa nasalis from the fossa orbitalis. The process is similar in extent in Rhothonemys and Phosphatochelys but is thinner than in Pelomedusidae and Cearachelyini . The ventral surface of the prefrontal in Rhothonemys forms the roof of the large fossa nasalis and is broadly concave ventrally on each side of the midline (fig. 211A). The shape of this paired depression is more distinct than in Phosphatochelys , Pelomedusidae , and Cearachelyini .

FRONTAL (figs. 208–211)

Preservation: Both frontals are present in AMNH 30521, and dorsal and ventral surfaces are visible. The parasagittal ridge delimiting the sulcus olfactorius is broken along its ventral margin in both prefrontals. A small part of the orbital edge is broken posteriorly on both prefrontals, but most of the edge is intact.

Contacts: The frontal in Rhothonemys contacts the prefrontal anteriorly, the parietal posteriorly, and the other frontal medially. In most bothremydids the frontal also contacts the postorbital, but in Rhothonemys the contact is separated by the parietal exposure in the orbital margin. In Phosphatochelys the contact is also lacking, due to the small size of the frontal and the broad prefrontal-parietal contact.

Structures: The frontal in Rhothonemys forms the dorsal part of the orbital margin, much as in Taphrosphys , and in strong contrast to Phosphatochelys , in which the frontal is widely separated from the orbital margin by the parietal-prefrontal contact. The frontal is smaller than the prefrontal in Rhothonemys , as it is in Phosphatochelys . In Taphrosphys , Azabbaremys , and Nigeremys , it is equal to or larger than the prefrontal. In Labrostochelys the prefrontal is very elongate.

The ventral surface of the frontal in Rhothonemys (fig. 211A) shows the anteromedial projection, actually broken off on both sides but determinable from the sutural surface on the prefrontal, extending under the prefrontal, as occurs in most bothremydids and many pleurodires. The parasagittal ridge, although incomplete, separating the sulcus olfactorius from the fossa nasalis is about the same size as in other bothremydids, such as Galianemys . The sulcus in Phosphatochelys seems to be formed by the parietal rather than the frontal.

PARIETAL (figs. 208–211)

Preservation: Both parietals are present in AMNH 30521, but they are incomplete ventrally, posteriorly, and laterally. On the left side, the posterior temporal roof margin is broken and slightly displaced, but it seems to have its natural edge posteriorly. Ventrally, only the dorsalmost part of the processus inferior parietalis is preserved. None of the actual supraoccipital contact is preserved.

Contacts of dorsal plate: The parietal in Rhothonemys contacts the frontal anteriorly, the other parietal medially, and the supraoccipital posteroventrally (displaced and filled with matrix). The postorbital contact is present on the left side and seems to run the length of the parietal. However, the sutures are difficult to discern and a quadratojugal suture cannot be made out, so some of this contact may be the quadratojugal.

Structures of dorsal plate: The edge of the temporal emargination is preserved partially on the left side. Although slightly displaced, the degree of emargination in Rhothonemys is similar in extent to that seen in Phosphatochelys and Azabbaremys , but it is less emarginate than in Taphrosphys .

The ventral surface of each parietal is visible in AMNH 30521 (fig. 211), and it has the ridges where the processus inferior parietalis and the fossa orbitalis form walls separating these spaces. Only the dorsal attachment ridges are present; the walls are broken off. The parietal in Rhothonemys is relatively farther forward in the skull roof than it is in all other Pelomedusoides except Phosphatochelys . This is also expressed in the exposure of the parietal in the orbital margin in Rhothonemys and Phosphatochelys . The parietal in Rhothonemys , Phosphatochelys , and Taphrosphys forms a pocket in the posterior part of the fossa orbitalis, posterior to the orbital margin and anterior to the septum orbitotemporale ridge (all that remains of this wall). In Rhothonemys this pocket (fig. 211) has its walls broken away, but it is very similar to that in Taphrosphys sulcatus (fig. 170C) and T. ippolitoi (fig. 183B).

Contacts and structures of processus inferior parietalis: The processus inferior parietalis in AMNH 30521 is largely missing, and its ventral contacts and structures cannot be determined.

JUGAL (figs. 208–211)

Preservation: Both jugals are present in AMNH 30521, but neither is complete. The left one is lacking some of its anterodorsal area, where it forms the orbital margin, and some of its medial process appears broken. The right jugal is preserved along its contact with the maxilla, but its posterodorsal area and most of its medial process are missing. The posterior margin of the right jugal is intact, but the left one is a broken surface.

Contacts of lateral plate: The anteroventral contact with the maxilla in Rhothonemys runs from the posteroventral orbital margin to the edge of the narrow cheek emargination as in Phosphatochelys . The posterior edge of the jugal is exposed on the cheek emargination ventrally, but dorsally its contacts are broken in AMNH 30521. Dorsally, the jugal contacts the postorbital, and although broken, this can be seen on the left side. Posterodorsally, it is likely that the jugal contacts the quadrate and/or the quadratojugal, but this cannot be determined due to crushing and missing bone in this area.

Structures of lateral plate: The fossa orbitalis in Rhothonemys forms a large pock- et ventral to the orbital margin (fig. 211). This pocket is also found in Phosphatochelys and Taphrosphys but not, as far as can be seen in the flattened specimens, Labrostochelys . In Rhothonemys the jugal forms the posterior half of the suborbital pocket; the maxilla forms the anterior half. The jugal in Rhothonemys and Phosphatochelys is relatively deep dorsoventrally, in contrast to the much shallower jugal in Taphrosphys and Labrostochelys . The suborbital pocket in Rhothonemys is deeper than in Phosphatochelys . These pockets are more accurately considered to be one feature, an expanded fossa orbitalis formed by multiple bones (jugal, postorbital, parietal, maxilla), although it is convenient to describe and label the particular areas as separate. In the character descriptions, the fossa orbitalis enlargement is considered one character, character 27.

The posterior margin of the jugal in Rhothonemys seems to be a free edge, ending in an acute ridge, not a sutural margin. This is very similar to the jugal in Phosphatochelys , which also has a free posterior margin. In Rhothonemys the dorsal and posterior elements of this cheek emargination are not known due to crushing and bone loss, but there does not seem to be room for a large quadratojugal or a large cheek emargination. The free edge on the jugal runs dorsally almost to the postorbital contact.

Contacts of medial process: Both jugals in Rhothonemys have no preserved contacts medially. The left jugal has a sutural surface on the medial edge of the medial process. This is presumably for the pterygoid and palatine contacts. The medial process does not really have a horizontal contact with the maxilla as in other Pelomedusoides; only the lateral plate of the jugal does. The lateral plate contacts the postorbital dorsally, but the medial jugal process is present only ventrally, so the postorbital does not participate in the reduced septum orbitotemporale. This condition also occurs in Phosphatochelys and Azabbaremys (the latter entirely lacks the septum orbitotemporale).

Structures of medial process: The jugal medial process in Rhothonemys is well developed ventrally, but incomplete more dorsally, so that the upper half of the septum orbitotemporale is largely absent, in contrast to bothremydids outside the Taphrosphyini . The medial process of the jugal is even smaller in Phosphatochelys and Azabbaremys . It is present but poorly preserved in Taphrosphys and Labrostochelys . The fossa orbitalis in Rhothonemys has a deep ventral pocket that is defined posteroventrally by the medial jugal process, but posterodorsally it is likely that the fossa is open. Nonetheless, the medial process in Rhothonemys is higher and extends farther medially than it does in Phosphatochelys .

QUADRATOJUGAL (figs. 208–210)

Preservation: The position and identification of the quadratojugal in AMNH 30521 is problematic. The posterolateral part of the skull roof is mostly missing on the right side and is cracked and poorly preserved on the left. The best guess for the quadratojugal is a plate of bone on the left side lying lateral to the parietal on the posterior margin of the temporal roof. The medial margin is a suture with the parietal; the lateral margin is a broken edge connecting to the quadrate. The anterior margin is a large crack, with no clear sign of a sutural contact. If Rhothonemys is similar to Phosphatochelys , then this plate of bone is in the same place as the quadratojugal in Phosphatochelys . However, if Rhothonemys is like Azabbaremys , the plate of bone is the posterior continuation of the postorbital and the quadratojugal is a small element, largely lost in the crushed and broken area lateral to the postorbital. However, careful examination of this area does not show any sutural edges either, and there does not seem to be enough bone missing to make up a complete quadratojugal, hence the identifications in figure 210 and the restoration in figure 208. It is still just a guess, though.

Contacts and structures: If this plate of bone is the quadratojugal, it is similar in contacts to those in Phosphatochelys and Taphrosphys , contacting the parietal medially, the postorbital anteriorly, and the quadrate laterally. The quadratojugal forms the lateral part of the temporal margin.

SQUAMOSAL (figs. 208–210)

Preservation: Most of the left squamosal and a small portion of the right squamosal are present in AMNH 30521. The left squamosal lacks its dorsal and posterior margins. Only the part of the right squamosal immediately posterior to the quadrate is preserved, and this is broken by crushing.

Contacts: The squamosal in Rhothonemys contacts the quadrate anteriorly and anteromedially and the opisthotic medially. A possible anterodorsal contact with the quadratojugal is indeterminate due to missing bone.

Structures: The squamosal forms the cone-shaped outer bone for the antrum postoticum. In Rhothonemys the antrum is ‘‘moderate’’, similar in size to Phosphatochelys and Taphrosphys , larger than in Labrostochelys , and not absent, as in Azabbaremys . The posterior limit of the squamosal is a broken edge on both sides, but it does not seem to be trending into a very elongated shape like that in Labrostochelys . There is a well-developed vertical flange on the ventral surface of the squamosal, as in Phosphatochelys , Labrostochelys , and Taphrosphys . The lateral tubercle also seen in these three genera is not determinable in Rhothonemys , as the anterolateral part of the squamosal is broken off.

POSTORBITAL (figs. 208–210)

Preservation: The area of the skull roof in AMNH 30521 posterior to the orbits is poorly preserved on both sides. On the right side only broken pieces of bone in the matrix are present that seem to be postorbital fragments, and no sutures are visible. On the left side, the anterodorsal and anteroventral areas of the postorbital are preserved and defined by sutures, but they are separated from each other by matrix. Posteriorly the postorbital is not clearly defined by sutures except medially, where it lies along the parietal. Most of its lateral edge is a broken margin. Posteriorly, a crack separates the postorbital from what may be the quadratojugal (see Quadratojugal) or a continuation of the postorbital. Either interpretation is possible, although the former is slightly more likely.

Contacts of lateral plate: The postorbital in Rhothonemys contacts the parietal medially for its entire length and the jugal anterolaterally. Posteriorly it may contact the quadratojugal (see Quadratojugal). The postorbital lateral contacts are dubious, but it is probable that there was a quadrate contact, as in Phosphatochelys .

Structures of lateral plate: The postorbital in Rhothonemys forms that posterodorsal margin of the orbit. If the quadratojugal is present behind the postorbital, then the postorbital does not reach the temporal margin. If there is no suture, however, then the postorbital extends posteriorly to the temporal margin, as in Azabbaremys .

Contacts and structures of medial process: The postorbital in AMNH 30521 is not well enough preserved to determine all the features of the medial process. The ventral portion of the internal postorbital surface is preserved on the left side, and it shows that the medial process is absent; there is no medial contact with another bone. The septum orbitotemporale is small or absent.

PREMAXILLA (figs. 208–211)

Preservation: The premaxilla in AMNH 30521 is present on both sides, but both lack the posterior projection.

Contacts: The premaxilla in Rhothonemys contacts the maxilla posterolaterally and the other premaxilla medially. A posteri- or contact with a vomer is indeterminate.

Structures on dorsal surface: The premaxilla in Rhothonemys forms the ventral margin of the very large apertura narium externa. On the midline there is a dorsal projection of both premaxillae that forms a midline separation of the apertura. This internarial process has a broken dorsal margin on both sides, but the broken area seems too small to support a complete internarial bar, particularly on the left side. It is likely that the apertura narium externa in Rhothonemys was not divided on the midline.

On the anterior surface of the premaxilla, the bone slopes dorsally into the apertura narium externa, very similar to Phosphatochelys . This condition also occurs on the maxilla of Rhothonemys (see Maxilla), but it is not so apparent on the Phosphatochelys maxilla.

Structures on ventral surface: The labial ridge on the premaxilla in Rhothonemys is thick in comparison to most other Taphrosphyini (e.g., Phosphatochelys ), but it is not as thick as in Azabbaremys . The actual edge of the labial ridge, however, is acute, not blunt. The internal surface of the premaxilla in Rhothonemys is nearly vertical but slightly concave. This concave surface continues the concave surface on the maxilla. The external surface of the labial ridge is convex, curving anterodorsally to meet the low, horizontal ridge that marks the beginning of the indented surface surrounding the apertura narium externa. The medial horizontal process of the premaxilla in Rhothonemys is missing.

MAXILLA (figs. 208–211)

Preservation: Both maxillae are present in AMNH 30521, but both lack most of their horizontal plate.

Contacts of vertical plate: The vertical plate of the maxilla in Rhothonemys contacts the premaxilla anteromedially, the jugal posterodorsally, and the prefrontal anterodorsally. The posterior margin of the maxilla below the jugal is a free edge, so there is no quadratojugal or quadrate contact.

Structures of vertical plate: The orbital margin in Rhothonemys is an acute ridge, well dorsal to the floor of the fossa orbitalis, in contrast to the low, blunt orbit of forms like Bothremys cooki . The orbit is placed high on the cheek in Rhothonemys , so that the maxilla is unusually deep, deeper than in any other bothremydid, although Phosphatochelys and Azabbaremys approach this condition. The maxilla in Rhothonemys is relatively large and forms a greater part of the cheek than in any other bothremydid.

The maxilla forms the lateral margin of the apertura narium externa. This margin in Rhothonemys is very high and the apertura is very large, relatively larger than in any other bothremydid. The maxilla bears a slightly recessed area around the margin of the apertura narium externa, as on the premaxilla. There is a similar, indistinct recessed area around the apertura in Phosphatochelys , but not in other bothremydids. It is possible that this is the attachment area for a large proboscis, showing that Rhothonemys is really an elephant. The fossa nasalis is not completely preserved in Rhothonemys , but the internal surface of the maxilla shows the approximate extent. As expected from the larger apertura, the fossa nasalis is also large, although not significantly larger than in some other Taphrosphyini .

Contacts and structures of horizontal plate: Most of the horizontal or medial plate of the maxilla in AMNH 30521 is missing, but its lateral position on the vertical plate and some structures are preserved. The contacts are not preserved.

The labial ridge, actually formed more by the horizontal plate, is an acutely edged ridge that becomes thick dorsally and is slightly concave on its medial surface. The labial ridge in Rhothonemys is unique in shape among bothremydids; it is thicker than in Phosphatochelys , Labrostochelys , and Taphrosphys and has a different cross section than in Arenila and Nigeremys . The cross section with a broadly curved outer surface and a concave inner surface is not found in other pleurodires. The maxilla and premaxilla labial ridge morphologies are the same.

Although most of the lingual ridge is missing in AMNH 30521, a small portion is preserved on both sides. This shows a very low lingual ridge and a narrow triturating surface, broadly curved from labial to lingual ridge, similar to Labrostochelys . The surface, however, has a series of very low, dorsoventral ridges. The narrow triturating surface of the maxilla is consistent with the narrow lower jaw associated with the skull of AMNH 30521.

The fossa orbitalis in Rhothonemys (fig. 211) has a large pocket formed anterior and ventral to the orbital margin. This pocket, although smaller, also occurs in Phosphatochelys and Taphrosphys . Combined with the posterodorsal parietal pocket, the result is an unusually large fossa orbitalis in Rhothonemys , even larger than that in Chedighaii . At the posteroventral corner of the fossa orbitalis, just anterior to the jugal suture, is a large foramen leading into a canal in the maxilla. This seems to be the foramen supramaxillare, or at least the foramen identified as this in Gaffney (1979a: fig. 54).

VOMER AND PALATINE

Preservation: The vomer and the palatine are not preserved in AMNH 30521.

QUADRATE (figs. 208–210)

Preservation: Parts of both quadrates are present in AMNH 30521, but both are incomplete. The right quadrate has only the antrum postoticum portion of the cavum tympani present. The left quadrate has all of the central part of the cavum tympani, incisura columellae auris, antrum postoticum, and sulcus eustachii, but none of its margins. All of its dorsal surface is preserved, but only part of the ventral surface, and the entire processus articularis is missing.

Contacts on lateral surface: At least anterodorsally, the quadrate contacts the postorbital and the squamosal posterodorsally. Poor preservation makes the position of the quadratojugal unclear, but it is likely that the quadrate contacts it dorsally, as in Phosphatochelys . Because the maxilla ends posteriorly in a free edge, there is no quadrate-maxilla contact. The ventral part of the jugal is also a free edge, but there may have been a quadrate contact above it, as in Phosphatochelys .

Structures on lateral surface: The dorsal edge of the quadrate is a broken margin, and it looks as if the quadrate were not exposed on the temporal margin, but this is unclear.

The cavum tympani shows a centrally located incisura columellae auris that is a foramen leading medially into the canal typical of bothremydids. The sulcus eustachii is an indentation or groove in the posteroventral margin of the cavum tympani, as in Phosphatochelys and Azabbaremys and other Taphrosphyini . The antrum postoticum of Rhothonemys is similar in size to that in Phosphatochelys and Taphrosphys , not very small, as in Labrostochelys , or absent, as in Azabbaremys . The edges of the sulcus eustachii are damaged, but it looks as if the ventral process, as found in Phosphatochelys , Labrostochelys , and Taphrosphys , was present.

Contacts on dorsal and anterior surfaces: The quadrate contacts in AMNH 30521 are not entirely clear due to poor preservation and possible fusion. The quadrate contacts the opisthotic posteromedially and the squamosal posterolaterally. The area of the prootic is damaged (only the left side preserves this area), but it looks as if there is no quadrate-supraoccipital contact and that the prootic contacts the opisthotic. This would be consistent with other Taphrosphyini . The anterior surface of the quadrate is missing ventrally and is partially covered by matrix.

Structures on dorsal and anterior surfaces: The foramen stapedio-temporale and the foramen nervi trigemini in Rhothonemys are below the broken edge and are missing.

Contacts and structures on ventral surface: This area, containing the foramen posterius canalis carotici interni and condylus mandibularis as well as pterygoid, basisphenoid, and basioccipital contacts, is missing. The broken surface through this area is poorly preserved.

Contacts on posterior surface: The quadrate contacts the opisthotic dorsomedially, the exoccipital medially, and the squamosal dorsolaterally, but other contacts are not preserved.

Structures on posterior surface: AMNH 30521 is broken off below the level of the fenestra postotica. However, the quadrate makes up the lateral half of that foramen; it is not subdivided for the stapedial artery and lateral head vein. Lateral to the fenestra postotica is a posteroventrally facing concavity that seems to have contained a bend in the lateral head vein in Rhothonemys . This concavity is not seen in Phosphatochelys or other Taphrosphyini .

PTERYGOID

Preservation: It is likely that some of the pterygoid in AMNH 30521 is present in the broken surface of the otic chamber, but no sutures or structures can be made out.

SUPRAOCCIPITAL (figs. 208–210)

Preservation: Most of the supraoccipital in AMNH 30521 is probably present, but it is crushed and poorly preserved. The crista supraoccipitalis is preserved only at its base.

Contacts: The supraoccipital in Rhothonemys contacts the parietal anterodorsally. It seems to be exposed on the skull roof just behind the parietals. The supraoccipital contacts the exoccipitals posteroventrally and the opisthotics posterolaterally. Although not definite, the supraoccipital seems to contact the prootic anterolaterally with no quadrate contact.

Structures: The base of the crista supraoccipitalis in Rhothonemys , even though crushed, shows a plate that is lower in height than the one in Phosphatochelys . The foramen magnum is present but crushed, obscuring its shape.

EXOCCIPITAL (figs. 208–210)

Preservation: The dorsal part of the left exoccipital and pieces of the right one are preserved in AMNH 30521.

Contacts: The exoccipital in Rhothonemys contacts the supraoccipital dorsally and the opisthotic laterally. The other contacts are missing.

Structures: The left exoccipital in Rhothonemys preserves some of the foramen magnum margin and the dorsal half of the foramen jugulare posterius. The foramen is closed laterally and separated by bone from the fenestra postotica. The condylus occipitalis is missing.

BASIOCCIPITAL

Preservation: None of the basioccipital is preserved in AMNH 30521.

PROOTIC (fig. 210)

Preservation: The left prootic is present in AMNH 30521, but sutures defining it are questionable due to poor preservation and crushing in this area.

Contacts and structures: The prootic in Rhothonemys contacts the quadrate laterally and the supraoccipital posterodorsally, but the opisthotic contact is questionable. The ventral half is missing. The foramen stapedio-temporale, foramen nervi trigemini, and other structures are not preserved.

OPISTHOTIC (figs. 208–210)

Preservation: The left opisthotic in AMNH 30521 is present and only missing some of its posterior margin. The right one is preserved dorsally but broken posteriorly and ventrally.

Contacts: The opisthotic in Rhothonemys contacts the supraoccipital anteromedially, the quadrate anterolaterally, the squamosal posterolaterally, and the exoccipital posteromedially. There is probably an anterior contact with the prootic.

Structures: The opisthotic in Rhothonemys forms much of the roof of the fenestra postotica. The fenestra is not subdivided (see Quadrate). The opisthotic forms the lateral wall of the foramen jugulare posterius. There is no ventrally opening groove on the opisthotic, as in Taphrosphys .

BASISPHENOID

Preservation: The basisphenoid is not preserved in AMNH 30521.

Azabbaremys moragjonesi

The single skull that represents this species is a large, well-preserved, uncrushed specimen that shows internal morphology. It was described by Gaffney, Moody, and Walker (2001), which is modified and expanded here. Azabbaremys is the sister taxon to an undescribed skull, CNRST SUNY 199, and in turn they are sister taxa to the remaining Taphrosphyina , with Nigeremys and Arenila outside this group.

PREFRONTAL (figs. 212, 215, 280A)

Preservation: Both prefrontals in BMNH R16370 are present and complete. The sutures are clearly defined and the dorsal and ventral surfaces are visible.

Contacts: The prefrontal in Azabbaremys contacts the maxilla anteroventrolaterally, the frontal posteriorly, and the other prefrontal medially.

Structures: The prefrontal in Azabbaremys is a relatively large element, forming most of the preorbital part of the skull in dorsal view. In contrast to the small prefrontals of pelomedusids and Araripemys , Azabbaremys and most bothremydids have a broad preorbital area and a large prefrontal. The dorsal surface is convex, forming most of the distinctive dorsal protuberance of the nose found in Azabbaremys . The ventral surface is broadly concave, without a distinctly defined sulcus olfactorius.

The prefrontal of Azabbaremys forms the anterior margin of the fossa orbitalis, as in other Pelomedusoides; however, in Azabbaremys the prefrontal has a well-developed posteroventral portion that occupies the area formed by the maxilla in other Pelomedusoides. This part of the prefrontal is comparable in extent to Taphrosphys and possibly Arenila , although the sutures are ambiguous in the latter. Ventrally the prefrontal contacts the maxilla in a nearly horizontal suture at the level of the ventral margin of the orbit. The anterior margin of the prefrontal forms the dorsal margin of the apertura narium externa, which is protuberant in Azabbaremys so that the dorsal margin overhangs the apertura, much as in Taphrosphys . The protuberance is broad, extending the entire width of the apertura.

FRONTAL (figs. 212, 215)

Preservation: Both frontals in BMNH R16370 are present and complete. The sutures are clearly defined and the dorsal and ventral surfaces are visible.

Contacts: The frontal in Azabbaremys contacts the prefrontal anteriorly, the other frontal medially, the parietal posteriorly, and the postorbital posterolaterally. It forms the dorsal margin of the fossa orbitalis between the prefrontal and postorbital contacts.

Structures: The dorsal surface of the frontal in Azabbaremys is broadly convex, continuing the convexity formed by the prefrontal.

On the ventral surface the frontal forms a deep and well-developed sulcus olfactorius. The parasagittal ridge is low anteriorly, beginning as a continuation of the prefrontal ridge that divides the fossa orbitalis from the fossa nasalis. The sulcus olfactorius ridge deepens posteriorly as it merges into the processus inferior parietalis. The orbits of Azabbaremys are widely separated and the margins are almost vertical, a strong contrast to other bothremydids such as Bothremys and Rosasia , which have orbits that face primarily dorsally. Nigeremys and Arenila have orbits with a more lateral orientation, but they do not approach the Azabbaremys condition.

PARIETAL (figs. 212, 215, 217)

Preservation: Both parietals in BMNH R16370 are present and complete. Sutures are clear on the dorsal surface. The right parietal is visible ventrally, but the left one has matrix on its ventral surface.

Contacts of dorsal plate: The dorsal plate of the parietal in Azabbaremys contacts the frontal anteriorly, the other parietal medially, and the postorbital laterally.

Structures of dorsal plate: The posterior edge of the parietal in Azabbaremys forms part of the anterior margin of the temporal emargination. The margin is a nearly straightedge trending anterolaterally along the parietal and continuing onto the postorbital. The temporal roof in Azabbaremys is relatively extensive, with most of the otic chamber being covered in dorsal view. In comparison to other bothremydids, the roof in Azabbaremys is more extensive, although Nigeremys approaches it. Nonetheless, the size and shape of the parietal itself are quite similar in Azabbaremys and Taphrosphys .

Contacts of processus inferior parietalis: Posteriorly the parietal in Azabbaremys lies above the supraoccipital, as in other turtles. The prootic contacts the parietal from midway along the supraoccipital suture and extends anteroventrally to the foramen nervi trigemini. In Azabbaremys the foramen nervi trigemini is preserved but damaged on both sides. As preserved, the parietal narrowly enters the foramen margin dorsally. The pterygoid forms nearly all of the anterior and ventral margins, with the prootic forming the posterior margin. The complete extent of the foramen is not known posteriorly due to damage.

Structure of processus inferior parietalis: The processus inferior parietalis in Azabbaremys is present on both sides, with medial and lateral surfaces being visible. It is unusually narrow, in strong contrast to pelomedusids, podocnemidids, and most bothremydids, but in common with Taphrosphys . The condition in Arenila and Nigeremys is indeterminate. The parietal extends ventrally, forming the processus inferior parietalis, and meets the crista pterygoidea of the pterygoid in a suture that drops anteroventrally from the foramen nervi trigemini to the edge of the processus inferior parietalis (fig. 217). The anterior margin of the processus forms the posterior margin of the foramen interorbitale. As a consequence of the high, arched palate, the deep sulcus olfactorius, and the short snout, the foramen interorbitale is relatively small in Azabbaremys .

JUGAL (figs. 212, 215, 280A)

Preservation: Both jugals in BMNH R16370 are present and complete. The sutures are clearly defined and the internal and external surfaces are visible except for a small area on the inside of the left jugal.

Contacts of lateral plate: The external or vertical plate of the jugal in Azabbaremys contacts the postorbital dorsally, the quadratojugal posterodorsally, the quadrate posteriorly, and the maxilla ventrally. The quadrate contact is found throughout the Taphrosphyini .

Structures of lateral plate: The jugal forms the posteroventral margin of the orbit in Azabbaremys . It is exposed to a greater extent than the jugal in Nigeremys and Arenila ( Lapparent de Broin and Werner, 1998) .

Contacts of medial process: The medial process of the jugal in Azabbaremys forms part of the septum orbitotemporale and contacts the palatine medially and the maxilla ventrally.

Structures of medial process: As in other Taphrosphyini , the jugal does not extend onto the triturating surface in Azabbaremys . The jugal forms part of the floor of the fossa orbitalis, contacting the maxilla anteriorly and the palatine medially (fig. 280A).

QUADRATOJUGAL (figs. 212, 215)

Preservation: Both quadratojugals are present and complete in BMNH R16370, but both have some damage due to cracking, and the sutures are not all clear.

Contacts: The quadratojugal of Azabbaremys contacts the postorbital dorsomedially, the jugal anteroventrally, the squamosal posteriorly, and the quadrate ventrally. In both Phosphatochelys and Taphrosphys the quadratojugal has a parietal contact, absent in Azabbaremys .

Structures: In the Bothremydini and Cearachelyini the quadratojugal is a large element extending from the ventral margin of the cheek to the temporal emargination on the skull roof. In Azabbaremys and other Taphrosphyini the quadratojugal is smaller and retracted dorsally above the quadrate, not exposed on the cheek edge, but it still enters the edge of the temporal emargination. Rosasia also has a retracted quadratojugal with no exposure on the cheek edge. The ventral edge of the quadratojugal in Azabbaremys contacts the quadrate. Just dorsal to this suture the edge of the concave cavum tympani extends up onto the quadratojugal. It is unusual in turtles to have the quadratojugal form a significant portion of the cavum tympani, and this does not occur in any other pleurodire.

SQUAMOSAL (figs. 212, 215, 218, 281, 286A)

Preservation: Both squamosals are preserved in BMNH R16370, but they are missing parts of their posterolateral margins. The right squamosal is more complete posteriorly, but it has breakage on its anterolateral process.

Contacts: The squamosal in Azabbaremys is the usual cone-shaped (but apparently not hollow) element sitting on the posterolateral corner of the quadrate, as in most turtles. It contacts the opisthotic medially on the dorsal, posterior, and ventral surfaces. A short process of the squamosal contacts the quadratojugal along the lateral edge of the temporal emargination. All of these contacts are as in pelomedusids, podocnemidids, and other bothremydids.

Structures: The squamosal in Azabbaremys is cone-shaped and relatively smooth, not flat, as in Pelusios . Azabbaremys lacks the ventral, parasagittal flange or process of the squamosal characteristic of Taphrosphys . The antrum postoticum is absent in Azabbaremys , and there is no depression on the cavum tympani (figs. 281, 286A). The squamosal is presumably solid, not hollow, as in other turtles.

POSTORBITAL (figs. 212, 215)

Preservation: Both postorbitals in BMNH R16370 are present and complete. The sutures are clearly defined and the internal surface is visible on the right side.

Contacts of lateral plate: The postorbital in Azabbaremys is a large, quadrangular element contacting the frontal anteromedially, the parietal posteromedially, the jugal anterolaterally, and the quadratojugal posterolaterally.

Structures of lateral plate: The postorbital forms the posterior margin of the orbit and the anterior edge of the temporal emargination. In the other Taphrosphyini in which the postorbital is completely known ( Phosphatochelys , Taphrosphys , Labrostochelys ), the postorbital is shorter and does not reach the temporal margin.

Contacts and structures of medial process: In most Pelomedusoides, the postorbital has a medial process that contacts the jugal and palatine to form the septum orbitotemporale, which is particularly large in the Bothremydini . In Azabbaremys this wall is absent, and the medial and ventral surfaces of the postorbital are smooth. An absent or reduced septum orbitotemporale is also seen in Taphrosphys , Phosphatochelys , and Rhothonemys , but not in Arenila and Nigeremys .

PREMAXILLA (figs. 212–216)

Preservation: Both premaxillae are preserved in BMNH R16370, being complete and visible on all surfaces.

Contacts: The premaxilla in Azabbaremys contacts the maxilla posterolaterally, the vomer posteriorly, and the other premaxilla medially, as in the other Taphrosphyini .

Structures on dorsal surface: The dorsal surface of the premaxilla in Azabbaremys forms the ventral margin of the apertura narium externa and the floor of the fossa nasalis. The apertura has paired grooves at the front that lead posteriorly into the shallow choanal channels that run posterolaterally from the fossa nasalis. Within the fossa the premaxillae on the midline are nearly flat anteriorly, but they rise dorsally very quickly to form a high, median projection dividing the choanal channels and meeting the vomer.

Structures on ventral surface: The premaxilla in Azabbaremys forms a deep, acute labial ridge with a median ventral process forming a short hook. On the anterior surface this hook is continuous with a low ridge running on the midline dorsally to the ventral margin of the apertura narium externa. The ridge is unpaired, but at the ventral margin of the apertura, on either side of this ridge, are paired troughs that cut into the lower margin of the apertura narium externa. This premaxillary morphology is unique in turtles. The labial ridge on the premaxilla of Azabbaremys is relatively deep, in contrast to the shallower labial ridge in Nigeremys . Nigeremys also has a very wide and thick labial ridge with a blunt margin. In Azabbaremys and Taphrosphys the ridges are much thinner and more acute. The premaxillae are missing in Arenila .

The premaxillae in Azabbaremys form a relatively deep concavity on the midline just posterior to the labial ridge. Nigeremys also has a midline concavity that is hemispherical rather than triangular as in Azabbaremys . The posterior margins of the premaxillae enter the apertura narium interna in Azabbaremys , but they are excluded from it in Nigeremys and Arenila by a vomer-maxilla contact.

MAXILLA (figs. 212–216, 280A)

Preservation: Both maxillae are complete and visible on all surfaces in BMNH R16370.

Contacts of vertical plate: The maxilla in Azabbaremys in lateral view shows a relatively flat, deep, bladelike bone, forming most of the labial ridge, the ventral part of the fossa orbitalis, and the lateral part of the fossa nasalis. The prefrontal and premaxilla contacts are clear on both sides. The maxilla has a long, roughly horizontal suture, with the jugal behind the orbit. In more typical bothremydids such as Bothremys and Rosasia , the jugal contact is not as long.

In Azabbaremys the posterior end of the maxilla contacts the quadrate, as in the other Taphrosphyini and most of the Bothremydini . The maxilla-quadrate contact in Azabbaremys is broken slightly on both sides. On the right side it is slightly broken but not displaced. The suture on the external surface can be seen ventrally, beginning in a broken part of the cheek margin and extending dorsally into an area of some breakage with a slight amount of breakage in the sutural contact itself until it reaches the jugal. On the internal surface of the right side, the maxilla-quadrate suture is less disturbed and is altered only by a slight amount of overlap between the two bones. The left side is damaged by anterior-posterior pressure forcing the maxilla and quadrate past each other so that they overlap for about a centimeter. The break did not occur precisely in the suture between the two bones, and part of the maxilla-quadrate suture is clearly visible on the internal surface running dorsally from the ventral margin of the cheek.

Structures of vertical plate: The maxilla in Azabbaremys is completely smooth on its external surface; there are no nutrient foramina, as in most other turtles, and there is no ridge or bone texture change indicating the position of the horny rhamphotheca. The dorsal process is wider in Azabbaremys than it is in Phosphatochelys and Taphrosphys ; it is about the same as in Nigeremys and Arenila . The labial ridge has an acute edge and curves convex ventrally. The ridge is thinner than in Phosphatochelys , but thicker than in Labrostochelys and Taphrosphys .

Contacts of horizontal plate: The maxilla in Azabbaremys contacts the premaxilla anteromedially, the palatine posteromedially, and the jugal posterolaterally. There is no vomer contact as in Labrostochelys , Nigeremys , and Arenila .

Structures of horizontal plate: The horizontal plate of the maxilla makes up the triturating surfaces, the floor of the fossa orbitalis, and the floor of the fossa nasalis. The fossa nasalis in Azabbaremys is a large chamber divided posteriorly into paired choanal channels leading into the apertura narium interna. The floor of the fossa orbitalis in Azabbaremys lies above the choanal channel and is angled anterodorsally, forming part of the arched palate seen in Azabbaremys . Only a small part of the orbital floor is actually made up of maxilla; most of it is palatine and jugal.

The maxilla bears most of the triturating surface in Azabbaremys . It is roughly parallel-sided; the labial ridge is equidistant from the lingual margin, bordering the apertura narium interna throughout its length. There is a very small contribution of the palatine to the triturating surface posteromedially. The width of the triturating surface in Azabbaremys is narrower than in Nigeremys and Arenila , particularly anteriorly but comparable to Phosphatochelys . In Nigeremys and Arenila the maxilla is broad enough to contact the vomer and prevent narial exposure of the premaxilla. In Azabbaremys the maxilla does not contact the vomer and the premaxilla does enter the apertura narium interna. It is likely that in Taphrosphys the maxilla does not contact the vomer either. The triturating surface in Azabbaremys is unusual in being very rugose and formed by a series of rough corrugations with the shape of small teeth.

VOMER (figs. 212–216)

Preservation: The vomer was originally complete and well preserved in BMNH R16370. It may have been broken off and lost by the senior author during a misguided cleaning attempt. However, after heroic efforts by the BMNH Curator of Plumbing, it was not found down the drain. Fortunately, the cast of the skull preserves the vomer’s original shape and position, and a series of photos in the AMNH archives also record its morphology (fig. 216).

Contacts: The vomer in Azabbaremys contacts only the premaxillae anteriorly. The posterior contact with the palatines figured in Gaffney, Moody, and Walker (2001: figs. 2B, 3) is in error. It is possible but unlikely that such a contact was been eroded away by postmortem damage, but as preserved (fig. 216), there is only the premaxillae contact.

Structures: The vomer in Azabbaremys is a thin, curved element, extending dorsally and posteriorly from a high midline process formed by the premaxillae. The vomer shows that the palate in Azabbaremys was highly arched. The dorsal margin of the vomer has a narrow groove in it, the sulcus vomeri of cryptodires that bears the septum nasalis ( Gaffney, 1979a: 92). In Labrostochelys , Nigeremys , and Arenila the vomer is dumbbell-shaped in contrast to the thin columnar shape in Azabbaremys . In Arenila and Nigeremys the vomer is also unusually thick and robust. As preserved, the vomer in Azabbaremys is only attached anteriorly to the premaxillae. It tapers posteriorly and there is a gap between the posterior end of the vomer and the palatines. The palatines have an anterior margin that is thin and lacks an expanded area for the reception of the vomer. The anterior margin of the palatines and the posterior margin of the vomer are not finished edges, and a palatine contact could have existed.

PALATINE (figs. 212–215, 280A)

Preservation: Both palatines are preserved in BMNH R16370, but their posterolateral contacts are slightly damaged and not completely clear. All of the ventral surfaces are visible, but only some of the dorsal surfaces can be clearly determined.

Contacts: The palatine in Azabbaremys contacts the maxilla anterolaterally, the other palatine medially, the pterygoid posteriorly, and the jugal laterally. Despite the published figures ( Gaffney, Moody, and Walker, 2001: figs. 2B, 3) a vomer contact is not present as preserved. In the floor of the fossa orbitalis, the palatine contacts the maxilla anterolaterally and the jugal laterally. There is no postorbital contact, because the septum orbitotemporale is absent in Azabbaremys .

Structures on dorsal surface: The lateral portion of the dorsal surface of the palatine in Azabbaremys is complex and helps form a unique morphology. The absence of the septum orbitotemporale (see Postorbital, Jugal) considerably alters the usual pleurodire morphology in Azabbaremys . The palatine forms most of a ridge that extends between the inside of the cheek laterally and the base of the processus trochlearis pterygoidei medially. This ridge is really the ventral remnant of the septum orbitotemporale. The medial process of the jugal does not extend far enough medially to reach the pterygoid, as in most other pleurodires. Instead, the palatine is widely exposed in the posterior face of the septum orbitotemporale between the jugal and the pterygoid. The palatine forms most of the floor of the orbit. This area is preserved and visible on both sides of Azabbaremys , but the right side is damaged along the lateral palatine suture, and the left side is intact.

The palatine in the orbital floor of Azabbaremys is not a flat element. It is high medially and posteriorly. It slopes ventrally, forming a concavity deepest at the jugal and maxilla sutures. Posteromedially the palatine forms a dorsal process, unknown in any other pleurodire, just anterior to the contact of the pterygoid at the base of the processus trochlearis pterygoidei. This process marks the anterolateral limit of the sulcus palatinopterygoideus, a structure that in Azabbaremys lacks the firm base seen in other pleurodires, due to the absence of the septum orbitotemporale.

Structures on ventral surface: Only a small part of the palatine enters the triturating surface in Azabbaremys , about the same as in Arenila and Nigeremys . The palatine extends from a free edge anteriorly, the margin of the apertura narium interna (actually the ill-defined foramen orbitonasale), to the posterior free edge on the margin of the fossa temporalis. In nearly all other pleurodires (except some other Taphrosphyini ) there is a septum orbitotemporale rather than a free margin posteriorly. Laterally the palatine contacts a short process of the jugal posterolaterally and a short process of the maxilla anterolaterally. Medially the palatine reaches the other palatine. The palatine and descending process of the prefrontal do not meet in the anteroventral part of the orbital floor; they are separated by the maxilla.

QUADRATE (figs. 212–216, 218, 281C, 286A)

Preservation: Both quadrates are present and well preserved in BMNH R16370.

Contacts on lateral surface: In lateral view the quadrate in Azabbaremys contacts the maxilla anteroventrally, the jugal anterodorsally, the quadratojugal dorsally, and the squamosal posterodorsally.

Structures on lateral surface: The cavum tympani in Azabbaremys has a completely enclosed incisura columellae auris, as in Nigeremys and Arenila . The cavum tympani is nearly circular, with a distinct notch in its posteroventral edge, presumably for the eustachian tube. The incisura columellae auris in Azabbaremys is a small round hole for the stapes. On the right side of the skull a remnant of the stapes is present. The cavum tympani in Azabbaremys is completely smooth in the area of the antrum postoticum. Neither Arenila nor Nigeremys has an antrum postoticum. Remaining Taphrosphyini do have an antrum postoticum.

The quadrate of Azabbaremys has a step or shelf at the ventral margin of the cavum tympani, just above the flat, vertical sheet forming the ventral margin of the quadrate. This shelf (fig. 286A) seems to be in Nigeremys also, although the area is not well preserved. It is indeterminate in Arenila . The quadrate ventral margin is a continuation of the ventral margin of the maxilla, resulting in a lateral profile for Azabbaremys that completely lacks any suggestion of a cheek emargination. Nigeremys has a shallow notch, possibly due to postmortem damage, in the cheek area. Arenila is missing bone in this region.

Contacts on dorsal and anterior surface: The quadrate in Azabbaremys contacts the prootic anteromedially, the opisthotic posteromedially, and the squamosal posteriorly. There is no supraoccipital contact, consistent with other Taphrosphyini , but in contrast to most other Bothremydidae .

Structures on dorsal and anterior surface: On the anterior face of the otic chamber the quadrate-pterygoid suture is visible on both sides but is best seen on the right. As in other turtles, the quadrate forms the lateral margin of the foramen stapedio-temporalis. The margins of the foramen are eroded on both sides of Azabbaremys , but it is best preserved on the left side. The canalis stapedio-temporalis is free of matrix on the right side, so that the aditus canalis stapedio-temporalis can be clearly seen, as can the foramen and canalis. As in other bothremydids, the foramen stapedio-temporalis in Azabbaremys opens on the anterior face of the otic chamber, not more dorsally as in other pleurodires. The quadrate forms the ventral margin of the foramen stapedio-temporale and the ventral and anterior portions of the canalis cavernosus as it extends medially from the region just beneath the canalis stapedio-temporalis.

In Azabbaremys the region between the foramen stapedio-temporale and the processus inferior parietalis is eroded and missing some of the thin bone that covers the canalis cavernosus and forms the margins of the foramen nervi trigemini. The ventral portion of the more lateral part of this area is formed by the quadrate.

Contacts on ventral surface: In ventral view the quadrate of Azabbaremys forms part of the very low but distinct tuberculum basioccipitale with the basioccipital. The paired tuberculum is essentially absent in Nigeremys and, apparently, in Arenila . The quadrate has a very narrow contact with the basisphenoid between the broader contacts with the basioccipital and pterygoid. The quadrate contacts the quadrate ramus of the pterygoid in a suture extending from the basisphenoid along the processus articularis of the quadrate, as in other pleurodires.

Structures on ventral surface: The foramen posterius canalis carotici interni in Azabbaremys is similar in position to that in Phosphatochelys , in contrast to the pterygoidbasisphenoid-quadrate position of Taphrosphys and the quadrate-only position in Labrostochelys . The foramen posterius canalis carotici interni (see Pterygoid) is formed almost entirely by the pterygoid and quadrate, with a very narrow extension of basisphenoid exposed in the roof and margin of the foramen. Although it is arguable, character 74, position of the foramen posterius canalis carotici interni, has been coded as formed by the pterygoid-quadrate for Azabbaremys , even though the basisphenoid does enter the margin very narrowly.

The condylus mandibularis in Azabbaremys lies anterior to the condylus occipitalis, as in Phosphatochelys , but not as far anterior as in Taphrosphys . The fossa pterygoidea is absent in Azabbaremys ; there is only a low ridge marking the attachment margin of the pterygoideus muscle.

Contacts and structures on posterior surface: Posteriorly in Azabbaremys the cone-shaped squamosal fits onto the posterodorsal part of the quadrate. The medial part of the quadrate meets the other braincase elements and forms structures enclosing the associated soft parts (fig. 218). The quadrate and opisthotic combine to form the ovoid fenestra postotica containing the stapedial artery and lateral head vein. In some forms (such as Taphrosphys ) the fenestra is subdivided around those structures, but in Azabbaremys this is not the case. However, the fenestra postotica is widely separated from the foramen jugulare posterius by a well-developed opisthotic-quadrate contact medial to the fenestra postotica. Also, in Nigeremys the fenestra postotica is an ovoid foramen and it is widely separated from the foramen jugulare posterius. In Azabbaremys the quadrate contacts the exoccipital and the basioccipital ventral to the foramen jugulare posterius on the occipital surface.

PTERYGOID (figs. 212–215, 218)

Preservation: Most of both pterygoids are present in BMNH R16370, but the processus trochlearis pterygoidei is broken on both. On the right side the processus is missing entirely, but on the left side the main body is present but displaced anteriorly. The distal portions of the processus are missing on the left side also. The dorsal structures of the pterygoid are visible but so affected by damage that some areas are missing. Some of the dorsal surface sutures are unclear, although all of the ones on the ventral surface can be seen.

Contacts on ventral surface: In ventral view the pterygoid contacts are as in other bothremydids: palatine anteriorly, quadrate posterolaterally, basisphenoid posteromedially, and other pterygoid medially. The palatine contact is not transverse, as in Arenila and Taphrosphys , but is curved and concave anteriorly. The midline pterygoid contact is relatively short compared to Taphrosphys , but Arenila has the pterygoids completely or nearly separated on the midline.

Structures on ventral surface: The foramen posterius canalis carotici interni in Azabbaremys lies at the contact of pterygoid and quadrate (see Quadrate) at the posterior edge of the pterygoid. The anterior margin of the foramen is formed by a C-shaped indentation of the pterygoid, while the quadrate and a small extension of the basisphenoid are narrow elongations exposed in the roof and the posterior edge of the foramen. The pterygoid underlies broader exposures of the quadrate and basisphenoid. When compared with the much simpler triple meeting of these three bones in Taphrosphys , it seems as if the pterygoid of Azabbaremys had been dragged posteriorly over the other two bones, pulling the foramen with it.

Anterior to the foramen in Azabbaremys is a shallow concavity formed mostly by the pterygoid, with an anteromedial margin that extends right across the pterygoid. This is the scar for the M. pterygoideus, being very shallow in Azabbaremys . In Arenila and Nigeremys this is a concavity, the fossa pterygoidea, much deeper and fully defined, with posteromedial walls completely lacking in Azabbaremys .

The processus trochlearis pterygoidei in Azabbaremys is preserved only on the left side and its distal margins are missing. Its base has been broken and the process displaced dorsally and anteriorly. As preserved, the processus is not extensive; it would appear to be similar in shape and extent to that in the living pelomedusids and smaller than in the podocnemidids. The orientation of the processus trochlearis pterygoidei in Azabbaremys is posterolateral, rather than mostly lateral as in many podocnemidids. The Azabbaremys processus differs from Pelomedusidae in having a ridge along its ventrolateral edge, rather than being curved. The processus trochlearis pterygoidei of Nigeremys is present on both sides but badly damaged. Nonetheless, it is consistent with what is known in Azabbaremys .

The foramen palatinum posterius is formed in the pterygoid-palatine suture in Azabbaremys , as in most bothremydids. It is preserved only on the left side and faces ventrolaterally rather than ventrally, at least partially as a result of postmortem deformation. However, the arching of the palate in Azabbaremys is somewhat greater than in Nigeremys and Arenila , so that the curve from the palate around the base of the processus trochlearis pterygoidei is more pronounced. Also, the foramen palatinum posterius is located more laterally than in Nigeremys and Arenila , resulting in a foramen that opens more laterally into the fossa temporalis in Azabbaremys than in the others.

Contacts on dorsal surface: The anteri- or contacts of the pterygoid at the base of the processus trochlearis pterygoidei in Azabbaremys can be seen on the left side. The pterygoid-palatine contact wraps around the base of the processus from the foramen palatinum posterius dorsally to the edge of the sulcus palatinopterygoideus. The pterygoid is clearly separated from the medial process of the jugal by the palatine. This is an unusual condition; in nearly all pleurodires the jugal contacts the pterygoid in this area. The septum orbitotemporale that is formed by medial processes of the postorbital and jugal in other pleurodires is absent in Azabbaremys and small in all other Taphrosphyini except Arenila and Nigeremys . The enlarged palatine forms the only remaining anterior support for the base of the processus trochlearis pterygoidei in Azabbaremys .

Structures on dorsal surface: The dorsal surface of the pterygoid bears the crista pterygoidea and forms the floor of the canalis cavernosus and foramen nervi trigemini. The crista pterygoidea is preserved and visible on both sides in Azabbaremys . It rises above the level of the pterygoid plate and meets the processus inferior parietalis, making up the lower third of the braincase wall (fig. 217). The crista pterygoidea plus processus inferior parietalis are relatively short in comparison with such forms as in Podocnemis . The foramen nervi trigemini has been eroded along its margins as well as the anterior margin of the processus inferior parietalis. However, the damage is not enough to significantly narrow this structure.

The bones, prootic plus pterygoid, making up the anterior wall of the canalis cavernosus at the anterior edge of the otic chamber are eroded, opening up the canalis on both sides. Nonetheless, the relative positions of the foramen nervi trigemini and the foramen stapedio-temporale can be determined. These foramina are relatively close to each other, as in other bothremydids.

SUPRAOCCIPITAL (figs. 212, 215, 217)

Preservation: The supraoccipital is present and nearly complete in BMNH R16370, but only the right side is visible; the left side is still covered with matrix.

Contacts: The median section of the supraoccipital underlies the two parietals along the midline. Their mutual contact slopes anteroventrally. The laterally projecting otic portion of the supraoccipital in Azabbaremys contacts the prootic anterolaterally, the opisthotic laterally, and the exoccipital posterolaterally. As in the other Taphrosphyini , there is no supraoccipital-quadrate contact.

Structures: The crista supraoccipitalis in Azabbaremys is relatively short, comparable to that in living pelomedusids, but slightly longer than in Taphrosphys . Although incomplete, the crista in Arenila seems to be longer than in Azabbaremys , compared to the position of the condylus occipitalis. The crista supraoccipitalis in Nigeremys is similar in length to Azabbaremys but seems to be incomplete. The crista supraoccipitalis in Azabbaremys is deeper anteriorly and narrows to an acute posterior end. In Taphrosphys the end of the crista is curved and blunt. The very end of the crista is broken off in Azabbaremys , but it is already very narrow and is probably missing only a small section.

EXOCCIPITAL (figs. 212, 215, 217, 218)

Preservation: Both exoccipitals are preserved in BMNH R16370; they are complete, but only the right one is entirely free of matrix. Both have clear sutures.

Contacts: The exoccipital in Azabbaremys contacts the supraoccipital dorsally, the opisthotic laterally, and the basioccipital ventrally and ventrolaterally. There is also a contact with the quadrate ventrolaterally between the basioccipital and opisthotic. This quadrate contact is found in all bothremydids and is absent in all other Pelomedusoides.

Structures: The exoccipital in Azabbaremys forms all of the condylus occipitalis; the basioccipital enters the neck of the condyle. The exoccipitals make up the condyle in all of the described bothremydids. However, this cannot be substantiated in either Nigeremys or Arenila due to poor preservation, but Lapparent de Broin and Werner (1998) described both taxa as having this feature. In Azabbaremys the exoccipitals are eroded on the midline, giving the condyle a bilobed appearance. The foramen jugulare posterius is formed almost entirely by the exoccipital, with a narrow section of opisthotic entering the foramen dorsolaterally. The foramen is entirely closed by bone, as in Taphrosphys , Arenila , and Bothremys , in contrast to the open condition in Foxemys . Between the foramen jugulare posterius and the condylus occipitalis are the two foramina nervi hypoglossi.

BASIOCCIPITAL (figs. 212–215, 217, 218)

Preservation: The basioccipital in BMNH R16370 is complete and clearly defined.

Contacts: The basioccipital is a wide but very short, triangular element in Azabbaremys . It makes up the medial half of the tuberculum basioccipitale along with the quadrate laterally. Its entire anterior margin is a transverse contact with the basisphenoid. Posteriorly and dorsally, the basioccipital contacts the exoccipitals, reaching only the base of the condylus occipitalis.

Structures: Between the paired tubercula basioccipitale is a shallow median concavity that extends slightly onto the basisphenoid. The basioccipital in Nigeremys is not clearly defined by sutures. This element in Taphrosphys is much larger and longer and is just as wide. Arenila has a longer basioccipital also, and its anterior contact with the basisphenoid is curved and concave posteriorly.

PROOTIC (figs. 212, 215)

Preservation: Both prootics are present and visible in BMNH R16370, although they are eroded in the area of the canalis cavernosus and the sutures are not always clear.

Contacts: The prootic in Azabbaremys contacts the supraoccipital posteromedially, the parietal medially, the pterygoid ventrally, the quadrate laterally, and the opisthotic posterolaterally. The prootic forms the medial margin of the foramen nervi trigemini, the parietal forms the anterodorsal corner, and the pterygoid forms the anterior and ventral margins.

Structures: The foramen nervi trigemini as preserved is incomplete on both sides due to erosion of the medial margin exposing the canalis cavernosus. The prootic-pterygoid suture is gone in this area, so its position and the relative amount of prootic versus pterygoid contribution to the margin of the foramen nervi trigemini is not determinable. The foramen stapedio-temporale is formed in the prootic-quadrate suture and opens directly anteriorly as in most bothremydids. Although much of the bone between the foramen stapedio-temporale and the foramen nervi trigemini is eroded away, it can be seen that these structures were relatively close together, as in Taphrosphys and Bothremys .

OPISTHOTIC (figs. 212, 215, 218)

Preservation: The opisthotic is preserved completely on both sides of BMNH R16370. It is clearly defined and visible except on the dorsal area of the left side where it is covered with matrix.

Contacts: The opisthotic in Azabbaremys has the usual contacts for bothremydids: supraoccipital dorsomedially, prootic anteromedially, quadrate anterolaterally, squamosal posterolaterally, quadrate (again) ventrolaterally, and exoccipital posteromedially.

Structures: The opisthotic in Azabbaremys forms the roof of the fenestra postotica; the ventral half of the fenestra is formed by the quadrate (fig. 218). The fenestra is an oblong oval, presumably with the stapedial artery lying in the upper part and the lateral head vein in the lower part. The processus interfenestralis of the opisthotic forms the relatively thick lateral wall of the foramen jugulare posterius, contacting the exoccipital ventromedially and the quadrate ventrolaterally. A foramen, probably the foramen externum nervi glossopharyngei, penetrates the middle of the processus interfenestralis. Above this foramen is a much smaller foramen, which could alternatively be interpreted as the foramen nervi glossopharyngei. Neither can be probed, so their identification is in doubt. Neither of these foramina is found in Bothremys .

BASISPHENOID (figs. 212–215)

Preservation: The basisphenoid is present and clearly defined in BMNH R16370.

Contacts and structures on ventral surface: The basisphenoid in Azabbaremys has the triangular shape typical for many bothremydids: wider than long, with a straight transverse posterior suture with the basioccipital, rather than the curved suture seen in Arenila . The lateral contact with the quadrate is relatively narrow. Anterolaterally the basisphenoid very narrowly enters the margin of the foramen posterius canalis carotici interni (see Pterygoid, Quadrate). The anterior contacts of the basisphenoid are with the pterygoids.

Contacts and structures on dorsal surface: The dorsal surface of the basisphenoid in Azabbaremys can be seen inside the nearly cleaned out cavum cranii. An endocast has been made to facilitate this study (fig. 217). The dorsum sellae is relatively high, compared to living Pelomedusoides and Bothremys . A well-developed processus clinoideus rises on each side of the dorsum, also in contrast to the lower processus clinoideus of Bothremys and living Pelomedusoides. The dorsum sellae in Azabbaremys does not overhang the sella turcica, also in contrast to the above taxa. The sella turcica is shorter due to the very short rostrum basisphenoidale, in contrast to the longer sella and very long rostrum in Bothremys and the living Pelomedusoides. The rostrum basisphenoidale in Azabbaremys is very short, with a deep concave anterior face. On either side thin walls may represent ossified portions of the trabeculae. The entire dorsal basisphenoid morphology in Azabbaremys is a foreshortened version of that seen in other Pelomedusoides. This is consistent with the ventral morphology, which also shows a foreshortened basisphenoid.

SUBTRIBE NIGEREMYDINA

Nigeremys gigantea

The single skull representing this taxon is nearly complete but has a very badly damaged surface, making sutures difficult or impossible to see. This skull has been described briefly by Bergounioux and Crouzel (1968), but only Lapparent de Broin and Werner (1998) actually have a useful description. Nigeremys is the sister taxon to the similar Arenila , together making up the Nigeremydina , which is the sister taxon to the Taphrosphyina .

PREFRONTAL (figs. 219, 222)

Preservation: Both prefrontals are present in MNHN (P) NIR 1, in which the surface is damaged by erosion and pitting, but little bone has been lost, and the marrow cavity is not exposed. The original shapes of the bones seem intact. Sutures are clear but are partially obscured by being drawn over in ink.

Contacts: The usual bothremydid contacts are present in Nigeremys : frontal posteriorly, maxilla anteroventrolaterally, and prefrontal medially. The contacts are similar to those in Azabbaremys , except that the prefrontal-maxilla contact is more dorsal in Nigeremys than in Azabbaremys . Only the posterior part of the prefrontal is present in Arenila and it agrees with Nigeremys .

Structures: The prefrontal forms a protruding dorsal margin for the apertura narium externa in Nigeremys , which is shaped like that in Azabbaremys . In Nigeremys the prefrontal does not extend over the premaxilla in dorsal view as it does in Azabbaremys . The prefrontal forms the anterior wall of the fossa orbitalis, but Nigeremys has less of the bone exposed in the orbital margin than in Arenila and Azabbaremys . The prefrontal exposure in the fossa orbitalis is obscured by matrix and breakage in both orbits of Nigeremys .

Although Bergounioux and Crouzel (1968) described paired nasal bones in ‘‘ Potamochelys ’’, neither Lapparent de Broin and Werner (1998) nor we have found any evidence for their presence in MNHN (P) NIR 1.

FRONTAL (figs. 219, 222)

Preservation: Both frontals are present in MNHN (P) NIR 1, in which the surface is damaged by erosion and pitting, but little bone has been lost. There is no evidence of deformation. Most of the sutures are clear but partially obscured by being drawn over in ink. The postorbital contacts are damaged enough to make their exact position unclear, but enough is preserved to show that there is a postorbital contact.

Contacts: Nigeremys has the usual bothremydid contacts: prefrontal anteriorly, postorbital posterolaterally, parietal posteriorly, and other frontal medially. There is no prefrontal-postorbital contact preventing frontal exposure in the orbit, as shown in Bergounioux and Crouzel (1968: fig. 1). The frontal-parietal suture is straight, not curved. The frontal-prefrontal suture does not protrude anteriorly on the midline as much in Nigeremys as it does in Arenila , and its midline length is slightly less in Nigeremys than in Arenila .

Structures: The frontal forms the dorsal part of the orbital margin in Nigeremys , as in Arenila . The internal surface is covered by matrix.

PARIETAL (figs. 219, 222)

Preservation: Both parietals are present in MNHN (P) NIR 1, in which the surface is damaged by erosion and pitting. The posterior margin of the left parietal appears intact, but some bone is broken from the right parietal. The processus inferior parietalis is not visible due to matrix. The sutures are clear.

Contacts of dorsal plate: As in other bothremydids, the parietal of Nigeremys contacts the frontal anteriorly, the postorbital laterally, and the other parietal medially. The possibility of a posterolateral contact with the quadratojugal as in Taphrosphys and Phosphatochelys cannot be ruled out. The contact with the supraoccipital is at the posteromedial margin of the parietal.

Structures of dorsal plate: The posterior temporal emargination in Nigeremys completely exposes the otic chamber and is more emarginate than in Azabbaremys , Phosphatochelys , and Taphrosphys . The margin on the parietal is straight, as in Azabbaremys . Arenila could have had a very similar emargination to Nigeremys , but it is missing much of the edge.

Contacts and structures of processus inferior parietalis: The parietal in Nigeremys overlies the supraoccipital posteriorly, but the suture is unclear and none of the processus inferior parietalis is exposed.

JUGAL (figs. 219, 222)

Preservation: The area of the jugal is badly damaged with no clear sutures in MNHN (P) NIR 1. The anterior part of the jugal-postorbital contact and the jugal-maxilla contact can be seen, but we are unable to confirm the jugal contacts of either Bergounioux and Crouzel (1968) or Lapparent de Broin and Werner (1998). The latter authors showed a large jugal extending from the orbital margin to the temporal emargination, a condition unknown in any other pleurodire. This area is badly damaged by cracking and overpreparation, with a number of cracks expanded by overzealous incising of presumed sutures. Nonetheless, careful examination of the supposed jugal sutures of Lapparent de Broin and Werner (1998) has failed to corroborate them. The posterior limits of the jugal in Nigeremys are not determinable in our interpretation.

Contacts of the lateral plate: On both sides of MNHN (P) NIR 1, just below the orbital margin, is a contact with the maxilla. On the right side, more dorsally in the orbital margin, is the postorbital contact. Both of these become indistinguishable posteriorly. These sutures in Nigeremys are in about the same position in Arenila and in Azabbaremys .

Structures of the lateral plate: The jugal in Nigeremys enters the orbital margin; whether it entered the cheek margin is unknown.

Contacts and structures of the medial process: We have been unable to identify sutures for the medial process of the jugal in MNHN (P) NIR 1, although the area is preserved in the lower temporal fossa on both sides. Lapparent de Broin and Werner (1998) did not identify any jugal in ventral view.

QUADRATOJUGAL

Preservation: Although both Bergounioux and Crouzel (1968) and Lapparent de Broin and Werner (1998) identified a quadratojugal in MNHN (P) NIR 1, we have been unable to confirm any sutures defining this element. The cheek area where Lapparent de Broin and Werner (1998) showed quadratojugal sutures certainly has deeply incised fissures in these positions. However, microscopic examination of these cracks has revealed only broken bone fragments and glue, and they have the appearance of mechanical abrasion. It is possible that some of these fissures represent natural sutures, but this cannot be corroborated by us. It is possible that a maxilla-quadrate contact was present on the cheek in Nigeremys , and that the quadratojugal was placed dorsally, as in Azabbaremys .

SQUAMOSAL (figs. 219, 222)

Preservation: Both squamosals are present in MNHN (P) NIR 1, but the surfaces are damaged and only a few sutures are visible. The posterior surface is eroded and little if any of the original surface is preserved.

Contacts: As in other bothremydids, the squamosal of Nigeremys contacts the quadrate anteriorly and the opisthotic medially. The quadratojugal contact is not distinguishable.

Structures: The squamosal of Nigeremys is not well preserved, but it is similar in size and position to the squamosal in Azabbaremys . There is no ventral flange as in Taphrosphys . The squamosal in Arenila is poorly preserved also, but it seems to be longer than the one in Nigeremys .

POSTORBITAL (figs. 219, 222)

Preservation: The postorbital is present on both sides of MNHN (P) NIR 1, but its defining sutures are obscured by poor preservation and overpreparation. The parietalpostorbital suture and the anterior part of the jugal-postorbital are clear, but the more posterolateral sutures are ambiguous.

Structures: The postorbital of Nigeremys seems to extend posteriorly to reach the posterior temporal emargination, as in Azabbaremys and in contrast to Phosphatochelys . However, the bone is too poorly preserved to be sure that the quadratojugal did not contact the parietal. Lapparent de Broin and Werner (1998: fig. 4f) showed an elongate postorbital with a parasagittal jugal suture. We have been unable to confirm this suture.

PREMAXILLA (figs. 219–222)

Preservation: Both premaxillae are present in MNHN (P) NIR 1, but the sutures are unclear and are determinable only near the labial ridge. The bone surface is eroded and pitted, but it seems to be close in shape to the original bone. The dorsal end of the dorsal process is broken in both, and there is a hole in the midline concavity on the triturating surface.

Contacts: As in other bothremydids the premaxilla contacts in Nigeremys are with the vomer posteriorly, the maxilla posterolaterally (visible only on the labial ridge and near the vomer on the left side), and with the other premaxilla on the midline (visible only in a few places). As in Arenila , Nigeremys has an unusually long premaxilla, so the midline premaxilla contact and the posterolateral maxilla contact are unusually long.

Structures in dorsal view: Very little of the dorsal surface of the premaxilla is visible in MNHN (P) NIR 1 due to matrix and breakage. The midline dorsal process can be seen, and it is large and broad, as in Arenila but in contrast to other Taphrosphyini . The dorsal end of the process is broken off.

Structures in ventral view: The labial ridge on the premaxilla in Nigeremys is very thick, as in Azabbaremys . However, in Nigeremys the ridge is blunt in contrast to acute in Azabbaremys . It is possible that this is due to weathering, but the relatively even nature of the ridge suggests it is naturally blunt even though very little original bone surface is present. Azabbaremys has a very shallow, parasagittal groove extending vertically on the anterior face of the premaxilla (see Azabbaremys, Premaxilla ); this groove is absent in Nigeremys . The anterior face of the premaxilla in Nigeremys is not as deep as in Azabbaremys .

The midline concavity formed by the premaxilla is very deep in Nigeremys , as it is in Arenila but in contrast to all other Taphrosphyini . As described by Lapparent de Broin and Werner (1998), the foramen praepalatinum is formed entirely by the premaxilla, as in Arenila and the other Taphrosphyini in which it is determinable.

MAXILLA (figs. 219–222)

Preservation: Both maxillae are present and nearly complete, but as with the other bones in MNHN (P) NIR 1, the maxilla surfaces are badly damaged by erosion and pitting, at least partly the result of overpreparation. The sutures are not clear, and we have not been able to confirm many of the contacts described in Lapparent de Broin and Werner (1998) (see below).

Contacts of vertical plate: In MNHN (P) NIR 1 the prefrontal contact is visible on both sides and the anterior part of the jugal suture. However, microscopic examination of the cheek does not confirm the jugal and quadratojugal contacts figured by Lapparent de Broin and Werner (1998: fig. 4f). There are grooves and fissures, but these seem to be natural cracks enlarged mechanically. It is quite possible that the quadratojugal did not contact the maxilla, and that there was a maxilla-quadrate contact, as in Azabbaremys and other Taphrosphyini .

Structures of vertical plate: The dorsal process of the maxilla in Nigeremys differs from Arenila and Azabbaremys in having a prefrontal contact that is inclined anteroventrally rather than posteroventrally, but the thickness of the process is similar in all three, and in Azabbaremys the maxilla forms the lateral margin of an apertura narium externa that is about as wide as it is in Arenila and narrower than in Azabbaremys .

The suborbital depth of the maxilla in Nigeremys is greater than in Arenila , but it is about the same as in Azabbaremys . Arenila has possibly been made shallower by damage (see Arenila , Maxilla). There is a slight emargination along the cheek margin in Nigeremys on the left side, and most of the right cheek margin is missing. This cheek emargination seems to have a broken edge, so it probably is not original.

Contacts of the horizontal plate: The premaxilla suture in Nigeremys is visible in a few places (see Premaxilla) and the vomer suture can be seen for both maxilla. Arenila also has a vomer-maxilla contact via a short median process of the maxilla, and both are absent in Azabbaremys . These sutures and the palatine sutures have been described and figured by Lapparent de Broin and Werner (1998: fig. 4b) and we can confirm them. The jugal suture, presumably exposed in the fossa temporalis inferior wall, is not discernable.

Structures of the horizontal plate: The triturating surface in Nigeremys has a very thick, blunt labial ridge as described and figured in Lapparent de Broin and Werner (1998) and Bergounioux and Crouzel (1968). This agrees with the very thick labial ridge of Arenila . Azabbaremys is also thick in comparison to other Taphrosphyini , but not as thick as in Nigeremys and Arenila . Nigeremys has a blunt margin, possibly due to postmortem erosion (see Premaxilla), in contrast to the acute edge seen in Arenila and Azabbaremys . The triturating surfaces of Nigeremys and Arenila are wider anteriorly due to a narrow medial process reaching the vomer. In both a low, thick lingual ridge is separated from the much higher labial ridge by a shallow trough. Lingual and labial ridges are parallel. Lapparent de Broin and Werner (1998) stated that the maxillary triturating surfaces are wider in Arenila than in Nigeremys , but our own measurements of the triturating surface at the position of the apertura narium interna (appendix 5) divided by the total length (partially restored for Arenila ) results in a ratio of 4.0 for Arenila and 3.8 for Nigeremys , suggesting that they are both nearly the same.

The apertura narium interna in Nigeremys is very similar in size and position to that in Arenila . The apertura is more triangular in Azabbaremys and oval in Nigeremys .

VOMER (figs. 219–222)

Preservation: The vomer in MNHN (P) NIR 1 is nearly complete, but with the usual surface damage seen in this specimen. The sutures are relatively clear except on the right anterior edge where there is some breakage.

Contacts: Nigeremys has the vomer contacts seen in Arenila : premaxilla anteriorly, maxilla anterolaterally, and palatine posteriorly.

Structures: As in Arenila , the vomer of Nigeremys is unusually thick and wide, in contrast to Azabbaremys and Labrostochelys , the only other Taphrosphyini in which it is known. Palatal arching is much less in Nigeremys and Arenila than in Azabbaremys (see Arenila, Vomer ). The apertura narium interna of Nigeremys is about the same size as in Arenila (see Arenila, Vomer ).

PALATINE (figs. 219–222)

Preservation: Both palatines are present in MNHN (P) NIR 1, but they are damaged by surface erosion and pitting. The anterolateral sutures are clear, but the posterior and lateral ones are not. No dorsal surface morphology is visible due to matrix.

Contacts: Palatine sutures in MNHN (P) NIR 1 are not clear despite the description and figure in Lapparent de Broin and Werner (1998: fig. 4b), but we have no reason to contradict these. There is an anterolateral one with the maxilla, and, on the right side, a posterior one with the pterygoid. No sutures are visible on the dorsal surface. The midline suture is not clear.

Structures on ventral surface: The palatine in Nigeremys contributes only a small part to the triturating surface, as in Arenila and Azabbaremys . The choanal passage is a large, curved trough on each palatine, confluent with the apertura narium interna. The choanal passage of Nigeremys is almost the same as in Arenila , and both have a low midline ridge absent in Azabbaremys . The foramen palatinum posterius in Nigeremys seems to be completely formed by the palatine, but this is not certain due to unclear sutures. The foramen lies at the top of a cone-shaped concavity, very similar to the one seen in Arenila and distinct from the flat surface of Azabbaremys .

QUADRATE (figs. 219–223)

Preservation: Both quadrates are present in MNHN (P) NIR 1, but the bone is poorly preserved and has been further damaged by overpreparation, ink lines, and glue. The left quadrate is more complete than the right one. It is crushed slightly dorsoventrally and part of the cavum tympani is broken away. The bone surface is eroded and pitted and sutures are unclear. The right quadrate is badly crushed dorsoventrally so that the cavum tympani is barely discernable and the sulcus eustachii is not discernable. The processus articularis is broken off the right side. Although Lapparent de Broin and Werner (1998: fig. 4b, f) described sutures for the quadrate, we have been unable to confirm them.

Contacts in lateral view: The cheek on both sides in MNHN (P) NIR 1 is badly damaged and poorly preserved. We have been unable to confirm the sutures figured by Lapparent de Broin and Werner (1998: fig. 4f) showing a large quadratojugal contacting the quadrate anteriorly. There are cracks and fissures, as well as ink lines, in this area, but a careful microscopic examination has failed to distinguish sutures. However, it is quite possible to find fissures and cracks consistent with a quadrate-maxilla contact, as seen in Azabbaremys . We also dispute the quadrate-squamosal suture drawn by Lapparent de Broin and Werner (1998: fig. 4f). This is a crack that extends ventromedially into the cavum tympani and is not a suture.

Structures in lateral view: The cavum tympani in Nigeremys , visible on the left side, is unusually deep and overhung by its margins. This is probably due to dorsoventral crushing that has made the cavum lower and more oval than it was originally. It is quite possible that the cavum tympani in Nigeremys was originally similar in size and shape to that in Azabbaremys , that is, nearly circular. The incisura columellae auris in Nigeremys is completely closed, as in Arenila . The sulcus eustachii is a narrow, clearly defined trough, also similar to that in Azabbaremys . There is no sign of an antrum postoticum in Nigeremys , although the poor preservation does not preclude the presence of a small one. The antrum is absent in Azabbaremys also. As in most bothremydids, the fossa precolumellaris is absent in Nigeremys . Nigeremys has a well-developed shelf below the cavum tympani as in Azabbaremys .

Contacts in dorsal view: Sutures on the dorsal surface of the otic chamber in MNHN (P) NIR 1 are not distinguishable.

Structures in dorsal view: The foramen stapedio-temporale in MNHN (P) NIR 1 is not determinable.

Contacts in ventral and posterior views: Although Lapparent de Broin and Werner (1998: fig. 4f) showed a number of quadrate sutures in MNHN (P) NIR 1, we have only been able to confirm the basioccipital-quadrate suture. This lies about halfway along the ridge forming the posterior wall of the fossa pterygoidea. The Lapparent de Broin and Werner (1998) figure shows this suture placed more anteromedially, as a function of a very narrow basioccipital. We interpret the basioccipital as wider (see Basioccipital) and place the quadrate suture more laterally, as in Arenila .

Structures in ventral and posterior views: The foramen posterius canalis carotici interni in Nigeremys was figured by Lapparent de Broin and Werner (1998: fig. 4f) as being in the roof of the fossa pterygoidea and formed by the quadrate and pterygoid (left side) and quadrate, pterygoid, and basisphenoid (right side). We have carefully examined this area in MNHN (P) NIR 1 and have been unable to confirm the position or presence of this foramen. The two areas indicated in Lapparent de Broin and Werner (1998) are badly eroded and pitted, with a few pits more heavily prepared than others. We see no sign of natural foramina margins or sutures.

Although not identified in a figure Lapparent de Broin and Werner (1998: 179) described the ‘‘facial foramen’’ (i.e., the foramen nervi facialis). However, we have been unable to confirm the presence of this foramen. The fossa pterygoidea is probably formed mostly by quadrate and pterygoid in Nigeremys . The fossa is relatively deep and large, as in Arenila but in strong contrast to all the other Taphrosphyini , which lack it.

The condylus mandibularis and processus articularis are present on the left side of MNHN (P) NIR 1, but the bone surface is damaged. The right quadrate lacks the condylus and most of the processus articularis. The condylus mandibularis in Nigeremys lies at about the level of the condylus occipitalis, as it does in Arenila , but in contrast to the more anterior position in Azabbaremys and the other Taphrosphyini . The occipital structures are badly preserved, but the fenestra postotica (fig. 223) is discernable and seems to consist of a single opening, although markings drawn on the specimen obscure some features.

PTERYGOID (figs. 219–222)

Preservation: Both pterygoids in MNHN (P) NIR 1 are present and nearly complete, but their surfaces are pitted and eroded. Sutures are unclear. Breakage is present along the margins, particularly the processus trochlearis pterygoidei and the quadrate ramus. Only the ventral surface is visible.

Contacts on ventral surface: Only a few sutures are discernable in MNHN (P) NIR 1. We agree with Lapparent de Broin and Werner (1998: fig. 4f) in the position of the anterior palatine-pterygoid suture, the midline suture, and the general position of the basisphenoid suture. However, the basisphenoid is slightly more anterior in our interpretation, resulting in a very narrow midline contact of the two pterygoids, similar to that in Arenila . We are unable to confirm the pterygoid-quadrate suture figured by Lapparent de Broin and Werner (1998: fig. 4f).

Structures on ventral surface: The processus trochlearis pterygoidei in Nigeremys trends posterolaterally, as in Arenila and Phosphatochelys . It is eroded around its margins (on both sides) but seems to be in its original position and to retain its original size. The quadrate ramus is a high wall forming the posterior margin for the fossa pterygoidea, as in Arenila . The fossa pterygoidea (see Quadrate) seems to be formed by the quadrate posteriorly, the pterygoid anteriorly, and the basisphenoid medially, as in Arenila . The foramen posterius canalis carotici interni is not identifiable (see Quadrate).

Contacts and structures on dorsal surface: The right fossa temporalis in MNHN (P) NIR 1 still contains matrix, but the left one shows the lateral surface of the processus trochlearis pterygoidei and some of the postorbital wall. Sutures are not discernable. Enough of the sulcus palatinopterygoideus can be made out to show that it is narrow, not wide as in Azabbaremys .

SUPRAOCCIPITAL (figs. 219, 222, 223)

Preservation: The supraoccipital in MNHN (P) NIR 1 is covered by matrix anteriorly, and much of the crista supraoccipitalis is missing. The bone is poorly preserved and badly pitted.

Contacts: No sutures are visible in MNHN (P) NIR 1.

Structures: The crista supraoccipitalis in Nigeremys can be seen in cross section, and the lower part is much thicker than the upper part, as in Arenila . The crista is broken posteriorly, so its length cannot be determined, but its longest part is at the ventral edge, which is longer than the ventral edge in Azabbaremys . The edges of the foramen magnum are not clear, as it is filled with matrix.

EXOCCIPITAL (figs. 219, 222, 223)

Preservation: Both exoccipitals are present in MNHN (P) NIR 1, but despite their bone surfaces being better preserved than most of the skull, sutures are not discernable.

Contacts: No sutures are visible in MNHN (P) NIR 1.

Structures: The edges of the foramen magnum in Nigeremys are not clear due to matrix. The bones making up the condylus occipitalis cannot be determined. Lapparent de Broin and Werner (1998: fig. 4f) showed a large basioccipital entering the condyle. However, it is just as likely that it was formed solely by exoccipitals. One foramen nervi hypoglossi could be interpreted on the right side of Nigeremys . The foramen jugulare posterius (fig. 223) is more certain, despite markings drawn on the specimen. The foramen is closed laterally and is separated from the fenestra postotica by a distance about the same as in Arenila and farther than in Azabbaremys .

BASIOCCIPITAL (figs. 219–222)

Preservation: The basioccipital is present and nearly complete in MNHN (P) NIR 1, but the surface is badly eroded and pitted. Nonetheless, we think that some sutures are discernable.

Contacts: Our interpretation of the basioccipital shape in MNHN (P) NIR 1 differs from that in Lapparent de Broin and Werner (1998). They showed an elongate, parallel-sided basioccipital not much thicker than the condylus occipitalis. Microscopic examination does not confirm these as sutures. In our interpretation, on the left anterior margin of the basioccipital is an anterolaterally trending suture paralleling the fossa pterygoidea. This is the basisphenoid contact, and posteriorly it is the basioccipital-quadrate contact. These sutures show that the basioccipital of Nigeremys is wider than long and has posterolateral processes extending along the fossa pterygoidea, all as in Arenila . In any case, the bone surface is poorly preserved, allowing a number of interpretations.

Structures: The contribution, if any, of the basioccipital to the condylus occipitalis cannot be determined in MNHN (P) NIR 1 (see Exoccipital). As in Arenila , there is no prominent tuberculum basioccipitale in Nigeremys . We have been unable to find a foramen posterius canalis carotici interni (see Quadrate). There is a shallow concavity on the surface of the basioccipital, as in Arenila . If the question of sutures is put aside, there is remarkably close agreement in the shape of the bone surface in the basicranium in both Nigeremys and Arenila .

PROOTIC

Preservation: Both prootics are presumably present in MNHN (P) NIR 1, but the absence of sutures and surface damage makes it hard to define them. The right one is still partially obscured by matrix, but the left one is visible.

Contacts: No sutures are discernable in MNHN (P) NIR 1.

Structures: Neither the foramen nervi trigemini nor the foramen stapedio temporale can be identified with assurance in Nigeremys . However, the dorsal surface of the left otic chamber is better preserved than most of the skull, and it seems likely that a foramen would be visible if present on the dorsal surface. The anterior surface has pits and damaged areas that could be a foramen.

OPISTHOTIC (figs. 219–223)

Preservation: Both opisthotics are present in MNHN (P) NIR 1, but their posterior surfaces are badly pitted and eroded. Most sutures are not discernable.

Contacts: The opisthotic-squamosal suture is visible on both sides in MNHN (P) NIR 1 but other contacts are not, despite the sutures figured in Lapparent de Broin and Werner (1998: fig. 4f).

Structures: The fenestra postotica in Nigeremys is a single oval placed laterally far from the foramen jugulare posterius (see Quadrate). The posterior edge of the opisthotic in Nigeremys has a large, horizontal ridge not found in Azabbaremys (the opisthotic is missing in Arenila ). The ridge continues onto the squamosal. Some of the ridge may be missing, as the margin has the appearance of a broken edge.

BASISPHENOID (figs. 219–222)

Preservation: The basisphenoid is complete in MNHN (P) NIR 1, but its surface is damaged by pitting and erosion. Some sutures are discernable on the left side. Only the ventral surface is visible.

Contacts: On the ventral surface, the basisphenoid in Nigeremys contacts the pterygoid anterolaterally and the basioccipital posteriorly. Presumably, there is a posterolateral contact with the quadrate.

The basisphenoid as figured in Lapparent de Broin and Werner (1998: fig. 4f) is an elongate, parallel-sided element with a transverse basioccipital suture. The sutures we have found show the basisphenoid to be a triangular element, wider than long, with a curved basioccipital suture. The degree to which it enters the fossa pterygoidea is not determinable.

Structures on the ventral surface: The foramen posterius canalis carotici interni is not determinable (see Quadrate, Pterygoid). It could have been in the pterygoid-basisphenoid suture and might be one of the many pits and defects in this area. The fossa pterygoidea is probably formed medially by the basisphenoid. The roof of the fossa is particularly damaged with no clear sutures or bone surface.

Arenila krebsi

Arenila is based on one partial skull from the Maastrichtian of Egypt. Enough of the palate is present to determine most of that area, but the quadrates are poorly preserved. This skull was described by Lapparent de Broin and Werner (1998), who provided good photographs but no line drawings to go with them. To take advantage of those photographs, we provide labeled line drawings for the views in Lapparent de Broin and Werner (1998), as well as our own photographs. Arenila is the sister to Nigeremys , together making up the subtribe Nigeremydina , which is the sister group to the subtribe Taphrosphyina .

PREFRONTAL (figs. 224, 227)

Preservation: Most of the prefrontal in TUB Vb-641 is preserved on the left side, but it lacks its anterior margin. The right one is completely missing. Despite Lapparent de Broin and Werner’s (1998) assertion, not enough of the prefrontal margin is preserved to show that the apertura narium externa in Arenila is larger than the apertura narium externa in Nigeremys . The sutures are clearly visible, with the bones being slightly displaced along them. The prefrontal is broken by a parasagittal crack running from the frontal suture to the anterior eroded edge of the prefrontal just medial to the midline.

Contacts: The preserved contacts in Arenila show the usual ones for Bothremydidae : frontal posteriorly and maxilla anteroventrolaterally. The midline contact with the other prefrontal is missing. The frontal-prefrontal suture trends anteromedially to posterolaterally to a greater degree than in Nigeremys , but about the same as in Azabbaremys .

Structures: The dorsal margin of the apertura narium externa is entirely missing and the length of the prefrontal on the midline is indeterminate. It is quite possible that the anterior margin of the prefrontal in Arenila was the same as in Nigeremys . The prefrontal part of the orbital margin is preserved with some sign of erosion, but not as much as the anterior prefrontal edge. Presumably the margin here is close to the original. Within the fossa orbitalis of Arenila can be seen the ventral process of the prefrontal. It forms about as much of the fossa as in Azabbaremys , but differs from Azabbaremys in having a wide maxilla contact. The prefrontal forms the lateral edge of the foramen orbitonasale, although this opening is otherwise poorly preserved.

FRONTAL (figs. 224, 227, 229)

Preservation: The left frontal is nearly complete and only slightly distorted in TUB Vb-641. The right frontal is broken off medial to the orbital margin.

Contacts: As in other bothremydids, the frontal of Arenila contacts the prefrontal anterolaterally, the postorbital posterolaterally, the parietal posteriorly, and the other frontal medially. The postorbital suture is short, as in Taphrosphys and Nigeremys , not long, as in Azabbaremys .

Structures: The frontal of Arenila forms the dorsal part of the orbital margin. The degree of exposure is about the same as in Nigeremys and Azabbaremys . The sulcus olfactorius is visible in lateral view with what seems to be a deep wall, as in Azabbaremys . The frontal is relatively thick, as in Azabbaremys but in contrast to Taphrosphys and Phosphatochelys .

PARIETAL (figs. 224, 227, 229)

Preservation: Most of the left parietal and a small part of the right are preserved in TUB Vb-641. As with almost the entire skull, they are broken and at least partially damaged. The left processus inferior parietalis and ventral part of the right one are preserved. The preserved lateral edge of the left parietal appears broken for its anterior half, but the posterior half has a tapered, finished edge that seems to be original, all as described in Lapparent de Broin and Werner (1998: 175).

Contacts of dorsal plate: As in other bothremydids, Arenila has a parietal with these contacts: frontal anteriorly, postorbital anterolaterally, and parietal medially. As in Nigeremys it cannot be determined due to breakage whether there was a quadratojugal contact in Arenila .

Structures of the dorsal plate: The degree of emargination in Arenila cannot be determined. The restoration shows the minimum amount of bone present, but the parietal roof was probably more extensive than shown. It could be comparable to that restored for Nigeremys .

Contacts of processus inferior parietalis: The processus inferior parietalis in Arenila contacts the frontal wall and continues the sulcus olfactorius anteriorly. The processus inferior parietalis extends ventrally along the anterior margin of the prootic to contact the pterygoid. The parietal appears to extend laterally under the parietal-frontal suture to contribute to the postorbital wall above the sulcus palatinopterygoideus. The presence of the postorbital in Arenila and Nigeremys is in contrast to the other members of the tribe Taphrosphyini . It is possible that the parietal in Arenila does have a descending process lateral to the sulcus palatinopterygoideus that reaches the pterygoid. However, the required suture is at present a crack and therefore ambiguous. In the dataset Arenila is given a question mark for this character.

On the right side of TUB Vb 641, most of the processus inferior parietalis is missing, but its posteroventral part is present. It shows the posterior prootic contact above the foramen nervi trigemini and the ventral pterygoid contact anterior to the foramen nervi trigemini. Posteriorly the parietal contacts the supraoccipital as in other turtles.

Structures of the processus inferior parietalis: The processus inferior parietalis is relatively broad in Arenila , with the margin of the foramen interorbitale more anterior than in those forms with a narrow processus, Azabbaremys and Phosphatochelys . It is not known for the other Taphrosphyini .

The foramen nervi trigemini is formed by the prootic posteriorly, the parietal anteriorly, and the pterygoid ventrally, as in most Bothremydidae . In Arenila there is a vertical ridge anterior to the foramen that is absent in Azabbaremys but present in Phosphatochelys . It is indeterminate in the other Taphrosphyini .

JUGAL (figs. 224, 227, 229)

Preservation: The medial process and anterior part of the lateral plate of the left jugal are present in TUB Vb 641. The posterior edge of the lateral plate of the jugal is a broken margin, so its original extent is unknown. The sutures are clear but there is displacement and breakage along the sutures, particularly with the maxilla. As with other elements in TUB Vb 641, the bone surface is eroded and damaged.

Contacts of lateral plate: The jugal in Arenila contacts the postorbital dorsally and the maxilla ventrally, as in other Bothremydidae . The possible quadrate and quadratojugal contacts are not determinable. Nigeremys has poor sutures, but the lateral plate of the jugal seems to contact the postorbital and maxilla as in Arenila .

Contacts of medial process: The floor of the left orbit is badly fractured, but parts of the jugal-maxilla suture can be seen. In the postorbital wall, the jugal-postorbital contact runs dorsally from the pterygoid, as in most bothremydids. The jugal contacts the pterygoid and, more ventrally, the maxilla, although this area is badly broken with some displaced fragments.

Structures of medial process: As in other bothremydids, the jugal of Arenila forms the lateral part of the septum orbitotemporale. The septum orbitotemporale in Arenila and Nigeremys is completely formed as in the tribe Bothremydini , but in contrast to the reduced or absent postorbital wall seen in the other Taphrosphyini .

QUADRATOJUGAL

Preservation: There is no identifiable part of the quadratojugal in TUB Vb 641.

SQUAMOSAL (figs. 224, 227)

Preservation: An eroded fragment of the right squamosal is present on the posterodorsal edge of the right quadrate and is the only remnant of the squamosal in TUB Vb 641. The only original bone surface is a small section anteromedially.

Contacts: The only contact remaining in TUB Vb 641 is the anterior one with the quadrate.

Structures: The squamosal remnant is probably too incomplete laterally to retain any of the antrum postoticum if one was present. In any case, there is no sign of an antrum postoticum. The squamosal in TUB Vb 641 does extend posteriorly more than the squamosal in Nigeremys , but Nigeremys is broken in this area and may have had more of a posterior extension than is preserved.

POSTORBITAL (figs. 224, 227, 229)

Preservation: The left postorbital is preserved in TUB Vb 642, but its posterior margin is broken and the medial process is damaged by cracks and displaced fragments. The lateral plate surface is eroded. The lateral plate sutures are clear, but the medial process ones are not.

Contacts of lateral plate: As in other bothremydids, the postorbital of Arenila contacts the frontal anteromedially, the parietal medially, and the jugal ventrolaterally. The quadratojugal contact is missing.

Structures of the lateral plate: As preserved, the orbital margin of Arenila shows an orbit relatively larger than the one in Nigeremys and Azabbaremys , but comparable to that in Phosphatochelys . Lapparent de Broin and Werner (1998) used the larger orbit of Arenila as a diagnostic difference from Nigeremys . However, close examination of TUB Vb 641 shows that all of the elements forming the orbital margin are abraded and eroded. The posteroventral elements (maxilla, jugal, and postorbital) seem to be particularly eroded and have edges exposing the cancellous marrow. The postorbital and jugal could be missing enough bone that, if restored, would result in an orbital margin of the size seen in Nigeremys , removing this feature as a difference between the two taxa ( Lapparent de Broin and Werner, 1998). The maxilla is also damaged and the frontal is clearly lacking its original margin. Only the prefrontal comes close to retaining an original margin. We have not compensated for these missing edges in the restoration (fig. 224), so the orbit was probably smaller than in our restoration.

Contacts of medial process: The medial postorbital process is not well preserved in TUB Vb 641, but the contacts with the pterygoid ventrally and the jugal laterally are clear. It is possible that there is a medial contact with the parietal (see Parietal) that prevents postorbital exposure in the lateral wall of the sulcus palatinopterygoideus.

Structures of medial process: The postorbital in Arenila forms a major part of the septum orbitotemporale. The surface exposed in the posterior wall of the fossa orbitalis is partially covered in matrix, and the sutures are not clear.

PREMAXILLA (figs. 224, 227)

Preservation: The posterior parts of both premaxillae are preserved. The right premaxilla has come to light subsequent to the description of Lapparent de Broin and Werner (1998).

Contacts: As preserved, there is the lateral contact with the maxilla, the posterior contact with a remnant of the vomer (on the right side), and the median contact with the other premaxilla.

Structures on dorsal and ventral surfaces: The dorsal surface is eroded and does not clearly show the margins of openings. Posteriorly the ventral surface forms the foramen praepalatinum and a groove entering it. The triturating surface formed by the premaxilla defines the posterior margin of the midline concavity, which is laterally and posterolaterally continuous with the lingual ridge of the maxilla. The premaxilla is unusually large, as in Nigeremys and in contrast to other Taphrosphyini .

MAXILLA (figs. 224, 227, 229)

Preservation: When originally described only the left maxilla was known for TUB Vb 641, but now a partial right maxilla is also available. The left maxilla is much more complete and better preserved than is the right one. The left maxilla is considerably eroded on its lateral surface. Almost all of the original bone surface is gone and the margins are broken. The labial ridge has been crushed medially and the bone displaced so that the ridge is more medial to the edge of the skull than it was premortem. The triturating surface is cracked, but most of it is not eroded.

Contacts of vertical plate: As in other bothremydids, the maxilla of Arenila contacts the premaxilla anteriorly, the prefrontal anterodorsally, and the jugal posterodorsally. The posterior contacts are unknown, but the restoration shows that the maxilla is relatively long in Arenila as it is in Azabbaremys , placing the known end of the maxilla close to the quadrate.

Structures of vertical plate: The anterior end of the maxilla in Arenila is a broken margin with no sign of the apertura narium externa. The apertura in Arenila could have been the same as in Nigeremys . The dorsal process of the maxilla in Arenila is about as thick and high as it is in Azabbaremys , in strong contrast to the conditions in Labrostochelys or Phosphatochelys . In Nigeremys the process is the same thickness, although the suture appears to be inclined posterodorsally rather than anterodorsally as in Arenila .

The size of the orbit has been discussed under Prefrontal. The maxilla edge is eroded in TUB Vb-641, but not as much as the postorbital and jugal margins. The labial ridge has been crushed dorsoventrally as well as medially. Considering both orbital erosion and labial ridge crushing, the suborbital depth of the maxilla must have been greater than it is now. One of the few differences between Arenila and Nigeremys is the suborbital depth of the maxilla, and we have not adjusted this for crushing in our Arenila restoration. There is no way to be sure, but it is possible that the suborbital depth was the same in both originally.

Contacts of horizontal plate: In Arenila the maxilla contacts the premaxilla anteromedially, the palatine posteromedially, and the jugal posterolaterally, all as in other Taphrosphyini . Arenila also has a small, medial contact with the vomer, very much the same as in Nigeremys . Labrostochelys also has a vomer contact, but Azabbaremys does not. The vomer is not known for Phosphatochelys and Taphrosphys .

Structures of horizontal plate: The triturating surface of Arenila is mostly formed by the maxilla with only a small posteromedial contribution by the palatine, as in Nigeremys and Azabbaremys . The labial ridge is very thick in Arenila and Nigeremys , thicker than in any other Taphrosphyini . The ridge is blunt in Nigeremys and acute in Arenila . The entire surface of the skull of Nigeremys has been damaged, and the labial ridge as preserved is irregular and could easily have been as acute as in Arenila before being damaged. The lingual ridge of Arenila is very low and parallel to the labial ridge, as in Nigeremys . Both have a very shallow trough between the ridges. The labial ridge is straight in Arenila and Nigeremys , not curved convex outward as in the other Taphrosphyini , except Labrostochelys . The shape of the triturating surface is similar in Arenila , Nigeremys , and Azabbaremys , except that Arenila and Nigeremys have a broader medial process reaching the vomer.

The apertura narium interna in Arenila is very similar in shape to Nigeremys , with the thick edge of the lingual ridge of the maxilla forming its anterolateral margin. The foramen orbitonasale is present on the left side, but the thin edges are mostly broken away.

VOMER (figs. 224, 227)

Preservation: The anterior and posterior ends of the vomer are present in TUB Vb- 641, but the central bar is missing. The parts preserved show clear sutures.

Contacts: The vomer of Arenila contacts the premaxilla anteriorly, the maxilla anterolaterally, and the palatine posteriorly, as in Nigeremys and Labrostochelys . In Azabbaremys there is no vomer-maxilla contact.

Structures: Even though the vomer of Arenila lacks the central bar, the preserved parts show that the bar was thick and short, as in Nigeremys and in contrast to the very thin vomer of Azabbaremys . Enough of the two foramina praepalatinum are preserved on the premaxillae to show that the vomer may participate in the formation of these canals as described by Lapparent de Broin and Werner (1998).

The foramen narium interna in Arenila is formed by the vomer medially, the maxilla laterally, and the palatine posteriorly. As restored, the apertura in Arenila is about the same size as in Nigeremys although much of the margin is broken and the palatine edge is largely missing. Lapparent de Broin and Werner (1998: 174) used ‘‘larger choanal’’ to distinguish Arenila from Nigeremys , but this condition is the result of postmortem breakage. Furthermore, our examination of Nigeremys shows that the apertura narium interna is larger than that depicted in Lapparent de Broin and Werner (1998: fig. 4b).

Neither Arenila nor Nigeremys has the high degree of palatal arching seen in Azabbaremys and reflected in the vomer orientation that is more inclined in Azabbaremys than in Arenila and Nigeremys .

PALATINE (figs. 224, 227)

Preservation: Most of both palatines are preserved in TUB Vb 641, with some cracking along the choanal passage. A section is missing posteriorly on the right side and both palatines lack their anteromedial edges. The sutures are distinct.

Contacts: The palatine contacts in Arenila are with the vomer anteromedially, the palatine medially, the pterygoid posteriorly, and the maxilla anterolaterally.

Structures on the dorsal surface: Some of the contact of the palatine and the processus inferior parietalis is visible on the left side in Arenila , but it is too crushed to make detailed comparisons. The dorsal opening of the foramen palatinum posterius is formed anteriorly by the palatine, and on the left side a trough formed by palatine leads into it from the floor of the sulcus palatinopterygoideus. The sulcus itself is partially obscured by broken bone and matrix, but a parietal contact is possible. The palatine forms a major part of the fossa orbitalis floor and contacts the maxilla laterally and, possibly, the prefrontal anteriorly. The foramen orbitonasale is medial to the palatine here.

Structures on ventral surface: As in Nigeremys and Azabbaremys , the palatine of Arenila only forms a small part of the triturating surface. A large, curved choanal passage is confluent with the apertura narium interna, as in Nigeremys .

The foramen palatinum posterius in Arenila is formed anteriorly by the palatine and posteriorly by the pterygoid. As preserved, the foramen is very large, and Lapparent de Broin and Werner (1998) used this to distinguish Nigeremys from Arenila . Howev- er, discovery of a partial left palatine and close examination of the right palatine suggest that a smaller foramen was original and that the large one is a preservational artifact. The edges of the left foramen palatinum posterius are intersected by matrix-filled cracks that have enlarged and dislocated the margins of the foramen. When the remaining original edges are adjusted for this, the foramen is less than half the size as preserved. The foramen in Arenila is still slightly larger than in Nigeremys .

QUADRATE (figs. 224–230)

Preservation: Only the left quadrate is present in TUB Vb 641and almost all of it is badly damaged by erosion that has removed a large percentage of the bone. Only a small contact with the rest of the skull remains, and only a few areas of original bone surface are present.

Lateral view contacts: None of the cheek or skull roof contacts is preserved in Arenila . A block of the eroded squamosal is present on the posterodorsal part of the quadrate. Only the internal contact is preserved.

Lateral view structures: Only a small vestige of the cavum tympani is preserved in Arenila , and this is the medial wall of the sulcus eustachii. The canal for the columella auris is eroded away where it reaches the cavum tympani as described by Lapparent de Broin and Werner (1998). The distance between the sulcus eustachii and the stapedial canal is slightly shorter in Arenila than it is in Nigeremys , as noted by Lapparent de Broin and Werner (1998), but in neither taxon is that distance very different from that in Azabbaremys . The antrum postoticum is entirely missing. This may not be due only to erosion, as it is likely that if a moderate or tubelike antrum were present there would be some of it visible. It is likely that the antrum postoticum was actually absent in Arenila as in Azabbaremys .

Dorsal view contacts: The dorsal parts of the quadrate are missing due to erosion in TUB Vb 641.

Dorsal view structures: The extensive erosion of the quadrate in Arenila has exposed the canalis stapedio-temporalis, but the foramen stapedio-temporale has its ventral margin preserved. The foramen faces anteriorly, as in Azabbaremys .

Ventral and posterior view contacts: The medial contacts with the pterygoid, basisphenoid, and basioccipital are preserved at least in part on the right side in Arenila . There is a broken section between the quadrate and basisphenoid-basioccipital. The pterygoid has a small fragment attached to the quadrate anteriorly just lateral to the broken section, allowing restoration of the pterygoid limits. In posterior view the quadrate contacts the exoccipital ventromedially. As described by Lapparent de Broin and Werner (1998), most or all of the opisthotic is missing on both sides.

Ventral and posterior view structures: Laterally, the quadrate in Arenila preserves a horizontal ridge on its posterior surface that seems to continue the line formed by the sulcus eustachii. Dorsomedially, the trough above this ridge runs into the fenestra postotica. Very little of the quadrate and none of the opisthotic are preserved to show the extent and shape of the fenestra postotica in Arenila .

Ventrally the processus articularis and condylus mandibularis are preserved on their posterior surface, but the anterior and lateral surfaces are gone. The position of the condylus mandibularis is slightly posterior to the condylus occipitalis in Arenila , as in Nigeremys but in contrast to the more anterior position of other Taphrosphyini . The fossa pterygoidea is large and deep in Arenila , as it is in Nigeremys . The pterygoid flange of the quadrate forms the posteromedial wall of this concavity, and this is preserved in Arenila . The foramen posterius canalis carotici interni in Arenila has no contribution by the quadrate (see Basisphenoid), contrary to Lapparent de Broin and Werner (1998).

PTERYGOID (figs. 224–229)

Preservation: Parts of both right and left pterygoids are present in TUB Vb 641, but both are incomplete. The anteromedial plate is present on both sides, and on the right side most of the quadrate process is present. When described by Lapparent de Broin and Werner (1998) only the left processus trochlearis pterygoidei was present, but the right one has now been found, which helps correct a misconception about its orientation (see below).

Ventral surface contacts: As in other Taphrosphyini , the pterygoid of Arenila contacts the palatine anteriorly, the basisphenoid posteromedially, the quadrate posterolaterally, and other pterygoid medially. As in Nigeremys , the midline pterygoid contact is very narrow in Arenila .

Ventral surface structures: The left processus trochlearis pterygoidei in TUB Vb 641 is distorted in its position as preserved. The posterolateral end has been rotated medially, making the processus seem larger and parallel to the midline. The apparent size increase is due to matrix and bone debris filling the opened contact between the pterygoid and palatine. The distortion has also affected the size of the foramen palatinum posterius, making it larger than in its original state (see Palatine). Also, the newly discovered right processus trochlearis pterygoidei shows a more ‘‘normal’’ position. When restored, the processus in Arenila is not unusually large, similar to Nigeremys when its erosion is accounted for.

The fossa pterygoidea is present in Arenila (see Quadrate), as in Nigeremys . The pterygoid forms the anterolateral part of the concavity. The foramen posterius canalis carotici interni (see Basisphenoid) is formed in the pterygoid-basisphenoid suture.

Dorsal surface contacts: Some of the dorsal part of the pterygoid in Arenila is visible. At the base of the processus trochlearis pterygoidei, there is a lateral contact with the jugal, an anterodorsal one with the postorbital, and probably a dorsomedial one with the parietal (see Parietal). The crista pterygoidea forms the ventral margin of the foramen nervi trigemini, with the parietal forming the anterior margin and the prootic the posterior margin. The foramen is relatively large, as in Azabbaremys (not visible in Nigeremys ).

SUPRAOCCIPITAL (figs. 224–230)

Preservation: Most of the supraoccipital is present in TUB Vb 641, but its posterior margin is a broken edge so its complete limits are not known.

Contacts: As in other Taphrosphyini , the supraoccipital of Arenila contacts the parietal dorsally, the prootic anterolaterally, and the exoccipital posteroventrally. The presumed quadrate and opisthotic contacts are not preserved.

Structures: The crista supraoccipitalis in Arenila is longer than in Nigeremys and Azabbaremys , although the bone in Nigeremys may be eroded. The crista in Arenila is nearly twice the length that it is in Azabbaremys . A low ridge is present on the crista supraoccipitalis in Arenila , about one-third of the way above the foramen magnum. The trough below this ridge opens into the large concavity on the exoccipital (see Exoccipital). This ridge is not in Azabbaremys or Nigeremys .

EXOCCIPITAL (figs. 224, 229, 230)

Preservation: Both exoccipitals are present in TUB Vb 641 and are nearly complete with clear sutures.

Contacts: As in other Taphrosphyini , the exoccipital of Arenila contacts the supraoccipital dorsally, the quadrate ventrolaterally, and the basioccipital ventrally. The presumed lateral contact with the opisthotic is missing.

Structures: The foramen magnum of Arenila is very similar in size and shape to that in Azabbaremys and Nigeremys (which is somewhat eroded). The condylus occipitalis is formed completely by the exoccipitals. There is one foramen nervi hypoglossi very close to the much larger foramen jugulare posterius. In Azabbaremys , there are two foramina nervi hypoglossi and they are separated from the foramen jugulare posterius. In Arenila , the foramen nervi hypoglossi is facing almost laterally, just on the margin of the foramen jugulare posterius, a condition similar to that seen in Taphrosphys .

BASIOCCIPITAL (figs. 224, 227, 230)

Preservation: The basioccipital in TUB Vb 641 is present, nearly complete, with clear sutures. Its posterolateral margins are damaged on both sides but not much can be missing, due to the presence of the quadrate.

Sutures: As in other Taphrosphyini , the basioccipital in Arenila contacts the basisphenoid anteriorly, the quadrate laterally, and the exoccipitals dorsally.

Structures: The basioccipital in Arenila is pinched off by the exoccipitals on the neck of the condylus occipitalis and does not reach that structure. The tuberculum basioccipitale is only a low, horizontal ridge at the posterior margin of the basioccipital, very similar to that in Azabbaremys . The area is damaged in Nigeremys . Most of the basioccipital in Arenila forms a broad concavity, much deeper than the one in Azabbaremys but similar to the poorly preserved one in Nigeremys . As in Azabbaremys and probably Nigeremys , the basioccipital of Arenila is wide and short, in contrast to the much longer one in Taphrosphys .

PROOTIC (figs. 224, 229)

Preservation: Both prootics are present in TUB Vb 641, but the right one is nearly complete and has clear sutures while the left one is more crushed and obscured by matrix. The right prootic has no dorsal contact and ends in a broken margin.

Contacts: As in other Taphrosphyini , the prootic of Arenila contacts the parietal dorsomedially (visible on left side), the supraoccipital posterodorsally, and the pterygoid ventrally. The opisthotic is missing its contacts represented by broken margins on the prootic. The left and most of the right quadrate are also missing and the possibility of a quadrate-supraoccipital cannot be excluded.

Structures: The foramen nervi trigemini in Arenila is formed by the usual suspects: parietal anteriorly, pterygoid ventrally, and prootic posteriorly. The foramen is oblong and, as preserved on the right side, has a larger posterior part and a smaller, anteroventral part, probably reflecting the division of the trigeminal ganglion into two branches at this point (one branch separating medially, see Gaffney, 1979a). However, the left foramen nervi trigemini (which does not seem to be as well preserved as the right one) is an oval. Although it is hard to judge the relative sizes of this foramen among skulls widely varying in size, it seems that Arenila and Azabbaremys have relatively large foramina compared to Taphrosphys . Phosphatochelys also seems to have a relatively large foramen. The opening is not visible in Nigeremys .

The foramen stapedio-temporale in Arenila is represented only by a part of the ventrolateral margin on the right quadrate. The area of the prootic-quadrate suture is damaged and the full extent of the foramen is not determinable; however, it is clearly on the anterior face of the otic chamber and close to the foramen nervi trigemini.

OPISTHOTIC

Preservation: The opisthotic is not preserved in TUB Vb 641, except possibly as fragments on the right side.

BASISPHENOID (figs. 224, 227)

Preservation: The basisphenoid in TUB Vb 641 is nearly complete, but its posterolateral margins end in broken edges. On the right side only a small amount of basisphenoid can be missing because the quadrate is separated from the basisphenoid only by a narrow gap. Only the ventral surface of the basisphenoid is visible.

Contacts in ventral view: As in other Taphrosphyini , the basisphenoid contacts in Arenila are: pterygoids anterolaterally, quadrate posterolaterally, and basioccipital posteriorly. The basisphenoid-quadrate contact in Arenila is very narrow, as in Azabbaremys and Labrostochelys . The contact is indeterminate in Nigeremys . The basioccipital-basisphenoid suture in Arenila and Nigeremys is curved sharply, convex anteriorly, in contrast to all other Taphrosphyini in which it is straight.

Structures in ventral view: The foramen posterius canalis carotici interni is formed mostly by the basisphenoid in Arenila , with the anterolateral margin formed by the pterygoid. A short groove formed by the basisphenoid leads anteromedially into the foramen. Lapparent de Broin and Werner (1998:179, fig. 12a) figured the foramen posterius canalis carotici as being at the junction of three elements: quadrate, basisphenoid, and pterygoid. Further preparation and close examination of this area confirms that the quadrate is some distance posterior to the foramen. The foramen and groove are visible on both sides. In our interpretation, the area described by Lapparent de Broin and Werner (1998) as forming the foramen posterius canalis carotici interni is the narrow gap between the quadrate and basisphenoid. Nonetheless, TUB Vb 641 is not well preserved in this area and multiple interpretations are possible.

The basisphenoid in Arenila forms the medial wall of the fossa pterygoidea (the ‘‘podocnemidid fossa’’ of Lapparent de Broin and Werner, 1998), a large concavity also found in Nigeremys but no other Taphrosphyini (see Pterygoid or Quadrate). At the anteriormost margin of the basisphenoid in TUB Vb 641 are small, paired foramina, just adjacent to the pterygoid suture. Preparation of these show that each leads into a small anteriorly trending canal. The contents of these foramina are unknown.

LOWER JAW MORPHOLOGY

The lower jaw in pleurodires has never been treated systematically, although Fuchs (1931), Poglayen-Neuwall (1953), and Gaffney (1979a) have useful coverage. As with the skull, the following descriptions follow a standard pattern outlined in appendix 1. Tables 20 and 21 give comparisons of lower jaws in pleurodires, particularly those described here.

FAMILY EURAXEMYDIDAE Euraxemys essweini MATERIAL AVAILABLE: In FR 4922 (figs. 231, 232), the right ramus of the lower jaw is complete and the left ramus is gone posterior to the coronoid process. Both rami are well preserved, free of matrix with clear sutures.

DENTARY

Preservation: The right dentary is complete; the left one is missing its posterior margin.

Contacts: The dentary in Euraxemys contacts the coronoid posterodorsally, the surangular posteriorly, and the angular posteroventrally. There is a narrow contact with the prearticular on the medial surface, below the coronoid.

Structures: The dentaries in Euraxemys seem to be separated on the midline symphysis by a suture. As preserved, they are separate, with the contact area on both rami being a mixture of what looks like a sutural surface and broken bone. The jaws are also sutured in Araripemys and most chelids, but not in Emydura , while the jaws in bothremydids and podocnemidids are fused.

The symphysis in Euraxemys is rounded and not pointed or protuberant as in many other pleurodires. The jaw is not greatly thickened at the symphysis. In Araripemys the symphysis and jaw are quite thin, similar to chelids like Chelodina and Hydromedusa . Euraxemys has a thicker jaw, very close in proportions to Emydura and Elseya , but without the protuberant symphysis.

The triturating surface in Euraxemys is roughly parallel-sided, slightly wider anteriorly and narrowing posteriorly. The labial ridge is sharp and much higher than the lingual ridge, which is only the medial edge of the triturating surface. The surface itself is slightly concave and sharply tilted medially, much as in Elseya latisternum (AMNH 103700).

On the external surface, the dentary of Euraxemys forms the processus coronoideus with the coronoid bone on the medial surface of the processus. The processus coronoideus in Pelomedusoides does not vary a great deal and its size in Euraxemys is about the same as in pelomedusids and Emydura . Just below the processus is a distinct foramen, the foramen dentofaciale majus. Posterodorsally the dentary contacts the surangular and posteroventrally it contacts the angular.

The medial surface of the dentary has a distinct groove below the triturating surface, the sulcus cartilaginis meckelii. In Euraxemys the sulcus becomes prominent posterior to the symphysis, as in Elseya , rather than extending closer to it, as in Pelusios . The sulcus leads posteriorly into the foramen intermandibularis medius. Within the dentary, just anterior to this foramen is the foramen alveolare inferius, as it is in chelids and other Pelomedusoides. The medial contacts of the dentary are with the coronoid posterodorsally, the prearticular posteriorly, and the angular posteroventrally.

ANGULAR

Preservation: The right angular is complete in FR 4922; the left one lacks its posterior end.

Contacts: The angular of Euraxemys contacts the dentary anteriorly, the prearticular dorsally, the articular posteriorly, and the surangular dorsally on the lateral surface.

Structures: The angular of Euraxemys is a long, thin bone that begins anteriorly on the medial surface and bends posteroventrally to end below the articular bone. The foramen intermandibularis caudalis is formed in the prearticular-angular suture and it opens into the fossa meckelii. The angular in Araripemys extends more anteriorly than in Euraxemys , which is similar in extent to Pelusios . In Emydura and Elseya the large

TABLE 20

Comparison of Lower Jaws 1 TABLE 21 Comparison of Lower Jaws 2

splenial lies above the angular. In all Pelomedusoides, including Euraxemys , the splenial is absent. The angular in Euraxemys goes posteriorly to the end of the jaw separating the prearticular and surangular and curves posterodorsally to contact the surangular on the lateral surface of the jaw.

SURANGULAR

Preservation: The right surangular is complete in FR 4922; only a fragment of the left one remains.

Contacts: The surangular in Euraxemys contacts the dentary anteriorly, the coronoid in a narrow contact anterodorsally, the angular posteroventrally, and the articular posteromedially.

Structures: The surangular in Euraxemys is a large element lying on the external surface at the posterior end of the jaw. The surangular is a flat plate that forms the lateral wall of the fossa meckelii and the lateral margin of the area articularis mandibularis. The fossa meckelii in Euraxemys is about the same size as in Emydura and slightly smaller than in pelomedusids. It is much smaller than in Proganochelys .

On the external surface of the surangular in turtles are a series of foramina that transmit branches of the mandibular branch of the trigeminal nerve (VII 3). These foramina are particularly variable in pleurodires but the largest is usually referred to as the foramen nervi auriculotemporalis. Fuchs (1931) described this area and these nerves in Podocnemis , and Poglayen-Neuwall (1953) described the area in a number of cryptodires and pleurodires. The foramen nervi auriculotemporalis transmits the nervus auriculotemporalis, which has at least two branches and an anastomosis ( Fuchs, 1931: figs. 8, 10). In podocnemidids the foramen and associat- ed canals are large and complex, in other pleurodires they are usually smaller. Howev- er, in Euraxemys , pelomedusids, and Hamadachelys , the foramen approaches the podocnemidid condition. In Emydura / Elseya , and Araripemys the foramen is small and comparatively inconspicuous. In Emydura / Elseya the foramen is a simple hole penetrating the surangular from the fossa meckelii to the external surface. In pelomedusids the foramen nervi auriculotemporalis opens on the external surface above a short canal that extends ventrally beneath a short strap of bone to open more ventrally still on the external surface of the surangular. The contents of this short canal are presumed to be a branch of the nervus auriculotemporalis, but the canal is not described in the literature. In Euraxemys there is a similar canal on the right surangular (the left one is missing), but it is opened laterally by erosion and does not form a completely enclosed canal as preserved. Nonetheless, it is very similar to that canal in pelomedusids. Anteriorly, the foramen nervi auriculotemporalis penetrates medially and anteriorly into another canal in the surangular that also opens on the external surface of the surangular a short distance from the foramen nervi auriculotemporalis proper. The senior author has not seen a parallel structure in other pleurodires. This canal also is presumed to contain a branch of the nervus auriculotemporalis. Fuchs (1931) described one of the multiple foramina nervi auriculotemporalis (the largest) as being combined with the apertura lateralis canalis transversi into a larger, combined foramen that opens on the external surface of the surangular. The senior author has not been able to determine its contents, but these structures clearly vary among podocnemidids and on each side of the jaw as well.

In Euraxemys there is only one foramen penetrating the surangular from the external surface into the fossa meckelii. This is also the case in Podocnemis . There is a posterior foramen and canal just behind the foramen nervi auriculotemporalis in AMNH specimens of Podocnemis , and this appears to be the apertura lateralis canalis transversi of Fuchs (1931). However, this canal goes posteriorly into the articular bone and does not enter the fossa meckelii. Is this the posterior path of the nervus auriculotemporalis? Is it Fuchs’ apertura lateralis canalis transversi? It does not sound like it.

The surangular forms the entire lateral wall of the fossa meckelii. The internal surface of the fossa is visible in Euraxemys . Its lateral wall is formed anteriorly by the dentary, which has a vertical suture with the surangular just anterior to the dorsal opening of the fossa meckelii. The anterior opening of the fossa is the foramen intermandibularis CORONOID

Preservation: Both coronoids are complete in FR 4922.

Contacts: The coronoid in Euraxemys contacts the dentary anterolaterally, the prearticular ventromedially, and the surangular posterolaterally.

Structures: The coronoid bone in Euraxemys lies on the medial surface of the dentary, forming the medial half of the processus coronoideus. It contacts the dentary anterolaterally, the prearticular ventromedially, and the surangular posterolaterally. The processus coronoideus does not vary a great deal in size among Euraxemys , Emydura / Elseya , pelomedusids, and Araripemys . It is slightly higher in these taxa than in Podocnemis . The coronoid extends anteriorly to a greater extent in pelomedusids, chelids, and Araripemys than it does in Euraxemys . As in chelids and pelomedusids, the coronoid is only barely visible in lateral view in Euraxemys . More of it is exposed laterally in Araripemys .

medius formed laterally by the dentary and medially by the prearticular. The very narrow floor of the fossa slopes ventrally into the foramen intermandibularis caudalis and is mostly formed by the angular. Between the angular and prearticular is the foramen intermandibularis caudalis. The posterior wall of the fossa meckelii is formed by the articular sandwiched between the prearticular and surangular. The canal for the chorda tympani lies lateral to the articular. The surangular forms the lateral quarter or so of the area articularis mandibularis, as in most turtles. In Euraxemys there is a clear suture between the articular and surangular.

ARTICULAR

Preservation: The right articular in FR 4922 is complete; the left one is gone.

Contacts: The articular in Euraxemys contacts the surangular laterally, the angular anteroventrally, and the prearticular medially.

Structures: The articular is a triangular-shaped element lying at the posterior end of the jaw. In Euraxemys it is exposed dorsally and posteriorly and is sandwiched between the surangular laterally and the prearticular medially. It contacts the angular anteroventrally. As in chelids, the area articularis mandibularis of Euraxemys is more flat than convex and faces posterodorsally. The area is not strongly convex as in pelomedusids and podocnemidids. Rather, the surface is gently convex, dropping off laterally to the part of the surface on the surangular.

The articular pinches out anteriorly where a thin section of it forms the posterior wall to the fossa meckelii. At the posteroventral end of the fossa meckelii, laterally in the suture between articular and prearticular, is the foramen anterius chorda tympani. The foramen posterius chorda tympani is on the outside of the jaw, on the posteromedial edge of the area articularis mandibularis. It is also formed by the prearticular medially and the articular laterally and lies at the base of a depression on the retroarticular process. This is similar to the chorda tympani foramina in Hamadachelys . In chelids and Araripemys the foramen posterius chorda tympani is more anterior and not adjacent to the articular surface. Emydura / Elseya and Hamadachelys have retroarticular processes much as in Euraxemys , while Araripemys and pelomedusids lack them. A very similar condition of the entire back end of the jaw is in AMNH 63579, Erymnochelys madagascariensis . However, in Podocnemis the retroarticular process is depressed and separated below the level of the area articularis mandibularis.

PREARTICULAR

Preservation: The right prearticular in FR 4922 is complete; the left one is missing its posterior half or so.

Contacts: The prearticular in Euraxemys contacts the coronoid anterodorsally, the articular posteromedially, and the angular ventrally.

Structures: The prearticular in Euraxemys is a long, thin sheet extending from the coronoid bone posteriorly on the medial surface to the retroarticular process. Anteriorly it forms the posterior margin of the foramen intermandibularis medius, which opens into the fossa meckelii. In Emydura / Elseya this area is formed by the splenial, but in Pelomedusoides the prearticular extends anteriorly to replace it. The prearticular in Euraxemys extends anteriorly farther than in Araripemys and pelomedusids. The dorsal edge of the prearticular forms the medial margin of the upper opening of the fossa meckelii. It is slightly higher than in Emydura / Elseya and slightly lower than in Araripemys .

The ventral margin of the prearticular in Euraxemys is a long straight suture with the angular. About halfway along it is an opening, the foramen intermandibularis caudalis. It opens between the medial jaw surface and the ventral part of the fossa meckelii. The foramen is in about the same position in chelids, pelomedusids, Araripemys , and Euraxemys . On the medial surface of the prearticular, in the floor of the fossa meckelii, the foramen anterius chorda tympani is formed between the prearticular and the articular.

The posterior end of the prearticular covers the articular laterally. It does not bear any of the area articularis mandibularis. A small part of the prearticular is exposed dorsally at the posteromedial corner of the area articularis mandibularis where it forms the medial half of the foramen posterius chorda tympani. The articular forms the lateral half of the foramen. Presumably the canalis chorda tympani is formed between the prearticular and the articular.

FAMILY BOTHREMYDIDAE

TRIBE KURMADEMYDINI Kurmademys kallamedensis (fig. 233) MATERIAL AVAILABLE: ISI 155E, a nearly complete jaw, lacking the left posterior end; ISI 155D, both rami and symphysis, lacking the left coronoid region and some of the triturating surface; ISI 155F, a partial right ramus.

DENTARY

Preservation: The dentary is almost entirely complete on the right side of ISI 155E, with some cracking. The left side is complete as well, except for a small part of the posterior end. In ISI 155D, the dentary is mostly present but is slightly eroded and missing the area around the left processus coronoideus and part of the right triturating surface. In ISI 155F, only the symphysis and most of the right dentary are present.

Contacts: The dentary contacts in Kurmademys are the same as in Cearachelys : coronoid posterodorsally, surangular posterolaterally, angular posteroventrally, and prearticular posteromedially. The surangular contact is vertical beneath the processus coronoideus, not as far anterior as in Bothremys , but more anterior than in Euraxemys .

Structures: The symphysis in Kurmademys is similar to that in Cearachelys , an upturned labial ridge with a U-shaped concavity behind a narrow triturating surface defined by a low lingual ridge. Based on ISI 155F and ISI 155D, the symphyseal anterior margin is projected dorsally into a curved hook, but in ISI 155E the symphysis is low, as in Cearachelys . The triturating surface of Kurmademys is very similar to that in Cearachelys , with a low labial ridge turned dorsally at its posterior end, and a higher lingual ridge. In ISI 155E the shallow concavity at the posterior end of the triturating surface is about as deep as it is in Cearachelys , but in ISI 155D the concavity is shallower and the triturating surface is narrower.

The sulcus cartilaginus meckelii does not meet on the symphysis; it is similar to that in Cearachelys , being formed anteriorly by the dentary, and merging with the symphyseal concavity. On the lateral surface of the dentary is the foramen dentofaciale majus, just below the posterior end of the triturating surface, as in Cearachelys .

ANGULAR

Preservation: A nearly complete right angular and a partial left one are present in ISI 155E. Both angulars are complete in ISI 155D, but the left one is slightly damaged. The anterior half of the right angular is present in ISI 155F.

Contacts: As in Cearachelys , the angular in Kurmademys contacts the dentary anteriorly, the prearticular dorsally, the articular posteriorly, and the surangular dorsolaterally. The prearticular contact is long, as in Cearachelys and Euraxemys but in contrast to Bothremydini and Taphrosphyini . The articular contacts are not visible.

Structures: The angular in Kurmademys is very similar to that in Cearachelys , wrapping around from the sulcus cartilaginis meckelli to the surangular laterally.

SURANGULAR

Preservation: The right surangular is complete in ISI 155E, with the left one missing posteriorly. In ISI 155D the right one is cracked and missing a few areas; the left one is only present ventrally.

Contacts: As in Cearachelys , the surangular of Kurmademys contacts the dentary anteriorly, the coronoid anterodorsally, the angular posteroventrally, and the articular posteromedially.

Structures: The surangular in both Cearachelys and Kurmademys has a shallow, laterally facing depression that covers the lateral surface of the jaw, which is better defined in Kurmademys . The foramen nervi auriculotemporalis is just anteroventral to the area articularis mandibularis, as in Cearachelys . Also as in Cearachelys , the surangular forms the anterolateral margin of the area. The surangular extends posteriorly on the lateral surface of the processus retroarticularis.

CORONOID

Preservation: Both coronoids are present and largely complete in ISI 155E. In ISI 155F the right coronoid is present, as it is in ISI 155D.

Contacts: The coronoid contacts in Kurmademys are very similar to those in Cearachelys . There is only a small lateral exposure of the coronoid above the dentary-surangular contact. On the medial surface the prearticular has a posteroventral contact, as in Cearachelys .

Structures: The processus coronoideus in Kurmademys is formed medially by the coronoid, as in Cearachelys . The processus is better defined and slightly higher in Cearachelys than in Kurmademys . The coronoid in Kurmademys extends anteroventrally onto the triturating surface, just dorsal to the concavity on the dentary. In Cearachelys the coronoid barely forms the margin of the triturating surface. A very narrow surangular-prearticular contact prevents the coronoid from entering the dorsal opening of the fossa meckelii.

ARTICULAR

Preservation: The articular is present and complete on the right side of ISI 155E and on the left side of ISI 155D.

Contacts: Most of the articular sutures in ISI 155D and ISI 155E are fused, but the surangular contact, anterolaterally, is visible.

Structures: The area articularis mandibularis in ISI 155E is slightly narrower than in Cearachelys , but in ISI 155D it is the same width. The surface shape is the same in both Kurmademys and Cearachelys .

PREARTICULAR

Preservation: None of the Kurmademys jaws has a complete prearticular, but the preserved areas are overlapping. The right prearticular in ISI 155E is the best, missing only some of the fossa meckelii margin. In the right prearticular of ISI 155D, the anterior part is missing. In ISI 155F, the anterior part is present on the right ramus.

Contacts: As in Cearachelys , the prearticular of Kurmademys contacts the coronoid anteriorly, the articular posteromedially, and the angular ventrally. The angular contact is long, as in Cearachelys , enclosing the fossa meckelii to a greater extent than in Bothremydini and Taphrosphyini . The foramen intermandibularis medius, therefore, is more anterior in Kurmademys and Cearachelys than in the latter tribes.

TRIBE CEARACHELYINI Cearachelys placidoi (fig. 234)

MATERIAL AVAILABLE: THUg 1798 , complete let ramus, symphysis, right ramus with articular region intact but most of remaining bone eroded away, associated with a skull and shell ; BSP 1976 I 160, a nearly complete lower jaw with symphysis fragmented and distorted, associated with skull and shell.

DENTARY

Preservation: The dentary of BSP 1976 I 160 is nearly complete except for the symphysis. The left dentary in THUg 1798 is nearly complete, but most of the right one is missing.

Contacts: The dentary in Cearachelys contacts the coronoid posteromedially, the surangular posterolaterally, and the angular posteroventrally. The surangular contact is nearly vertical directly beneath the processus coronoideus, as in Kurmademys . The anterior extent of the surangular and restriction of the dentary exposure in lateral view occur in the Bothremydini , Taphrosphyini , and Kurmademys , but not in Euraxemys , chelids, or pelomedusids.

Structures: As in all bothremydids, the symphysis is fused, with no sign of a suture in Cearachelys . The symphyseal area in Cearachelys is similar to that in Kurmademys , with a narrow triturating surface and a deep Ushaped concavity behind it. The concavity is below the level of the triturating surface, not raised into a symphyseal wedge, as in Bothremys cooki .

The dentary in Cearachelys extends posteriorly on the lateral surface ventral to the surangular to a point below the middle of the area articularis. On the medial surface it is hidden from view below the coronoid process by contact between the prearticular and angular. The labial ridge is lower than the lingual ridge except at the symphysis. Its lateral edge is slightly concave in outline. The lingual ridge is taller than the labial ridge except where they are even at the symphysis. It is tallest where it contacts the coronoid and then becomes progressively lower anteriorly. It remains distinct to the symphysis where it produces the U-shaped central concavity.

The triturating surface in Cearachelys is wider posteriorly than anteriorly. There is a very weakly developed pit in the posterolateral part of the dentary. The medial and posterior walls of the pit are formed by the dentary. There is no lateral wall, no roof, and no contribution by the coronoid. Although the symphysis is pointed anteriorly, there is no symphyseal hook. The symphysis is blunt, as in Kurmademys .

The sulcus cartilaginis meckelii in Cearachelys is open for a very short distance. It is closed at the symphysis and where the prearticular meets the angular ventral to the coronoid process. The foramen intermandibularis medius and the foramen alveolare inferius are not well defined. A foramen dentofaciale majus is present on the lateral side of the dentary just ventral to the small dentary pit. The presence of nutritive foramina indicates that the rhamphotheca of the mandible covered the dorsalmost part of the dentary and the adjacent coronoid.

The dentary of Cearachelys differs significantly from that of Bothremys . Well, not all that significantly. In general, the triturating surface is narrower throughout its length. Furthermore, the labial ridge of the dentary is much lower, and neither it or the coronoid contributes to the dentary pit as they do in Bothremys . Cearachelys is similar to B. maghrebiana but differs from B. cooki in having the U-shaped structure formed by the labial ridges meeting on the symphysis. The symphyseal wedge seen in B. cooki is absent in Cearachelys . This U-shaped structure is found in other bothremydid jaws, including those of Araiochelys , Chedighaii barberi , and AMNH 29989. It is absent from jaws of the Taphrosphyini .

The dentary of Cearachelys is like that of B. cooki , Araiochelys , and Rhothonemys in having the sulcus cartilaginis meckelii closed anteriorly. It is like Araripemys and Kurmademys in having the posterior limit of the sulcus below the coronoid process where it is closed by an anterior meeting of the prearticular and angular.

The dentary of Cearachelys is most similar to that of Kurmademys . They have a high lingual ridge and a low labial ridge rising posteriorly to form a distinctive, anterodorsally facing concave portion of the triturating surface. This morphology may be interpreted as primitive for Bothremydidae . A greater degree of depression at the posterior end of the triturating surface would produce the pit seen in Araiochelys and Bothremys .

The lower jaws of BSP 1976 I 160 and THUg 1798 differ in the width of the triturating surface and the posterior height of the lingual ridge. The larger jaw, BSP 1976 I 160, has a wider triturating surface posteriorly, and the lingual ridge here is higher. The jaw of THUg 1798 is about one-third smaller than that of BSP 1976 I 160 and these differences may be due to growth, as in many living turtles.

The foramen dentofaciale majus is formed entirely by the dentary, anterior to the surangular contact, and below the posterior end of the labial ridge.

ANGULAR

Preservation: The angular is nearly complete on both sides of BSP 1976 I 160 and on the left side of THUg 1798 (only part of the right one remains). The posterior sutures are unclear in both jaws.

Contacts: The angular in Cearachelys contacts the dentary anteromedially, the prearticular dorsally, the articular posteriorly, and the surangular dorsolaterally. These contacts are very similar to those in Kurmademys . In the tribes Bothremydini and Taphrosphyini the prearticular contact is much shorter. The posterior contacts are not well defined but are visible in parts of both specimens.

Structures: The angular forms the ventral margin of the fossa meckelii and its anterior continuation, the sulcus cartilaginis meckelii. As in Kurmademys , the sulcus stops well short of the symphysis. The angular wraps ventrally around the posterior part of the jaw. This area is very similar in both Cearachelys and Kurmademys .

SURANGULAR

Preservation: The surangular is present on both sides of BSP 1976 I 160, but there is postmortem damage in the form of pitting and cracks. The left surangular is present in

THUg 1798; its surface is slightly damaged, but it is nearly complete.

Contacts: The surangular in Cearachelys contacts the dentary anteriorly in a nearly vertical suture below the coronoid, as in Kurmademys . The surangular is more extensive anteriorly than in Euraxemys , chelids, and pelomedusids, but not as much as in the Bothremydini . The surangular contacts the angular posteroventrally and the articular posteriorly.

Structures: In the lateral surface of the surangular, just anteroventral to the area articularis mandibularis, is the foramen nervi auriculotemporalis. The foramen can be traced within the surangular to its entrance into the fossa meckelii, anterior to the articular. The Bothremydini and Taphrosphyini seem to lack this foramen.

The surangular in Cearachelys extends posterolaterally around the articular and forms part of the edge of the area articularis mandibularis. It does not seem to form as much of the area as it does in Euraxemys . The surangular extends laterally onto the processus retroarticularis, which is almost entirely formed by the articular. The fossa meckelii in Cearachelys is narrow, as in other bothremydids, with the lateral surangular wall being higher than the medial prearticular wall.

CORONOID

Preservation: The coronoid is nearly complete on the left side of BSP 1976 I 160; it is slightly damaged on the right side. In THUg 1798 the left coronoid is nearly complete, but the right one is only represent- ed by a fragment .

Contacts: The coronoid in Cearachelys contacts the dentary anterolaterally, the prearticular ventromedially, and the surangular posterolaterally. The coronoid is less exposed laterally in both Cearachelys and Kurmademys than it is in Bothremydini and Taphrosphyini , due to a greater coronoidsurangular contact. The coronoid-dentary contact in Cearachelys does not extend as far laterally as it does in Kurmademys .

Structures: The coronoid forms the short but distinct processus coronoideus in Cearachelys . The coronoid extends ventrally on the medial surface to form the dorsal edge of the sulcus cartilaginis meckelii. It does not form any of the triturating surface.

ARTICULAR

Preservation: The articular is present on both sides of BSP 1976 I 160 and THUg 1798, with some damage on the left side of BSP 1976 I 160.

Contacts: The articular contacts the surangular laterally, the angular ventrally, and the prearticular medially.

Structures: The area articularis mandibularis in Cearachelys is convex, with a very low ridge anteroposteriorly, as in most Bothremys . The area is wider in BSP 1976 I 160 than in Kurmademys , but in THUg 1798 it is intermediate in width. The articular forms a narrow part of the posterior edge of the fossa meckelii.

The processus retroarticularis in Cearachelys is short, wide, and relatively massive, similar to that in Chedighaii . It is as long as in Kurmademys but wider. There is a well-defined groove on the medial surface for the chorda tympani, which is a foramen in Kurmademys .

PREARTICULAR

Preservation: The prearticular is nearly complete on the left side of BSP 1976 I 160; the right one has some damage anteriorly. In THUg 1798 the prearticular is present on both sides ; the left one is more complete than the right one.

Contacts: The prearticular in Cearachelys contacts the coronoid anterolaterally and the articular posteriorly. The angular contact runs for nearly the entire length of the prearticular, as in Kurmademys and Euraxemys , but in contrast to the much shorter contact of Bothremydini and Taphrosphyini .

Structures: The prearticular is a flat bone that forms the medial wall of the fossa meckelii, enclosing it for the entire length of the bone, in contrast to the more open fossa of Bothremydini and Taphrosphyini . In the angular-prearticular suture is a small foramen, the foramen intermandibularis caudalis.

TRIBE BOTHREMYDINI Foxemys mechinorum (figs. 235, 236)

MATERIAL AVAILABLE: PAM 511B, lower jaws lacking articular ends, figured in

Tong et al. (1998: figs. 7, 8); MC M2114, jaw lacking both articular ends; MC M2115, left ramus lacking articulation; MC M2116, left ramus lacking articulation; MC M2117, jaw lacking articular ends; MC M2118, nearly complete jaw, lacking left articular end, some cracks with displacement in symphysis.

DENTARY

Preservation: The six specimens all have most of the dentary preserved, although the preservation is best in MC M2118. In the four small jaws, MC M2114, MC M2115, MC M2116, and MC M2117, the dentary is either present only on one side or it is damaged. The largest jaw, PAM 511B, is complete on the right side but is missing the dorsal part of the left ramus. The best jaw, MC M2118, has a nearly perfect right ramus, but there is breakage with some displacement on the symphysis.

Contacts: The dentary in Foxemys contacts the coronoid posterodorsally, with more of the dentary being exposed ventral to the coronoid than in Kurmademys but not as much as in Bothremys . The surangular contact posterolaterally is more extensive anteriorly in Foxemys than in Kurmademys , but not as extensive as in Bothremys . The dentary contacts the angular posteroventrally, as in the other bothremydids.

Structures: The dentary is preserved in all six Foxemys jaws and it shows a size range, as measured from the symphysis to the processus coronoideus (see appendix 8), from 39 to 19 mm, presumably related to age. This increase in size is correlated with an increase in width of the triturating surface, so that the largest, PAM 511B, also has the relatively widest triturating surface. The width increase seems to be expressed in the lingual shelf in the smaller jaws. In the larger jaws, the labial edge is significantly swollen.

The symphysis of Foxemys has the symphyseal concavity of Kurmademys , but the lingual ridge on each side stops short of forming an anterior margin for the concavity, as in Kurmademys . The concavity is a sloping surface from the labial ridge posteriorly to the end of the symphysis. The labial ridge is upturned slightly in Foxemys , as in Kurmademys , rather than being flat, as in Araiochelys and Bothremys . The symphysis thickness is parallel trough are unique to Foxemys . In PAM 511B ( Tong et al., 1998: figs. 7, 8), the labial ridge is rounded, the trough is shallower and wider, and the lingual shelf is wider than in MC M2118.

In lateral view, the lingual shelf is higher than the labial ridge, agreeing with Cearachelys , Kurmademys , and other Bothremydini . The foramen dentofaciale majus lies entirely in the dentary, just anterior to the surangular contact. The medial surface of the dentary has a sulcus cartilaginis meckelii that stops short of the symphysis, as in Kurmademys and Araiochelys .

greater in Foxemys than in the other bothremydids.

The labial ridge in Foxemys is an acute ridge with a slight upturn that decreases in height posteriorly. In Kurmademys , Cearachelys , Araiochelys , Bothremys , and Chedighaii , the labial ridge is the edge of a relatively flat triturating surface, but in Foxemys the ridge is distinct on the margin of a trough that runs parallel to the ridge. None of the other bothremydids has the trough seen in Foxemys . Medial to the trough there is a wide, lingual shelf beginning anteriorly just behind the symphysis and sloping dorsally to the processus coronoideus. This shelf might be considered a flattened and broadened lingual ridge. The medial edge of the lingual shelf has no distinct ridge. This raised lingual shelf and ANGULAR

Preservation: A nearly complete angular is present only on the right side of MC M2118, with the anterior part remaining on the left side. PAM 511B also has the anterior parts of both angulars. Small parts of the angular are also in MC M2114, MC M2116, and MC M2117.

Contacts: The angular in Foxemys is very similar to that in Bothremys maghrebiana . It contacts the dentary anteriorly, the prearticular dorsally, the surangular dorsolaterally, and, presumably, the articular posteriorly, although the articular sutures are fused in MC M2118, the only specimen with an articular. The angular in Foxemys lacks the long prearticular contact seen in Kurmademys and Cearachelys . This short contact in Foxemys is the same as in Bothremys and other Bothremydini and Rhothonemys .

Structures: The angular forms the ventral margin of the sulcus cartilaginis meckelii and lower edge of the jaw. The fossa meckelii has the more open condition, seen in other Bothremydini , rather than the more closed fossa of Cearachelys and Kurmademys and other pleurodires. The suture between the angular and prearticular is not perfectly preserved in MC M2118, but there is a foramen intermandibularis caudalis.

SURANGULAR

Preservation: The surangular is nearly complete only on the right side of MC M2118, but the anterior part is present on the left side as well in PAM 511B and MC M2114.

rower. The surangular thickening is extensive and forms an upper ventrolaterally sloping surface and a ventral, medially sloping surface with a low ridge between them, all unique to Foxemys , among bothremydids. There is a small foramen on the upper surface.

Contacts: The surangular in Foxemys contacts the dentary anteriorly, the angular ventrally, the coronoid anterodorsally, and, presumably, the articular posteromedially, although the sutures are fused.

Structures: The surangular lateral exposure in Foxemys is not as extensive as in other Bothremydini , but it is more extensive than in Kurmademys and Cearachelys . In those taxa the dentary-surangular contact is vertical beneath the processus coronoideus, while in Foxemys , the surangular sends a process anteriorly. The surangular forms the lateral wall of the fossa meckelii and the lateral margin of the upper opening to the fossa. In Foxemys , in contrast to other bothremydids, the surangular is very thick, and the upper opening is displaced medially and is nar- CORONOID

Preservation: Both coronoids are nearly complete in MC M2118. The right one is present in PAM 511B, and the left one is present in MC M2114.

Contacts: The coronoid contacts the dentary anteriorly and ventrally, the surangular posterolaterally, and the prearticular posteromedially. The surangular contact is narrow, as in Kurmademys , not wide, as in Bothremys .

Structures: The coronoid in Foxemys forms the posteromedial part of the triturating surface, in particular, the posterior part of the lingual shelf, ending just medial to the triturating surface trough. The posterior edge of the triturating surface is a ridge running just anterior to the very low processus coronoideus; the ridge is actually higher than the processus in lateral view. In Bothremys , there is no ridge at the edge of the triturating surface and the processus coronoideus is much higher. In Kurmademys and Cearachelys the morphology is more similar to that of Foxemys , but the processus is higher and not as distinct from the triturating surface ridge. The coronoid narrowly enters the margin of the upper opening of the fossa meckelii, as in other bothremydids.

ARTICULAR

Preservation: The articular is present only on the right side of MC M2118.

Contacts: Sutures are fused in the available jaw.

Structures: The area articularis mandibularis of Foxemys is roughly spherical, with the two facets separated by a low, parasagittal ridge seen in other bothremydids. The processus retroarticularis is short and broad, as in Cearachelys , broader than in Rhothonemys and much broader and shorter than in Bothremys maghrebiana . The foramen posterius chorda tympani is either a true foramen or a notch lateral to the processus retroarticularis; breakage prevents being sure which is the case.

PREARTICULAR

Preservation: A nearly complete prearticular is present only on the right side of MC M2118; the anterior part of the bone is on the left ramus.

Contacts: The prearticular in Foxemys contacts the coronoid anterodorsally and the articular posteriorly, although no articular sutures are present. The ventral contact with the angular is only at the posterior end of the prearticular, as in the other Bothremydini and Rhothonemys . This is in contrast to the longer contact seen in Cearachelys and Kurmademys .

Structures: The prearticular in Foxemys is very similar to that bone in Bothremys and Araiochelys . In these taxa the prearticular exposes the anteroventral part of the fossa meckelii, in contrast to Cearachelys and Kurmademys , which have the fossa covered by the prearticular. Nonetheless, the Foxemys prearticular is slightly larger than in B. maghrebiana and has a small foramen anterodorsally near the foramen alveolare inferius. However, the foramen intermandibularis caudalis is also present, as in Cearachelys and Kurmademys and in contrast to other Bothremydini .

Araiochelys hirayamai (figs. 237, 238)

MATERIAL AVAILABLE: THUg 3338, a nearly complete lower jaw.

DENTARY

Preservation: A complete dentary is present.

Contacts: As in Bothremys maghrebiana .

Structures: The dentary in Araiochelys is closest in morphology to that in Bothremys maghrebiana , among the known lower jaws. There is a well-developed pit formed medially by a high lingual ridge and laterally by the labial ridge. The pit and the entire jaw ramus are narrower than in B. maghrebiana , or any other Bothremydini , by at least one-half. The relative height of the lingual ridge and the thickness of both ridges are comparable to those in B. maghrebiana , so Araiochelys is not simply a less ossified version of B. maghrebiana . The horizontal part of the triturating surface is a curved trough in Araiochelys , not flat, as in B. maghrebiana and B. cooki . The symphyseal wedge, low in B. maghrebiana , is much thinner in Araiochelys , although its extent on the symphysis is the same. Due to the narrower triturating surfaces in Araiochelys , the area posterior to the lingual ridges along the front of the symphyseal groove is wider than in B. maghrebiana . On the posterior and medial surface of the dentary, the sulcus cartilaginis meckelii does not meet on the midline as it does in B. maghrebiana .

ANGULAR

Preservation: Both angulars are present and nearly complete.

Contacts: As in Bothremys maghrebiana . The posterior sutures of the angular, articular, and surangular are unclear, however.

Structures: The angular in Araiochelys is similar to that bone in B. maghrebiana .

SURANGULAR

Preservation: The surangular is complete on both sides, although the sutures are not entirely clear on the right side.

Structures: The articular in Araiochelys is very similar to that in B. maghrebiana .

PREARTICULAR

Preservation: Both prearticulars in TH- Ug 3338 are present but have some damage to their medial surfaces.

Contacts: As in B. maghrebiana .

Structures: The prearticular in Araiochelys is very similar to that in B. maghrebiana .

Bothremys cooki (fig. 239) MATERIAL AVAILABLE: AMNH 2521, a lower jaw lacking the posterior ends, part of type of Bothremys cooki Leidy, 1865 (pl. 18, figs. 5, 8), described in Hay (1908: fig. 97, pl. 23, fig. 3) and in Gaffney and Zangerl (1968: fig. 22).

Structures: The surangular of Araiochelys is similar to that bone in Bothremys maghrebiana . The extent of surangular lying on the lateral side of the processus retroarticularis in Araiochelys is less than that in B. maghrebiana .

CORONOID

Preservation: Both coronoids are complete, although the right one is slightly damaged in THUg 3338.

Contacts: As in Bothremys maghrebiana .

Structures: The coronoid in Araiochelys is very similar to that in B. maghrebiana .

ARTICULAR

Preservation: The left articular is complete; the right one is missing some of the processus retroarticularis.

Contacts: As in Bothremys maghrebiana .

DENTARY

Preservation: The dentary and other bones of the lower jaw in AMNH 2521 are pitted and covered in places with a thin layer of plaster, presumably applied in the days of Leidy as a preservative. Both in 1966 and at the present the senior author has been unable to find any damn sutures. However, there are some cracks on the ventral surface and below the processus coronoideus in the right place for sutures. The right dentary has some pitting near the symphysis and the labial edge, and the left one has cracks and some lateral pits. The symphyseal damage is on the surface; the original morphology does not appear to have been affected significantly. On the ventral surface, the pitting is deeper and more extensive, particularly on the right side.

Contacts: Sutures are not visible.

Structures: The jaw in Bothremys cooki is most similar to the jaw of B. maghrebiana . The high lingual ridge and rising labial ridge forming the large cone-shaped pit are very similar. The extent that the pit extends posteriorly under the coronoid is the same. Where the labial and lingual ridges form the edges of the pit, they are slightly more flared in B. cooki than in B. maghrebiana .

The only prominent difference between the two jaws is in the symphyseal area. In B. cooki the lingual ridge is low and not defined anteriorly because of a thick wedge of bone medially. In B. maghrebiana , the lingual ridge is clearly defined, meeting on the midline the jaw in B. cooki ends in an obtuse angle with a terminal point, different from the rounded jaw margin in other Bothremydini . This point seems to be original; there is no evidence that it is a postmortem preservational or preparational artifact.

On the medial surface of the dentary, the sulcus cartilaginis meckelii does not appear to extend to the symphysis and join, as in B. maghrebiana . The sulcus is not well preserved, but the foramen alveolare inferius is visible on both sides.

ANGULAR

Preservation: The anterior part of the angular is present on both sides in Bothremys cooki , although much of this area is covered by plaster or is missing on the left side.

Contacts: Except for a suspicious crack on the ventral surface of the right ramus, there are no visible sutures for the angular.

Structures: The angular in Bothremys cooki forms the ventral margin of the sulcus cartilaginis meckelii and is similar to that area in B. maghrebiana .

SURANGULAR

Preservation: The anterior part of the surangular is present on both sides in Bothremys cooki , although the left side is more damaged and has a layer of plaster on it.

Contacts: Sutures are not visible.

Structures: The surangular in Bothremys cooki forms the lateral margin of the fossa meckelii. The anterior part of the dorsal opening of the fossa is visible on both sides in B. cooki . The area of the bone is very similar to that in B. maghrebiana .

anteriorly to define a posteromedial concavity, absent in B. cooki . This symphyseal wedge also occurs in UA 8708, a partial lower jaw from Madagascar (fig. 247; see also Gaffney and Forster, 2003). The anterior margin of CORONOID

Preservation: Most of both coronoids seem to be present in Bothremys cooki . The processus coronoideus is broken on the left side and damaged medially on the right.

Structures: The processus coronoideus of Bothremys cooki is higher and narrower in lateral view than in B. maghrebiana , Chedighaii , and Araiochelys . None of the bothremydid jaws, including AMNH 29989, has a high processus coronoideus.

ARTICULAR

Preservation: The articular is missing in AMNH 2521.

PREARTICULAR

Preservation: The anterior part of both prearticulars is present in Bothremys cooki ; the right one has more bone, however.

Contacts: Sutures are not visible.

Structures: As in Bothremys maghrebiana , the prearticular of B. cooki shows an anteriorly open fossa meckelii, although the area is not well preserved.

Bothremys maghrebiana (figs. 240, 241)

MATERIAL AVAILABLE: AMNH 30522, right ramus and symphysis nearly complete, left ramus lacking labial ridge and processus retroarticularis.

DENTARY

Preservation: The dentary in AMNH 30522 is nearly complete on the right side, missing only the tip of the symphysis. On the left side the lateral one-third of the dentary is missing from the symphysis to the coronoid.

Contacts: The dentary in Bothremys maghrebiana contacts the coronoid posterodorsally via broad sutures both medially and laterally where the two elements join in the formation of a deep pit. It contacts the surangular posterolaterally in a relatively short vertical suture ventral to the coronoid and a long horizontal ventral suture where the dentary extends posteriorly between the surangular and the angular. The dentary contacts the angular posteroventrally along a long suture. A narrow surangular contact with the dentary in Bothremys maghrebiana is visible in the medial wall of the sulcus cartilaginis meckelii.

Structures: As in all Bothremydidae , the two rami of the dentary in Bothremys maghrebiana are fused in a long symphysis with no evidence of a suture. Dorsally, the posterior extent of the dentary is limited by the coronoid on the triturating surface; the coronoid makes up the posterior part of this surface. Ventrally, the dentary extends far more posteriorly, between the surangular and angular. It reaches posteriorly to the level of the anterior edge of the area articularis mandibularis. Within the sulcus cartilaginis meckelii the dentary can be seen to extend posteriorly below the fossa meckelii.

The labial ridge of the dentary in Bothremys maghrebiana is an anteriorly facing horizontal ridge, except where it rises vertically at its posterior end to the coronoid process. This very low anterior edge gives the jaw an open, plowlike profile. The lingual ridge is high adjacent to the coronoid, but becomes lower anteriorly as it approaches the symphysis. The two lingual ridges meet in a narrow but well-defined U-shaped ridge that opens posteriorly. This U-shaped structure extends anteriorly nearly to the tip of the symphysis. The symphysis in Bothremys maghrebiana is deeply concave between the lingual ridges, forming a moderately thick symphyseal wedge, thicker than in Araiochelys .

The large pits are the most obvious feature of the lower jaws, with the whole anterior half of the jaw being involved in their formation. The triturating pit in Bothremys has been described and commented on by Leidy (1865), Baur (1891), Hay (1908), and Gaffney and Zangerl (1968). The pit is formed mostly by the dentary. There is no distinct labial ridge anteriorly; the surface is essentially horizontal. Posteriorly the labial margin curves upward and forms the lateral side of the conical-shaped pit. The lingual ridge extends posteriorly and dorsally, forming the medial edge of the pit. The coronoid bridges the two ridges and forms most of the roof of the pit. The pit in Bothremys maghrebiana is very similar to that in B. cooki . The pit in Araiochelys is much narrower and occupies relatively less of the triturating surface. However, the lower jaw of AMNH 29989, unassociated with a skull, is a massive expression of the morphology seen in Bothremys . The lower jaws of Chedighaii have pits similar to Bothremys , and the species barberi was placed in Bothremys by Gaffney and Zangerl (1968) on the basis of the lower jaw pits.

The anterior tip of the symphysis in B. maghrebiana is missing, but enough is preserved to show that a symphyseal hook was absent. In posterior view the dentary can be seen to have an open sulcus cartilaginis meckelii that remains open throughout its length to the symphysis, in contrast to Chedighaii and Araiochelys .

The foramen intermandibularis medius and foramen alveolare inferius are not defined by bone, apparently because the sulcus foramina suggest that it extended posteriorly over the dentary pit and at least to the anterior edge of the coronoid. The coronoid makes up the entire dorsal quarter of the dentary pit and all of the coronoid process. Therefore, it seems unlikely that the dentary itself served as a site of insertion of the jaw adductor musculature.

Although the dentary of Bothremys maghrebiana is similar to that of the type of the genus, B. cooki , the most apparent difference is the U-shaped area formed by the lingual ridges meeting on the symphysis, the symphyseal wedge. In B. cooki the symphyseal wedge is much thicker than in B. maghrebiana , which has an excavated concavity between the ridges rather than bone that is as high as the lingual ridges as in B. cooki . The dentary of B. maghrebiana differs further from other bothremydids in having a sulcus cartilaginis meckelii that is open to the symphysis. In other bothremydids this sulcus closes posterior to the symphysis.

cartilaginis meckelii is widely open between the coronoid and prearticular dorsally and the angular ventrally. A well-developed foramen dentofaciale majus is visible on the right dentary in Bothremys maghrebiana , ventral to the labial ridge at a point just anterior to the sutures of the surangular and coronoid.

The limits of the rhamphotheca are difficult to determine on this specimen of Bothremys maghrebiana . Enlarged nutritive ANGULAR

Preservation: The angular of Bothremys maghrebiana is completely preserved and undistorted on the right side of the lower jaw. On the left side it is nearly complete, missing only the portion that makes up the retroarticular process.

Contacts: The angular in Bothremys maghrebiana contacts the dentary all along its medial surface. Ventrally there is a long diagonal suture between these elements that extends from below the anterior edge of the area articularis mandibularis nearly to the symphysis. A dorsal suture between these elements marks the ventral limit of the sulcus cartilaginis meckelii. The angular forms a suture with the prearticular dorsally. This suture is horizontal just posterior to the closure of the sulcus cartilaginis meckelii, but it turns vertically on the medial surface of the area articularis where it contacts the articular. The angular meets the surangular posteroventrally in a long suture that extends from below the anterior edge of the area articularis to the posterior end of the long retroarticular process. This suture continues dorsally and then anteriorly to the posterior edge of the articular.

meckelii in Bothremys maghrebiana . Howev- er, since this sulcus is widely open, neither the foramen intermandibularis caudalis nor the foramen intermandibularis oralis is defined by bone. Posterior to the sulcus this element expands medially to support the medial side of the articular. The angular extends posteriorly along the ventral margin of the processus retroarticularis. The angular of Bothremys maghrebiana is nearly identical to the anterior portion of this element that is preserved in B. cooki . The angular is slightly thicker in B. cooki and contacts the prearticular more anteriorly. The posterior parts of the jaw of B. cooki are not preserved.

Structures: The angular forms most of the ventral margin of the sulcus cartilaginis

SURANGULAR

Preservation: The surangular of Bothremys maghrebiana is complete on the right side and nearly so on the left. On the left side only the portion posterior to the area articularis is missing.

Contacts: The surangular of Bothremys maghrebiana has a short anterior contact with the dentary just posterior to the foramen dentofaciale majus. It also has a long ventral suture with the dentary that extends posteriorly to a point below the anterior edge of the area articularis mandibularis. It has an anterodorsal suture with the coronoid that extends from near the foramen dentofaciale majus to the fossa meckelii. The surangular meets the angular posteroventrally where these elements join to form the ventral margin of the processus retroarticularis. Posteromedially the surangular meets the articular, also on the processus retroarticularis.

Structures: The surangular of Bothremys maghrebiana forms the lateral wall of the fossa meckelii. There is no foramen nervi auriculotemporalis visible on the lateral surface of this element. The foramen nervi auriculotemporalis is absent in Bothremydini and Taphrosphyini , but it is present in Cearachelys and Kurmademys , as well as pelomedusids, Araripemys , euraxemydids, and chelids. The surangular does not contribute directly to the area articularis mandibularis, but it supports the articular laterally. It also lies on the lateral wall of the processus retroarticularis, but it does not contribute significantly to this structure.

The surangular in Bothremys maghrebiana is a relatively large bone, extending anteriorly below the coronoid and below the processus coronoideus. This is also the case in the other Bothremydini and Rhothonemys , one of the Taphrosphyini . In Kurmademys and Cearachelys , the surangular is more extensive anteriorly than in Euraxemys and pelomedusids, but not to the degree seen in Bothremydini and Taphrosphyini . In Cearachelys , Kurmademys , chelids, pelomedusids, Araripemys , Euraxemys , and podocnemidids, the dentary extends posteriorly on the lateral surface of the jaw to separate the surangular and angular. In Bothremydini and Taphrosphyini the dentary is less extensive and more restricted to the ventral surface.

CORONOID

Preservation: The coronoid of Bothremys maghrebiana is completely preserved on the right side but is missing the anterolateral part that articulates with the lingual ridge of the dentary on the left side.

Contacts: The coronoid in Bothremys maghrebiana contacts the dentary anteriorly where it forms the dorsal part of the triturating pit. It broadly contacts the dentary both on the lingual ridge medially and on the labial ridge laterally. It contacts the surangular posterolaterally from just posterior to the foramen dentofaciale majus to the anterior end of the fossa meckelii. It also contacts the prearticular posteromedially between the fossa meckelii and the sulcus cartilaginis meckelii.

Structures: The coronoid in Bothremys maghrebiana forms all of the low, rounded processus coronoideus. It also forms a narrow part of the anterior end of the fossa meckelii between the surangular and the prearticular, and it contributes significantly to the mandibular triturating surface by forming the overhanging, posterodorsal one-third of the large, deep pit.

The coronoid of Bothremys maghrebiana is smaller than that in B. cooki . The coronoid extends farther anteroventrally in B. maghrebiana , but it has a very much lower coronoid process. It is more similar in size, height, and contacts to the coronoid of Araiochelys . The coronoid of Foxemys is smaller and lower than that in Bothremys and Araiochelys .

ARTICULAR

Preservation: The articular of Bothremys maghrebiana is completely preserved on the right side of the jaw, but it is missing its lateral half on the left side.

Contacts: The articular contacts the surangular laterally between the fossa meckelii and the processus retroarticularis. It contacts the angular medially and ventrally and the prearticular anteromedially. It is not possible to see any contact between the articular and the dentary.

Structures: The articular of Bothremys maghrebiana forms all of the area articularis mandibularis. The area is nearly round, being convex dorsally with an anteroposterior ridge in the middle. The articular also forms the posterior part of the fossa meckelii between the surangular and prearticular. It forms nearly all of the processus retroarticularis. A small foramen posterius chorda tympani is visible on the posterior suture between the articular and angular at the anterior end of the processus retroarticularis.

The area articularis mandibularis of Bothremys maghrebiana can be compared to only a few other bothremydids that have this element preserved. It is round in Chedighaii barberi (FMNH PR 247, the type) and Araiochelys , as well as in B. maghrebiana . In Rhothonemys the area articularis mandibularis is oval-shaped, with the long axis of the oval being oriented about 45 ° medial to the midline.

The processus retroarticularis in Bothremys maghrebiana is formed almost entirely by the articular, with a narrow layer of the angular on the ventral surface. The processus is long in B. maghrebiana , about as long as in Araiochelys . Both have a shallow concavity on the dorsal surface. In Rhothonemys the processus retroarticularis is shorter and broader, similar to that in Cearachelys and Kurmademys . In chelids, pelomedusids, Euraxemys , and Araripemys , the processus is very short or absent. Chedighaii barberi is not well preserved, but it has a large, broad processus, similar to that in Rhothonemys .

PREARTICULAR

Preservation: The prearticular of Bothremys maghrebiana is well preserved and complete on both sides.

Contacts: The prearticular contacts the coronoid anteromedially, the articular posteromedially, and the angular posteroventrally.

Structures: The prearticular in Bothremys maghrebiana forms the medial wall of the fossa meckelii and the posterodorsal limits of the sulcus cartilaginis meckelii and the foramen intermandibularis medius. The sulcus is widely open and, as in the other Bothremydini and Taphrosphyini , the foramen intermandibularis oralis and foramen intermandibularis caudalis are not defined by bone. The prearticular of Bothremys maghrebiana is comparable to that of Araiochelys and B. cooki insofar as the latter is preserved. In these forms the extent of contact between the prearticular and angular is reduced relative to that seen in Araripemys , Kurmademys , and Cearachelys . This contact is reduced to the point that the sulcus cartilaginis meckelii closes below the fossa meckelii; in the latter genera the sulcus cartilaginis meckelii closes below the coronoid process.

SPLENIAL

Preservation: The anteromedial part of the mandible of Bothremys maghrebiana is very well preserved and complete on both sides. It can be said confidently that it does not have a splenial.

Chedighaii barberi (figs. 242–244)

(See Note Added in Proof) MATERIAL AVAILABLE: FMNH PR 247, complete lower jaws (fig. 242) with badly eroded surface and some edges missing, associated with partial skull and shell, jaw described and figured in Gaffney and Zangerl (1968: fig. 22); ALAB PV 2001.2, right ramus missing symphysis and posterior end, left ramus missing posterior end and part of lingual ridge (figs. 242–244), associated with skull and partial shell; CSU K-90-6-2, jaw symphysis; NJSM 12704, cast (Denton Collection) of damaged right ramus.

DENTARY

Preservation: Only the symphysis and the anterior part of the right ramus are preserved in CSU K-90-6-2. The bone surface is only slightly eroded and cracked; most of it seems to be the original surface. In FMNH PR 247 the dentary is nearly complete, lacking the anterior end, but almost the entire bone surface has been eroded. The left dentary of ALAB 2001.2 is missing the midline, the anterior part of the lingual ridge, and a small part of its posterior end, but the bone surface is well preserved. In the right ramus, the entire anterior half is missing, as is the posterior end, but the remainder is well preserved.

Contacts: Sutures are clear in ALAB 2001.2 and most are visible on one side or the other in FMNH PR 247. The dentary contacts in Chedighaii barberi are very similar to those in Bothremys maghrebiana .

Structures: The dentary of Chedighaii barberi that can be seen in CSU K-90-6-2 and ALAB 2001.2 has an anterior, flat surface, which is the triturating surface, marked by many nutrient foramina, and a posterior, smooth surface, with a shallow concavity, the area of the symphyseal wedge. The concavity is defined anteriorly and laterally by the lingual ridge; posteriorly there is no margin. The symphysis in CSU K-90-6-2 is very similar to that in ALAB 2001.2, which is also flat anteriorly with a posterior concavity. The anterior margin seems slightly more acute in ALAB 2001.2, but it is not complete, so this may be misleading. The triturating surface on the symphysis is slightly wider anteroposteriorly in CSU K-90-6-2 than in ALAB 2001.2. In both the area is about half the length of the concavity behind it.

The labial ridge in Chedighaii barberi is very low anteriorly, but the lingual ridge shows a dorsal rise, as in Bothremys cooki and B. maghrebiana . The symphyseal concavity in all the Chedighaii jaws does not form the thickened symphyseal wedge seen in B. cooki , and, to a lesser extent, B. maghrebiana . FMNH PR 247 seems to differ from CSU K-90-6-2 and ALAB 2001.2 in lacking a midline union of the lingual ridges, but most of this jaw is worn by abrasion (the anterior part particularly), and the anterior lingual ridges are very low, probably due to postmortem damage.

On the ventral surface of CSU K-90-6-2, the nutrient-rich area covered by the horny beak is differentiated from the smooth posterior surface by a very low ridge that thickens laterally. The same ridge is in ALAB 2001.2, but FMNH PR 247 seems to be smooth, although this may be due to damage, as the surface is clearly eroded. Bothremys cooki and B. maghrebiana do not have this ridge where the foramina-rich surface changes to a smooth surface. On the medial surface of the symphysis in Chedighaii barberi , the anteriormost part of the sulcus meckelii extends close to the symphyseal area, but it does not meet the other sulcus on the midline, as in Bothremys maghrebiana .

Posteriorly, the lingual ridge in Chedighaii barberi rises to form the medial wall of the triturating pit, as in Bothremys . The labial ridge is only the thickened margin of the triturating surface as it extends posteriorly, until it rises abruptly to form the lateral wall of the pit, as in B. maghrebiana as well. In FMNH PR 247 the ridges and pit are very similar to B. maghrebiana except for a greater thickness of the bone. In ALAB 2001.2, however, the lingual ridge wall is much thicker than the labial ridge wall and the pit is not as deep as in FMNH PR 247, which has the pit wall equal in thickness. The jaw width is narrower in ALAB 2001.2 than it is in FMNH PR 247. It seems best to attribute these differences to individual variation at present, in the absence of better material. NJSM 12704 is more like FMNH PR 247, but it has also been damaged by postmortem erosion. Thus, it is possible to characterize FMNH PR 247 and NJSM 12704 as having slightly wider jaws, deeper pits, and a poorly defined symphyseal concavity due to low lingual ridges, in contrast to ALAB 2001.2 and CSU K-90-6-2. It is possible that these differences represent different taxa. The incomplete nature of the specimens and the lack of supporting characters make recognizing two taxa dubious at present, and all these jaws are identified as Chedighaii barberi . Nonetheless, it should be kept in mind that ALAB 2001.2, the specimen with the narrower jaws and shallower pits, has a skull lacking pits and is the basis for moving the species barberi from Bothremys to Chedighaii . If FMNH PR 247 proves to be a jaw type that is found with a pitted skull in the future, uh oh, a change is in the wind.

ANGULAR

Preservation: Both FMNH PR 247 and ALAB 2001.2 have the angular preserved, although it is damaged on both sides. In FMNH PR 247 the bone is nearly complete but damaged on its surface.

Contacts: The angular contacts in Chedighaii barberi are very similar to those in Bothremys maghrebiana . The prearticular is missing in both Chedighaii specimens, however, and the articular-angular suture appears fused, as is often the case in pleurodire jaws.

Structures: The angular in Chedighaii is very similar to that bone in B. maghrebiana . SURANGULAR

Preservation: The anterior part of the surangular is present on both sides of ALAB 2001.2. Both surangulars are nearly complete in FMNH PR 247, but they are eroded on their surfaces.

Contacts: As in Bothremys maghrebiana . The articular suture is not discernable.

Structures: The foramen dentofaciale majus in Chedighaii barberi is in the surangular-dentary suture in ALAB 2001.2, but it seems to be in the dentary in FMNH PR 247, although poor preservation makes it hard to be sure. In Bothremys maghrebiana the foramen is well onto the dentary. The surangular forms the lateral wall of the fossa meckelii, which in shape and size seem to be very similar to B. maghrebiana . The surangular thickness, however, is much greater in C. barberi than it is in B. maghrebiana and Araiochelys . This is determinable only in FMNH PR 247. The lateral wall drops straight ventrally in B. maghrebiana , but in C. barberi the surangular is thicker dorsally than ventrally. B. cooki appears to agree with C. barberi in this, although the area is not well preserved.

CORONOID

Preservation: The coronoid is preserved on both sides in FMNH PR 247 and ALAB 2001.2.

Contacts: The coronoid of Chedighaii barberi differs slightly from Bothremys maghrebiana in having less lateral exposure. The prearticular is missing in all specimens, but the sutural surface shows its position to be very similar to that in B. maghrebiana . The dentary and surangular contacts are otherwise as in B. maghrebiana .

Structures: The processus coronoideus in Chedighaii barberi is very similar to that in B. maghrebiana , not as high as in B. cooki . The coronoid forms the roof of the triturating pit in FMNH PR 247, as in Bothremys , with the dentary extending into the floor. In ALAB 2001.2, however, the shallower pit is formed almost entirely by the coronoid, and the dentary does not extend into the pit floor.

ARTICULAR

Preservation: The articular is preserved on both sides of FMNH PR 247, although it is somewhat eroded and no sutures defining it are visible.

Contacts: In FMNH PR 247, sutures that define the articular are not visible, a common condition in turtles.

Structures: The area articularis mandibularis in Chedighaii barberi is eroded and pitted, but it appears to be the same shape as in Bothremys maghrebiana , rounded with PREARTICULAR

Preservation: The prearticular is absent in all Chedighaii barberi specimens, but the sutural contacts in FMNH PR 247 suggest a bone very similar to that in Bothremys maghrebiana .

TRIBE BOTHREMYDINI

Genus indeterminate

MATERIAL AVAILABLE: AMNH 29989, lower jaw lacking posterior ends (figs. 245, 246), from the Ypresian Eocene Couche 1 phosphates of Ouled Abdoun, Morocco. This specimen cannot be physically fitted into the available skull material of Bothremys kellyi or other known taxa and probably represents an undescribed species.

a low anteroposterior ridge. The processus retroarticularis seems to be complete on the left side. It is shorter, broader, and more massive than in B. maghrebiana , similar to the one in Rhothonemys .

DENTARY

Preservation: The dentary in AMNH 29989 is nearly complete.

Contacts: The dentary in AMNH 29989 has similar contacts to those in Bothremys maghrebiana , but the coronoid contact is much smaller on the lateral jaw surface. Laterally, the dentary has a long, vertical suture with the surangular, restricting the coronoid contact dorsally.

Structures: The paired pits and anteriorly open fossa meckelii in AMNH 29989 suggest a close relationship to the Bothremydini , but the massiveness and detailed triturating surface shape show that AMNH 29989 is a new taxon not represented by known cranial material.

The triturating surface in AMNH 29989 has the pit formed by labial and lingual ridges. The pit is about the same relative size as in Bothremys , although it is higher than wide rather than equidimensional as in Bothremys . The lingual ridge that forms the medial wall of the pit is much thicker than it is in Bothremys . The anterior continuation of the lingual ridge curves laterally and meets the labial ridge rather than meeting on the symphysis as in B. maghrebiana and Araiochelys . The area between the lingual ridges forms a deep concavity with a low transverse ridge posteriorly and an upturned labial margin anteriorly. This concavity is the symphyseal wedge, which in other Bothremydini is not as deep and is bordered by the lingual ridges anteriorly. The symphysis in AMNH 29989 is much longer than in other bothremydids. It extends posteriorly to the level of the processus coronoideus, about twice the length of the symphysis in Bothremys .

The ventral surface of the dentary in Bothremys is relatively flat, but in AMNH 29989 it is deep posteriorly, where the bone is very thick, and tapers anteriorly to the labial margin. The symphyseal ventral surface is deeper and convex rather than flat as in Bothremys . The sulcus cartilaginis meckelii is relatively broad in AMNH 29989, related to the much higher and more massive jaw, compared to Bothremys . The sulcus does not reach the symphysis, as in B. maghrebiana . The foramen alveolare inferius and the contact with the surangular can be seen on the lateral wall of the fossa meckelii.

In contrast to other Bothremydini, AMNH 29989 has a highly vascularized triturating surface with nutrient foramina varying in size. The rhamphotheca-covered area does not have distinct margins, agreeing with other Bothremydini .

The foramen dentofaciale majus is formed entirely by the dentary and is contained in a shallow depression on the lateral surface of the dentary. There are what appear to be at least two foramina leading into the canalis alveolaris inferior from this depression.

ANGULAR

Preservation: The anterior part of the angular is preserved on the right side in AMNH 29989; only a small part is preserved on the left one.

Contacts: The angular in AMNH 29989 contacts the dentary in a short suture, rather than a long one as in Bothremys maghrebiana . The prearticular contact is more dorsal due to the greater height of AMNH 29989.

Structures: The angular in AMNH 29989 is deeper and shorter than in Bothremys maghrebiana , but the restricted prearticular contact still results in a relatively open fossa meckelii.

SURANGULAR

Preservation: The surangular is present on both sides of AMNH 29989, but both lack their posterior portions, broken off behind the fossa meckelii.

Contacts: The dentary contact is extensive and long, compared to Bothremys maghrebiana . The coronoid contact is much smaller than in B. maghrebiana .

Structures: AMNH 29989 is a deep and short jaw compared to Bothremys , and the surangular and dentary form most of the surface in lateral view. The relations of the surangular, however, are the same as in B. maghrebiana : it forms the lateral wall and dorsal opening of the fossa meckelii.

CORONOID

Preservation: Both coronoids in AMNH 29989 are complete.

Contacts: The coronoid in AMNH 29989 extends farther anteriorly along the lingual ridge and does not reach the labial ridge, in contrast to Bothremys . The medial exposure of the coronoid is less also because the dentary suture is placed more posteriorly. Laterally, the coronoid is much less extensive than in other Bothremydini because of the dorsal position of the dentary and surangular contacts.

Structures: Despite its massive form, the processus coronoideus is slightly lower than in B. maghrebiana and much lower than in B. cooki .

ARTICULAR

Preservation: A small part of articular in AMNH 29989 is exposed as a broken edge on the right side; otherwise, the bone is missing.

PREARTICULAR

Preservation: Only the right prearticular is present in AMNH 29989, and it is missing its posterior margin.

Contacts: The prearticular in AMNH 29989 contacts the coronoid anterodorsally and the angular posteroventrally.

Structures: The prearticular in AMNH 29989 is a higher, shorter bone than it is in Bothremys , and it is thicker and more massive. It forms the medial wall of the fossa meckelii. The fossa is more open than in Euraxemys , pelomedusids, and chelids, agreeing with other Bothremydini and Taphrosphyini .

TRIBE TAPHROSPHYINI Rhothonemys brinkmani (figs. 248, 249)

MATERIAL AVAILABLE: AMNH 30521, lower jaw lacking coronoid areas and both prearticulars, associated with skull.

DENTARY

Preservation: Nearly all of the left dentary in AMNH 30521 is present; the right one lacks most of its posterior contacts and is separated from the posterior jaw elements without good contacts. A posteroventral piece of dentary is present in the right posterior jaw piece.

Contacts: The dentary in Rhothonemys has only one piece of coronoid remaining; most of the coronoid attachment area is either broken or a sutural surface. The coronoid contact remaining is on the dorsomedial side of the dentary, just above and lateral to the fossa meckelii. Although incomplete, the coronoid contact in Rhothonemys appears to be less extensive medially and laterally compared to Bothremys maghrebiana ; it seems to have been similar to that in Euraxemys . The surangular contact is more anterior than in Euraxemys and is similar to Bothremydini . In Rhothonemys this suture runs along the posterior edge of the foramina-rich surface for the rhamphotheca, in contrast to other Pelomedusoides. The angular contact is similar to Bothremydini as well. The prearticular is missing.

Structures: The dentary in Rhothonemys is fused at the symphysis, as in all other bothremydids. The dentary does not extend posteriorly to the degree seen in Euraxemys ; it is similar to other bothremydids. The lateral side shows the foramen dentofaciale majus as a large foramen formed by the dentary anteriorly and the surangular posteriorly. The foramina-rich lateral surface extends to the posterior edge of the dentary; the surangular-dentary suture marks the limits of the rhamphotheca. This is in contrast to other bothremydids in which the rhamphotheca ends anterior to the edge of the dentary. The nutrient canal-rich area does not extend onto the ventral surface of the dentary.

The triturating surface in the lower jaw of Rhothonemys is very unusual for pleurodires in general and for bothremydids in particular. The dentary is roughly parallel-sided; it does not widen posterolaterally, and it is narrow. The triturating surface, as determined by foramen-rich bone and lingual and labial ridges, is very narrow, less than one-third the width of the dentary in dorsal view. The labial ridge is distinct and forms a low hook at the symphysis, but it nearly disappears posteriorly, becoming barely identifiable at the posterior edge of the dentary. The lingual ridge, very low at the symphysis, rises posteriorly so that it is the dorsalmost part of the jaw posterior to the symphysis. The lingual ridge is still blunt and low, and it barely rises above the level of the rest of the jaw. The triturating surface of Rhothonemys is very different from that in the Bothremydini . The surface is unusually narrow (rather than wide), it does not rise dorsally to a high processus coronoideus, and there is a symphyseal hook. The only similarity to Bothremydini is the lingual ridge that rises posteriorly to be dorsal to the labial ridge.

A possible tribe Taphrosphyini lower jaw is described in Bardet et al. (2000: 281, fig. 7d, e) as a ‘‘Chelonioidea gen. and sp. indet’’. Associated with this jaw are shell elements also described and figured ( Bardet et al., 2000: 281, fig.7a–c, g), one of which (fig. 7a) has the iliac scar small, round, and at the shell margin, features probably diagnostic for the tribe Taphrosphyini . The lower jaw is very similar to those of Rhothonemys and Taphrosphys congolensis (see figs. 248–250).

The dentary surface medial to the lingual ridge in Rhothonemys is smooth and slopes ventromedially, with a slightly concave shape. This surface is about three times wider than the triturating surface itself. On the medial surface, the dentary forms the sulcus cartilaginis meckelii, which is a sharply defined, V-shaped trough similar to that in other bothremydids, but not meeting on the midline as in Bothremys maghrebiana . The foramen alveolare inferius in Rhothonemys lies near the posterior end of the dentary on the medial surface in the region where the fossa meckelii grades into the sulcus cartilaginis meckelii. It is exposed on the right side, but on the left it is behind the displaced fragment of coronoid. Presumably, it was originally visible in medial view.

ANGULAR

Preservation: Both angulars are present in AMNH 30521, but neither is complete. The right one is complete posteriorly but is missing its anterior third. The missing part can be determined from the matching sutural surface on the dentary. The left angular is more complete, but its anterior third is broken and slightly displaced from its dentary contact.

Contacts: The anterior dentary contact in Rhothonemys is clear and similar to that in other bothremydids. It forms a V-shaped process of the angular lying below the sulcus cartilaginis meckelii. The prearticular is missing in AMNH 30521, but the angular and articular show a short, sutural contact for it, similar in position to that in the Bothremydini . The angular merges into the articular with no sign of a suture in AMNH 30521, as in many adult pleurodires. Posteroventrally, the angular in Rhothonemys has a long contact with the surangular, also as in other pleurodires.

Structures: The angular in Rhothonemys is a splintlike element forming the ventromedial surface of the jaw, below the fossa meckelii and sulcus cartilaginis meckelii. In Rhothonemys the angular has a short contact with the prearticular dorsally, with a free edge exposing most of the ventral part of the fossa meckelii, all as in Bothremys maghrebiana and other Bothremydini and in contrast to other pleurodires. The angular in Rhothonemys curves dorsomedially to form the floor of the fossa meckelii. Because the fossa meckelii is relatively open anteriorly, the foramen intermandibularis medius is placed posteriorly and the foramen intermandibularis caudalis is not formed, all as in Bothremys maghrebiana .

SURANGULAR

Preservation: The left surangular in AMNH 30521 is nearly complete, lacking only some of the dorsal margin. The right surangular is less complete, lacking its anterior and anterodorsal margins.

Contacts: The surangular in Rhothonemys contacts the dentary anteriorly, the angular posteroventrally, and the articular posteromedially. The coronoid is largely missing in AMNH 30521, and the surangular-coronoid contact surface is damaged.

Structures: The surangular in Rhothonemys is a large, sheetlike element at the posterior end of the jaw, forming the lateral wall of the fossa meckelii. As preserved, the fossa is similar to that in Bothremys maghrebiana , but the dorsal opening of the fossa is not preserved, so its shape is indeterminate. Posteriorly the surangular lies lateral to the blocklike articular, and in Rhothonemys it barely extends onto the area articularis mandibularis. The surangular extends posteriorly onto part of the processus retroarticudentary, it indicates a posterodorsally sloping dentary contact in Rhothonemys , in contrast to the more vertical one of Bothremys maghrebiana .

ARTICULAR

Preservation: Both articulars are present and seem to be complete in Rhothonemys . Due to the fusion of some sutures, however, this is not certain.

Contacts: The articular in Rhothonemys contacts the surangular laterally and is fused with the angular ventrally and anteroventrally. The prearticular is missing, but a sutural surface shows a medial contact, as in other pleurodires.

Structures: The articular forms nearly all of the area articularis mandibularis in Rhothonemys . The surface is roughly divided into two facets by a very low anteroposterior ridge, as in Bothremys maghrebiana . The medial facet is larger and more anterior than the lateral one. In B. maghrebiana they are more equal in size.

The articular in Rhothonemys forms nearly all of the processus retroarticularis; the angular contribution is uncertain due to sutural fusion. The processus is large in Rhothonemys , much larger than in Euraxemys and the Pelomedusidae . In Rhothonemys it is shorter, broader, and not concave dorsally, in contrast to Bothremys maghrebiana . There is no foramen posterius chorda tympani. The articular forms the posterior wall of the fossa meckelii.

laris, as in other bothremydids. There is no foramen nervi auriculotemporalis.

CORONOID

Preservation: Only a fragment of the coronoid is present anteriorly on the medial surface of the left ramus in AMNH 30521. It is slightly displaced ventrally. Along with the contact area on the medial surface of the PREARTICULAR

Preservation: The prearticular is missing in AMNH 30521, but a sutural surface on the articular and angular shows that one was present.

EPIFAMILY PODOCNEMIDINURA

Hamadachelys escuilliei MATERIAL AVAILABLE: AMNH 30029, left ramus (fig. 251), broken off approximately at or near the symphysis. This unassociated lower jaw is identified as Hamadachelys . Originally suspected to be the lower jaw of Galianemys , further preparation of Hamadachelys material shows that it belongs to this contemporary genus. We retain it for comparison with bothremydids.

DENTARY

Preservation: The left dentary in AMNH 30029 lacks the posterior ventral process and is preserved up to what seems to be the midline, although it is hard to be sure.

Contacts: The dentary in AMNH 30029 contacts the coronoid posterodorsally, the surangular posteriorly, but more dorsally than in Cearachelys or Kurmademys , and the angular posteroventrally. The surangular contact is vertical dorsally, but it becomes horizontal midway in its height, rather than at the bottom edge.

Structures: The symphyseal area as preserved in AMNH 30029 seems to be very similar to Kurmademys , particularly ISI 155D, which has a symphyseal hook. The hook is likely in AMNH 30029 because the labial ridge is rising at its broken edge. The narrow triturating surface, defined posteriorly by a Ushaped concavity with a low lingual ridge as its border, is very similar to this area in Cearachelys and Kurmademys . The labial ridge is slightly higher than the lingual ridge anteriorly, as in Cearachelys , THUg 1798. Also as in this Cearachelys specimen, the triturating surface is narrower than in Cearachelys, BSP 1976 I 160, and Kurmademys , although clearly there is variation in this in both Cearachelys and Kurmademys . The lingual ridge in AMNH 30029 rises posteriorly and is higher than the labial ridge, as in bothremydids. The triturating surface is slightly concave along its length, similar to that in Kurmademys and Cearachelys , except it is not as flat anteriorly. The jaw ramus is slightly deeper and more massive than in Cearachelys , similar to Kurmademys .

The sulcus cartilaginis meckelii is similar to that in Kurmademys and Cearachelys . The foramen dentofaciale majus is at the upper edge of the ramus, in the processus coronoideus, rather than more ventrally, as in Kurmademys and Cearachelys .

ANGULAR

Preservation: The anterior half of the left angular is present in AMNH 30029.

Contacts: The angular in AMNH 30029 contacts the prearticular dorsally and the dentary anteriorly and ventrally. The prearticular contact is long, as in Kurmademys and Cearachelys .

Structures: The angular in AMNH 30029 borders the sulcus cartilaginis meckelii, as in Kurmademys and Cearachelys . The angular wraps ventrally around the jaw, as in other pleurodires.

SURANGULAR

Preservation: Only a small part of the surangular, just posterior to the processus coronoideus, is present in AMNH 30029.

Contacts: The surangular in AMNH 30029 contacts the dentary anteriorly and dorsally and the coronoid anteromedially, as in Kurmademys and Cearachelys .

Structures: The surangular forms the lateral wall and dorsal margin of the fossa meckelii, the anterior part of which is present in AMNH 30029. The dentary and surangular form the lateral side of the processus coronoideus. There is not enough of the surangular preserved to determine other features.

CORONOID

Preservation: All of the coronoid is present in AMNH 30029.

Contacts: The coronoid in AMNH 30029 contacts the dentary anteriorly, the prearticular posteroventrally, and the surangular posteriorly.

Structures: The processus coronoideus in AMNH 30029 is slightly lower than that in Kurmademys and Cearachelys , but its shape is still similar to that in those taxa. The coronoid forms part of the dorsal edge of the foramen intermandibularis medius. The coronoid does not extend onto the triturating surface to the extent seen in Kurmademys , but it does form part of the medial margin and lingual ridge as in Cearachelys .

ARTICULAR

Preservation: The articular is missing in AMNH 30029.

PREARTICULAR

Preservation: The anterior part of the prearticular is present in AMNH 30029.

Contacts: The prearticular in AMNH 30029 contacts the coronoid dorsally and the angular ventrally. Other contacts are not preserved.

Structures: The prearticular in AMNH 30029 has the anterior extent seen in Kurmademys and Cearachelys and is not short as in the Bothremydini and Taphrosphyini .

SHELL MORPHOLOGY

FAMILY EURAXEMYDIDAE

Euraxemys essweini , n. gen. et sp.

(figs. 252–254)

MATERIAL AVAILABLE: FR 4922, a nearly complete shell. The shell of Euraxemys was originally in a concretion broken in half transversely (not into part and counterpart, the usual pattern for Santana fossils), and it was repaired with a solvent-based putty that also covered the ventral surface before being acquired by the Senkenburg Museum. This putty was removed at AMNH, the entire surface cleaned, and the rearticulated shell was embedded in a clear plastic that covered the carapace. Before embedding in plastic, a cast, AMNH 30568, was made of the carapace. With the ventral surface exposed, the shell was acid prepared using techniques described in Rutzky et al. (1994). The outline and margins of FR 4922 do not appear to be distorted, as the plastron and bridge have what seem to be their natural curvature. The central area of the carapace, however, is crushed ventrally with much of the bone broken. The original curvature is retained along the posterior margin. Due to the fragmented condition of the bone of the carapace, complete removal of the matrix was halted in order to leave a layer just beneath most of the carapace. Despite the breakage, many of the carapacial sutures are visible, but the scale sulci are largely obscured except along the well-preserved posterior margin. The anterior part of the ventral surface is visible in FR 4922, but the posterior area medial to the peripherals is covered by matrix.

The published reconstruction of the shell (Gaffney and Meylan, 1991) is incorrect in a number of areas, and a new description is presented here (fig. 254).

CARAPACE

Shell surface texture is largely obscured by breakage in FR 4922, but the distal portion of left costals 1–3 has the surface preserved (particularly visible on the cast). The surface texture has the form of low ridges and troughs roughly parallel to the lateral margins of the pleural scales, a surface ornamentation commonly seen in cryptodires (cf. Ernst et al., 1994: pl. 24, Clemmys muhlenbergii ) but not often in pleurodires. Nonetheless, it is a common chelonian pattern seemingly related to growth of the pleural scales.

Although damaged, the nuchal bone can be restored because its right lateral and posterior limits are determinable. It has the common six-sided shape of most Pelomedusoides, such as Taphrosphys , but it does not have the slight nuchal embayment seen in Chedighaii or in the original reconstruction of FR 4922 (Gaffney and Meylan, 1991), which also shows the nuchal as too long. The first two peripheral bones can be made out on the left side of FR 4922, and they are very similar to most other Pelomedusoides, such as Taphrosphys . They differ significantly from Araripemys , which has greatly reduced the anterior peripherals and lost contact with the nuchal. The bridge peripherals in FR 4922 are damaged on the dorsal surface but are better preserved ventrally, particularly on the left side. The bridge peripherals do not have guttering, and there is no indication of fontanelles as in Araripemys . The posterior peripherals 8–11 and pygal are well preserved and show the original shell curvature.

Peripheral 11 has an undercut separating the external and internal surfaces on the ventral side. The body wall margin separates these two surfaces. The internal surface is concave anteriorly and is nearly vertical. On the dorsal surface the pygal and posterior peripherals have a distinct flare that curves dorsally, suggesting that the shell may have been higher domed than in Cearachelys and was similar to the higher domed Pseudemys species. The crushing has greatly altered the original curvature.

Euraxemys has eight pairs of costals, with 2–7 being parallel-sided. The first costal is expanded anteriorly, as in nearly all turtles, and the eighth costal is slightly expanded distally. Many of the costal sutures are indeterminate due to breakage, but some of their limits can be determined by assuming bilateral symmetry and restoring by transposition. Costals 3–5 are particularly poorly preserved because this is the region where the concretion containing the skeleton was broken in half transversely. None of the costals meets on the midline. The suprapygal is six-sided, contacting the eighth neural anteriorly, the eighth costals laterally, the eleventh peripherals in a narrow pair of contacts, and the pygal posteriorly. The pygal is four-sided and nearly square, as in many Pelomedusoides butin contrast to the long pygal of Araripemys and the short pygal of Notoemys .

The neurals of Euraxemys are a complete set of eight with the common Pelomedusoides condition of 4, 6, 6, 6, 6, 6, 6, 6. Although many of the neurals are not well preserved, their shapes have been determined by reconstruction that assumes bilateral symmetry. The shortest neural is probably the seventh, with the eighth being about the length of the sixth. The first neural is the common shape seen in living Pelomedusoides, that is, rectangular with slightly curved margins. Neurals 2–5 are ‘‘coffinshaped’’ hexagonal: neural 6 is wide and short, neural 7 is wider than long, and neural 8 is similar in shape to neural 2 but smaller.

The ventral surface of the carapace in FR 4922 shows some morphology, although the area largely consists of finely broken bone. The nuchal, peripherals 1–4, costal 1, and the first neural can all be made out. Although broken, the first thoracic centrum is preserved. The anterior articulation and zygapophyses are missing, but its ventral surface is flat. The first thoracic ribs are preserved and they extend sharply posterolaterally, as in pelomedusids. Distally they curve laterally and are longer than in pelomedusids. In pelomedusids and all other living pleurodires the ribs are relatively flat and more closely sutured to the first costal than in Euraxemys , which has ribs that are rounded and seem to be free of the costal for a greater length. The Euraxemys rib does not have the short anterior projection seen in Pelomedusa . The posterior contact with the second thoracic is a flat ankylosis, with the first costal rib contacting both the first and second thoracic centra as in all other pleurodires.

The scale pattern on the carapace of Euraxemys is clear around the posterior edge but is not well preserved elsewhere. There does not seem to be a cervical scale, and there is the normal set of 12 marginals as in other Pelomedusoides. Marginals 1, 2, and 8–12 are determinable and lie entirely on the peripherals. The elongate marginal 2 of the Gaffney and Meylan (1991) reconstruction was in error. This marginal is similar in size to the first. The only vertebral scales that have margins preserved are the second and the fifth. The sulci for the second vertebral are visible laterally on the right and left sides where they cross the costal 1–2 suture and anteriorly where it crosses the neural 1. The second vertebral is similar in width to that in other Pelomedusoides such as Taphrosphys . The fifth vertebral is also similar to Taphrosphys but it may be slightly wider. The new reconstruction of FR 4922 indicates wider vertebrals than in the Gaffney and Meylan (1991) figure. The pleural scales are poorly preserved, and only the pleural 1–2, 3–4, and 4–5 sulci are preserved. There are no supramarginal scales as in Proterochersis and Platychelys .

PLASTRON

The plastron of Euraxemys is made up of the usual 11 bones found in most Pelomedusoides. The anterior lobe is shorter than the posterior lobe, but it reaches the anterior edge of the carapace, similar to that in Dortoka . The size and shape of the entoplastron, epiplastra, and anterior lobe in Gaffney and Meylan (1991) are incorrect (fig. 254). There is a relatively short midline contact of the epiplastra, as in most Pelomedusoides. The entoplastron of Euraxemys is quadrangular, as in most pleurodires. It is slightly longer than wide, in contrast to Chedighaii and Taphrosphys in which it is slightly wider than long. The hyoplastron forms the axillary notch. The hypoplastron forms the inguinal notch, which is relatively open. Neither notch is as open as in Araripemys , however.

Euraxemys has paired, laterally placed mesoplastra, in contrast to Araripemys , Dortoka , and chelids, which lack them, and in common with all other Pelomedusoides. Although neither mesoplastron in FR 4922 is complete, their limits can be reconstructed. The mesoplastron is slightly larger than in such forms as Bothremys , Pelomedusa , and Podocnemis . The mesoplastron does not extend farther medially than in most Pelomedusoides. There are no fontanelles in the Euraxemys plastron, and there are no foramina for musk ducts visible.

The xiphiplastron in Euraxemys forms most of the posterior lobe, which tapers posteriorly as in Platychelys and Notoemys and in contrast to forms like Taphrosphys , which are roughly parallel-sided. The posteriorly tapering plastron of Euraxemys is nearly the only similarity it has to that of Araripemys . There is a shallow anal notch, also similar to Araripemys and Platychelys and distinct from the deeper notch commonly seen in other Pelomedusoides.

There is the common Pelomedusoides complement of 13 scales on the plastron of Euraxemys , with no indication of axillary, inguinal, or inframarginal scales. There is a large, parallel-sided, intergular overlapping onto the entoplastron completely separating the smaller gulars and partially separating the humerals. This is essentially the same pattern seen in Pelusios , Podocnemis , and bothremydids like Cearachelys , Foxemys , Chedighaii , and Polysternon , but distinct from forms like Taphrosphys , which has a larger intergular scale completely separating the humeral scales, and Pelomedusa , which has very large humerals only slightly separated by the intergular. The gular scales in Euraxemys lie entirely on the epiplastra and are triangular.

The humeral scales are separated anteriorly by the gulars and intergular and are in contact medially for about half their length. The humeral scales extend onto the entoplastron. The humeral-pectoral sulcus is more posterior in Euraxemys than in forms like Bothremys , so that the humeral scale dominates the anterior lobe margin. The pectoral scales do not overlap onto the posterior half of the entoplastron, in contrast to the reconstruction of Gaffney and Meylan (1991). The pectoral-abdominal sulcus runs along the posterior part of the hyoplastra onto the mesoplastra. The abdominal and femoral scales in Euraxemys are similar to those in Cearachelys and most Pelomedusoides, as these scales do not exhibit much variation. The femoral-anal sulcus runs anteromedially rather than straight transversely in Euraxemys .

FAMILY BOTHREMYDIDAE

TRIBE KURMADEMYDINI Kurmademys kallamedensis (figs. 255–257) MATERIAL AVAILABLE: ISIR 278, partial shell; ISIR 152A, left costal 1; ISIR 152B, left costal 4; ISIR 152C, left costal 2; ISIR 152D, right hypoplastron; ISIR 152E, hyoplastron fragment; ISIR 152F, left ninth or tenth peripheral; ISIR 152G, left tenth or eleventh peripheral; ISIR 152H, costal fragment; ISIR 152I, nuchal; ISIR 152J, right costal 3; ISIR 152K, right peripheral 3; ISIR 152L, right peripheral 8; ISIR 152M, costal; ISIR 152N, bridge peripheral; ISIR 152O, right costal 1; ISIR 152Q, left hypoplastron; ISIR 152R, left hypoplastron; ISIR 152S, right hyoplastron; ISIR 152T, right peripheral 2; ISIR 153A, left costals 1 and 2, peripherals 3 and 4; ISIR 153B, neural 3; ISIR 153C, right costal 1; ISIR 153E, right costal 1; ISIR 153F, left peripheral 7; ISIR 153G, right costal 1; ISIR 153H, right xiphiplastron; ISIR 157B, right hyoplastron; ISIR 157D, costal fragment; ISIR 157E, left costal 1; ISIR 157F, costal fragment; ISIR 157G, right? eighth peripheral; ISIR 157H, costal; ISIR 157I, buttress or peripheral; ISIR 157J, buttress; ISIR 157K, buttress; ISIR 157L, right xiphiplastron; ISIR 157M, mesoplastron; ISIR 157P, costal 5; ISIR 157Q, costal 5; ISIR 157R, right costal 1; ISIR 157U, left costal 1; ISIR 157V, costal fragment; ISIR 157W, left costal 1; ISIR 157X, costal; ISIR 157Y, peripheral; ISIR 157Z, costal; ISIR 157AA, costal; ISIR 157AB, right hypoplastron; ISIR 157AC, neural 3 or 5?; ISIR 157AD, peripheral 9 or 10?; ISIR 157AE, neural 3 or 5?; ISIR 157AF, peripheral 2; ISIR 157AG, bridge peripheral; ISIR 157AH, neural; ISIR 157AJ, neural; ISIR 157AK, peripheral; ISIR 157AL, costal; ISIR 157AM, peripheral; ISIR 157AN, costal; ISIR 157AO, left xiphiplastron; ISIR 157AP, left peripheral 6; ISIR 157AQ, left hypoplastron; ISIR 157 AS, costal 5; ISIR 157AT, bridge peripheral; ISIR 157AU, right costal 1; ISIR 157AV, costal 5; ISIR 157AW, peripheral 8; ISIR 157AX, neural; ISIR 157AY, neural; ISIR 157AZ, right costal 5; ISIR 157BA, right costal 7; ISIR 157BB, first neural; ISIR 157BC, fourth neural; ISIR 157BD, second neural; ISIR 157BE, nuchal; ISIR 157BH, left costal 5; ISIR 157BI, right peripheral 2; ISIR 157BJ, left costal 1; ISIR 157BK, right hypoplastron; ISIR 157BL, neural; ISIR 157BM, right costal 1; ISIR 157BN, left costal 5; ISIR 157BO, right xiphiplastron; ISIR 157BS, costal; ISIR 157BT, costal 1; ISIR 157BU, costal.

CARAPACE

The shell of Kurmademys has not previously been illustrated. The carapace is incomplete in ISIR 278; the anterior part is damaged. The reconstruction is also based on ISIR 157BE (nuchal), ISIR 157W (costal 1), ISIR 152T (peripheral 2), and ISIR 152K (peripheral 3). The carapace is oval in shape and apparently lower than in most bothremydids, although crushing may obscure its original depth. The shell surface shows the ‘‘pelomedusoid’’ pattern to some extent ( Broin, 1977), at least on the bridge region.

The carapace consists of one nuchal, six neurals, one suprapygal, one pygal, eight pairs of costals, and 11pairs of peripherals, although the first peripheral is not known. The nuchal is pentagonal in shape without nuchal emargination, and it is wider than long as in Chedighaii . The first neural is more elongated than the following neurals and is roughly rectangular in shape. It does not contact the second costal, as in most bothremydids. The second to fifth neurals are all elongated and six-sided with the short anterolateral sides. The sixth neural is five-sided and as long as wide; it does not contact the suprapygal. The first costal is about as long as the second and the third costals together. Its inner side bears the axillary buttress scar, where a rounded swelling links the scar to the first and second thoracic rib head. The fifth costal bears the inguinal buttress scar on its inner side, which extends nearly half the width of the plate along its center, as in most bothremydids. The posterior part of the sixth costal and the seventh and eighth costals meet on the midline, separating the neural series from the suprapygal. The suprapygal is triangular in shape, as in other bothremydids. The pygal is slightly longer than wide.

The carapacial scales of Kurmademys do not differ significantly from those in Chedighaii and other bothremydids. The cervical scale is absent.

PLASTRON

The plastron in ISIR 278 consists of the posterior part of the left hyoplastron, left hypoplastron, left mesoplastron, and partial right hypoplastron. Other elements of this specimen are too badly broken to be recognizable. The plastron reconstruction is based mainly on ISIR 278, with information from ISIR 152S (right hyoplastron) and ISIR 157L (xiphiplastron). The entoplastron and both epiplastra are not known.

The anterior lobe in Kurmademys is probably relatively long, as in Cearachelys , not short, as in the Bothremydini and Taphrosphyini , based on the lateral edges of the hyoplastra. The mesoplastron is equidimensional but larger than that bone in Chedighaii . The posterior lobe is wide, as in Chedighaii , with a semicircular-shaped anal notch.

The plastral scales of Kurmademys are not well preserved. Only a lateral portion of the humeropectoral sulcus is present, and its position is very similar to that in Araiochelys , suggesting that the pectoral scale covers a small part of the epiplastron and a large part of the entoplastron. The pectoroabdominal sulcus is placed far anterior to the mesoplastron, also as in Araiochelys . The pelvic scars are similar to those in Chedighaii ; the ischiac scar is triangular in shape and situated anterior to the posterior margin of the plastron.

TRIBE CEARACHELYINI Cearachelys placidoi (fig. 258, table 22) MATERIAL AVAILABLE: THUg 1798 ( Gaffney, Campos, and Hirayama, 2001: figs. 8, 9), a complete shell; BSP 1976 I 160, a complete shell; MPSC uncatalogued, type specimen ( Gaffney, Campos, and Hirayama, 2001: figs. 6, 7), plastron and partial carapace.

CARAPACE

The carapace of Cearachelys is nearly complete in THUg 1798 and BSP 1976 I 160. Fragments of the left and right bridge peripherals are present in the MPSC specimen.

The carapace of Cearachelys is moderately domed, much as in recent Pelomedusa . The carapace is composed of a nuchal, eight neurals, eight pairs of costals, 11 pairs of peripherals, a single suprapygal, and a single pygal. There are no fontanelles as in Araripemys , and all the bones are tightly sutured as in most pleurodires. The principal distinguishing features of the carapace in Cearachelys lie in the neural bones. The first neural in most Pelomedusoides is four-sided and contacts only the nuchal, first costals, and second neural. In Cearachelys the first neural is six-sided and has short, paired contacts with the second costals. In THUg 1798 the second neural is four-sided, rather than six-sided as in most Pelomedusoides, and as a consequence it does not contact the first costals. In Araripemys the second neural also does not contact the first costals.

TABLE 22

Possible Cearachelyini Shells

However, in BSP 1976 I 160 the second neural also does not contact the first costals. However, in Munich BSP I 160 the second neural is asymmetric with five sides and a third costal contact only on the left side. The third neural in this shell is also asymmetric with five sides and a second costal contact only on the right side. Neurals 4–6 are the usual six-sided, coffin-shaped bones. Neurals 7 and 8 are smaller and more irregular, varying in the two carapaces. Neural 7 is six-sided, but the two lateral sides are nearly parallel rather than converging as in the other neurals. In both shells neurals 7 and 8 occupy the area between costals 7 and 8, but the neurals vary in size. In THUg 1798, neural 7 is much longer than neural 8 while in BSP 1976 I 160 they are nearly the same length. The triangular suprapygal contacts the last neural and is a bit wider in BSP 1976 I 160 than in THUg 1798.

The eight costals of Cearachelys are similar to those in Chedighaii and Taphrosphys as well as Euraxemys . The 11 peripherals are also similar to the other bothremydids, being wider posteriorly. The carapacial scales of Cearachelys are quite similar to the generalized condition for Pelomedusoides seen in Chedighaii , Taphrosphys , and Podocnemis . Because of a complete neural series, the sulcus between vertebrals 4 and 5 falls on the seventh neural in THUg 1798 and nearly on the seventh-eighth neural suture in BSP 1976 I 160. As expected, the scales in Cearachelys differ from the unique condition in Araripemys in which the first vertebral enters the nuchal emargination and the first two marginals are widely separated.

PLASTRON

The plastron in Cearachelys is known in all three specimens. The plastron in BSP 1976 I 160 is complete, but the bone surface is damaged in some areas, precluding sulci preservation. THUg 1798 is nearly complete with all sutures and sulci preserved. The MPSC specimen is missing some of the anterior edges of the plastron, and the posterior margins of the xiphiplastra are broken off.

The plastron of Cearachelys has a broad, semicircular anterior lobe and a tapering posterior lobe with a shallow xiphiplastral notch. The anterior lobe in Cearachelys is much broader than in Euraxemys and Araripemys . It agrees with Euraxemys and most pleurodires in being rounded and differs strongly from that in Araripemys , which is pointed. The epiplastra in Cearachelys meet on the midline for a length that is much more than in Araripemys but less than in Euraxemys . The entoplastron in Cearachelys is trapezoidal, not V-shaped as in Araripemys , and it does not have a curved posterior margin as in Euraxemys . The anterior lobe in Cearachelys is not as short as in the Taphrosphyini and Bothremydini , agreeing with Kurmademys .

Paired, laterally placed mesoplastra are present in Cearachelys , as in Euraxemys and in contrast to Araripemys , which lacks them. The mesoplastra of Cearachelys are similar in size and shape to those in Podocnemis , Taphrosphys , and Chedighaii and are not very large as in Euraxemys . The axillary and inguinal buttress attachments are not visible in any of the specimens at their current stage of preparation. The xiphiplastron has a moderate posterior projection and a shallow xiphiplastral notch, much as in Chedighaii but in contrast to the pointed projections and C-shaped notch in Taphrosphys .

The plastral scales in Cearachelys are much as in other Pelomedusoides. The intergular is roughly V-shaped and extends onto the entoplastron partially separating the humerals, as in Chedighaii and Podocnemis . The intergular extends onto the entoplastron slightly more than in Euraxemys but not as much as in Taphrosphys in which the intergular is large and completely separates the humerals. The humero-pectoral sulcus is behind the entoplastron, as in Euraxemys and Rosasia , but not crossing it, as in Foxemys , Polysternon , and Taphrosphys . The pectoral-abdominal sulcus crosses the anterior part of the mesoplastron, as in Chedighaii and most other Pelomedusoides. The other plastral scales are very similar to those in Chedighaii .

TRIBE BOTHREMYDINI Foxemys mechinorum (fig. 259) MATERIAL AVAILABLE: MDEt 10, shell (figured in Tong et al., 1998: figs. 1, 2); PAM 548, shell (figured Tong et al., 1998: figs. 9, 10); MDEt 09, complete plastron; MDEt 11, right scapuloprecoracoid and several isolated plates ( Tong et al., 1998); MDEt uncatalogued, nearly complete shell; MHNM uncatalogued specimen, a nearly complete shell.

CARAPACE

The shell morphology of Foxemys is very similar to that in Polysternon , although there are some differences (see Tong et al., 1998: table 1). The surface of the carapace and plastron in Foxemys is covered by the so-called ‘‘pelomedusoid’’ ornamentation, state 1 of character 175, that consists of fine forking and irregular vascular grooves, but the parallel striations seen in Polysternon are absent in Foxemys . The carapace outline is an elongated oval rather than more rounded, as in Polysternon and most other bothremydids, with a straight anterior margin and without a nuchal emargination. A wide posterior emargination on the carapace is present on the holotype ( Tong et al., 1998), which is apparently an individual anomaly and is absent in other specimens.

The nuchal is trapezoidal with a straight anterior margin. The anterior portions of the lateral margins of the nuchal are nearly parallel to each other in most specimens. The greatest width of the nuchal is usually about two times its anterior margin. There are seven neurals; their size and shape are similar to Polysternon , with the seventh and eighth pairs of costals meeting on the midline. The first neural is the largest and is roughly rectangular with slightly convex lateral margins. Its posterior end is usually slightly wider than the anterior end. The single, triangular suprapygal contacts the pygal and eleventh peripherals posteriorly. Eleven peripherals are present. The first one is roughly triangular with a short contact to the first costal.

There is no cervical scale, as in all Pelomedusoides. The vertebral scales are all slightly wider than long. The first vertebral is restricted to the first peripheral in most specimens, but some shells, such as PAM 548 and the MHNM specimen, have the first vertebral extending onto the second peripheral plate. The second vertebral is as wide as the first in all Foxemys specimens, while the second vertebral is clearly narrower than the first one in the Polysternon from the Fuveau Basin. All marginal scales are restricted to the peripheral bones.

PLASTRON

The axillary buttress in Foxemys is large and inserts on the lateral part of the first costal plate. On the inner side of the first costal plate, a low, blunt, posteriorly convex ridge links the axillary scar to the first and second dorsal rib heads. The inguinal buttress contacts the eighth peripheral plate and the lateral third of the fifth costal.

The plastron is slightly shorter than the carapace. The anterior lobe is short and wide at the base, as in other Bothremydini , with a nearly straight anterior margin. The bridge is longer than the posterior lobe, and the posterior lobe is longer than the anterior one. The lateral margins of the posterior lobe of Foxemys are less rounded than in Polysternon . The anal notch is large and wide, wider than in Polysternon .

The length of the epiplastral symphysis is about one-third to one-half that of the entoplastron. The large and diamond-shaped entoplastron is wider than long in most specimens, but some of them, such as the MHNM specimen and MDEt 09, have the entoplastron as long as wide. The posterior end of the entoplastron reaches the level of the bridge, as in Polysternon , but differs from Rosasia , in which the entoplastron is more anteriorly placed. The hyoplastra and hypoplastra are about the same length. The mesoplastron is roughly semicircular in shape and longer than wide. The size of the mesoplastron varies compared to the entoplastron, from smaller (MDEt 10) to equal (MDEt 09, MNHM specimen, and PAM 548). On the inner side of the xiphiplastron, the pubic scar is an anteromedially orientat- ed, elongate oval, with an acute anterior tip. The ischiac scar is elongated and triangular in shape. Its anterior margin is transverse and its posterior tip is anterior to the anterior margin of the anal notch.

The plastron has the usual 13 scales. The gular scales are large and triangular in shape, reaching but not crossing the anterior margin of the entoplastron. In PAM 548, MDEt 09, and most other specimens, the intergular is large, widely separating the gulars and extending for half the length of the entoplastron. In contrast, the shells in MDEt 10 and the MHNM specimen have a smaller intergular scale that reaches only the anterior third of the entoplastron. The humeral scale has a variable but short midline length, from 40 % to 8 % of the entoplastral length. The humeropectoral sulcus crosses the entoplastron posteriorly, about one-third to one-fourth of the length of the entoplastron, and then crosses the epihyoplastral suture about halfway along the length of the suture. The humeropectoral sulcus is entirely anteri- or to the epihyoplastral suture in MDEt 10. The pectoral scales cover the anterior two-thirds of the hyoplastra. The pectoroabdominal sulcus is straight and does not reach or barely reaches the mesoplastra. The abdominals are the largest plastral scales. They cover the posterior third of the hyoplastra and the anterior half of the hypoplastra and mesoplastra. The abdominofemoral sulcus is straight or slightly convex anteriorly, with the lateral ends curved posteriorly. The femoral scales cover the posterior half of the hypoplastra and the anterior part of the xiphiplastra. The femoroanal sulcus is convex anteriorly. The anal scales cover the posterior half of the xiphiplastra.

Polysternon provinciale (fig. 260) MATERIAL AVAILABLE: Two shell fragments including an anterior portion of carapace (MHNM 1982-853), from ‘‘lignite de la grande Mène’’, Matheron collection ( Matheron, 1869; Broin, 1977: fig. 1, pl. 1, fig. 3a,b; the other specimen was already lost when Broin [1977] reviewed the material); uncatalogued partial shell in the collection of the Musée Cantonal de Géologie de Lausanne ( Portis, 1882); MHNM 1982-857, holotype of ‘‘ Elochelys major ’’ ( Nopcsa, 1931: pl. XIII; Broin, 1977: fig. 2, pl. 1, fig. 1); MHNM 1982-855, internal cast showing the plastron with pubic and ischiac scars, collection Comte de Gérin-Ricard ( Nopcsa, 1931: fig. 1; Broin, 1977: pl. 1, fig. 2, from ‘‘Valdonne’’); Costa collection, a complete shell ( Buffetaut et al., 1996: fig. 4) and numerous plates.

CARAPACE

The shell surface ornamentation is figured by Portis (1882: pl. 28) and Broin (1977:

pl. 1, fig. 3b). It consists of anteroposteriorly directed, thin parallel striations, along with the ‘‘pelomedusoid’’ texture pattern, consisting of fine forking and irregular vascular grooves (state 1, character 175). The parallel striations, state 3, are unique among bothremydids; they are well marked and cover the entire surface of the carapace on the Portis specimen. On the holotype and Villeveyrac specimens, they are well marked on the neurals and the medial part of the costals. They are weaker on the plastron than on the carapace and absent on the plastron. The ‘‘pelomedusoid’’ ornamentation is also visible on the peripherals and the plastron.

The shell is low and a shortened oval in outline, different from that in Foxemys , which is an elongated oval. The nuchal emargination is present and shallow. The nuchal is trapezoidal and roughly as long as wide. There are seven neurals, with the first being the longest and having the laterally convex lateral margins as in Foxemys . The lengths of the neurals decrease from the front to the back. The second to sixth neurals are six-sided with short anterolateral sides. The seventh neural is five-sided and smaller. Eight costals are present; the first is the longest, but not as long as in Taphrosphys . The seventh and eighth pairs of costals meet on the midline. A single, triangular suprapygal contacts the pygal and the eleventh peripheral posteriorly, except in the Villeveyrac specimen, in which the suprapygal is narrower and does not contact the eleventh peripheral. There are 11 peripherals, all longer mediolaterally than anteroposteriorly in the Villeveyrac specimen, and the preserved posterior peripherals in Portis’ specimens ( Portis, 1882: pl. 28). The first peripheral is roughly triangular with a short contact with the first costal, as in Foxemys but in contrast to Cearachelys , Chedighaii , and Rosasia in which the first peripheral is more rectangular with a longer contact with the first costal.

The cervical scale is absent, as in all Pelomedusoides. The first vertebral is wider than the second one, in contrast to Foxemys , in which it is as wide as the second vertebral, and Cearachelys , Taphrosphys , Elochelys , and Rosasia , in which the first vertebral is narrower than the second one. The vertebrals 2–4 are longer than wide in the type shell, but wider than long in the Villeveyrac specimen.

PLASTRON

The plastron is shorter than the carapace. The anterior lobe is relatively short and wide at the base, as in other Bothremydini and Taphrosphyini . The bridge is longer than the posterior lobe; the latter is longer than the anterior lobe. The posterior lobe is wide with laterally convex lateral margins, differing from the nearly straight lateral margins of Araiochelys . The anal notch is deep and narrow. The pubic and ischiac scars are similar to those in Foxemys . The epiplastron is large with a rather long midline suture. The large entoplastron is diamond-shaped and slightly wider than long. The mesoplastron is laterally placed and smaller than the entoplastron.

Thirteen scales cover the plastron. The intergular scale is roughly as wide as the gular scale and covers little of the entoplastron. The gular scale reaches the entoplastron but does not overlap it in the type. In the Villeveyrac locality, among seven epiplastra preserved, only one has the large gular scale reaching the entoplastron; on the others the gular scale is smaller and does not reach the entoplastron. The humeropectoral sulcus is located half to a third of the length along the entoplastron and close to the epihyoplastral suture, in contrast to Cearachelys , Galianemys , and Rosasia in which the humeropectoral sulcus is located far behind the epihyoplastral suture and posterior to the entoplastron. The pectoroabdominal sulcus barely reaches the mesoplastron or is located slightly anterior to it, in contrast to Araiochelys and Kurmademys in which this sulcus is located well anterior to the mesoplastron.

Rosasia soutoi (fig. 261) MATERIAL AVAILABLE: Museu Minerologico e Geologico de la Faculdade de Ciencias, Universidade do Porto, uncatalogued type shell (figured in Carrington da Costa, 1940: pls. I, III); Universidade Nova de Lisboa, Portugal, MTA 1, shell ( Antunes and Broin, 1988: pl. 3, figs. 1, 2), MTA 2, shell ( Antunes and Broin, 1988: pl. 4, figs. 1, 2); two shells (only one figured) in the Serviços Geológicos de Portugal Lisbonne ( Carrington da Costa, 1958: pls. 3, 4); one shell in the Faculdade de Ciencias e Tecnologia, Universidade de Coimbra ( Carrington da Costa, 1958: pls. 5, 6).

CARAPACE

The present description is based on the descriptions and illustrations of Carrington da Costa (1940, 1958) and Antunes and Broin (1988). The shell surface of Rosasia is figured in detail by Carrington da Costa (1940: pl. II; 1958: pl. II). The texture is the ‘‘pelomedusoid’’ pattern, fine forking and irregular vascular grooves, as in many Pleurodira , especially bothremydids, but not as dense as in Taphrosphys ( Antunes and Broin, 1988) .

The shell is low, with a rounded outline. The nuchal emargination is narrow and deep, mostly involving the nuchal. Antunes and Broin (1988) mentioned the absence of the nuchal emargination in the specimen of Coimbra; this may be due to damage of the anterior margin of the shell. The nuchal is wider than long, but the anterior margin is roughly equal to half of the maximum width. There are seven neurals: the first is four-sided, the following ones are six-sided with short anterolateral sides, and the last neural is five-sided. The seventh and eighth pairs of costals meet on the midline. The suprapygal is small and triangular in shape, with its length slightly larger than the width. The pygal is slightly longer than wide. The first costal is longer than the second and third costals together. The first peripheral is roughly rectangular in shape, with the anterior margin being only slightly longer than the posterior one, in contrast to Polysternon and Foxemys in which the first peripheral is more triangular.

The cervical scale is absent, as in all Pelomedusoides. The second and third vertebral scales are roughly as long as wide. All marginal scales are restricted to the peripheral plates. The carapace has strong buttresses, as in podocnemidids and other bothremydids. The axillary buttress extends to the lateral half of the first costal plate and the inguinal buttress attaches to the fifth costal plate ( Antunes and Broin, 1988). The pelvic girdle is sutured to the shell, as in all Pleurodira . Its contact with the carapace cannot be seen. The pubic scar is oval in shape, oblique, and relatively short and wide, not elongated, as in Taphrosphys . The ischiac scar is roughly triangular in shape and placed anterior to the anal notch.

PLASTRON

The anterior lobe of the plastron is short and wide at its base, as in most bothremydids, with a semicircular outline. The bridge is long, longer than the posterior lobe and longer than in other bothremydids. The posterior lobe is longer than the anterior one and has straight lateral margins convergent posteriorly. The anal notch is very shallow and wide. The entoplastron is small, diamond-shaped, and as long as wide. Its posterior end does not reach the level of the axillary notch. The mesoplastron is larger than the entoplastron and is slightly wider than it is long.

The plastron is covered by 13 scales. The intergular scale overlaps about a third of the entoplastron. The gular scale is small and does not reach the entoplastron. The humeropectoral sulcus crosses the hyoplastron just posterior to the end of the entoplastron and is placed far behind the epihyoplastral suture, as in Cearachelys and Galianemys . The pectoroabdominal sulcus crosses the anterior part of the mesoplastron. The abdominofemoral sulcus is about midway along the hypoplastron, as in Cearachelys , Polysternon , and Foxemys , while in Araiochelys and Taphrosphys it is more anterior. The femoroanal sulcus is as in other bothremydids, crossing the xiphiplastron about midway.

Araiochelys hirayamai (figs. 262, 263)

MATERIAL AVAILABLE: THUg 3338, partially disarticulated shell (fig. 262), including left first and fifth costals, left third, seventh, and eighth peripherals and other fragments, nearly complete plastron and limb bones; MDEt 25, an incomplete and disarticulated shell, neurals 1–2, left costals 1–4 (the fourth is incomplete), right costals 2–5 (costals 3 and 5 are incomplete), left tenth and eleventh peripherals, one peripheral of bridge region, right hypoplastron and xiphiplastron, and other fragments. MDEt 25 is tentatively identified as Araiochelys hirayamai because of the similar size, the shape and the structure of the first and fifth costal plates, similar development of the axillary and inguinal buttress scars, the shape of the posterior lobe of the plastron, and the shape and position of the pubic and ischiac scars. Both MDEt 25 and THUg 3338 are from the Danian beds of the Ouled Abdoun Basin, Morocco. None of the plastron and only part of the carapace restoration are dependent on the identification of MDEt 25 as Araiochelys . CARAPACE

The surface texture of the carapace and plastron is the usual network of fine furrows, the ‘‘pelomedusoid’’ texture (state 1, character 175). This pattern is less pronounced than that seen in Taphrosphys ; the polygonal areas delimited by the furrows are flat rather than raised, as in Taphrosphys . The carapace is very thin in MDEt 25, measuring from 6 mm for neurals to less than 2 mm for the lateral end of the costal plates. In MDEt 25, only the first and the second neurals are preserved. The first neural is elongated and roughly rectangular in shape; it is slightly expanded at the midpoint and does not contact the second costal. The second neural is much shorter than the first and is pentagonal in shape with short anterolateral sides. It is in contact with the first neural anteriorly and the first costal anterolaterally. The following neurals are not preserved, but the complete medial margin of the second to fifth costal plates allows the reconstruction of the third to fifth neurals. The third to fifth neurals are all diamond-shaped and longer than wide, with their size decreasing from front to back. The posterior part of the second to fifth costals meets on the midline between the neurals. The sixth neural was present but not preserved, judging by the posteromedial margin of the fifth costal. It is probably smaller than the fifth neural.

The complete left first costal is preserved in THUg 3338. In MDEt 25 the left first costal is nearly complete, but only a small fragment of the right one is preserved. The first costal is relatively long, with its length being greater than that of the second and third costals together, as in Taphrosphys sulcatus ( Gaffney, 1975a) . The inner surface of this plate bears a large scar for axillary buttress insertion that reaches halfway along the costal. A posteriorly convex, low ridge with an acute anterior edge links the axillary buttress scar to the first and second dorsal rib heads. This structure is present on both THUg 3338 and MDEt 25. The second to fourth costals are short, as in Taphrosphys and Chedighaii . The fifth costal is preserved in both THUg 3338 and MDEt 25, in which it widens distally; the inguinal buttress reaches its midpoint on the ventral side. The third peripheral is rectangular and longer than wide. The seventh, eighth, and tenth peripherals are roughly square in shape. The eleventh peripheral is wider than long. The posterior margin of the carapace is undulant.

The scute sulci of the carapace are shallow but easily distinguishable. The first vertebral scale is not complete; only the posterior part of the left side is preserved. The first vertebral is clearly narrower than the second one. The second vertebral scale is very large and longer than wide. It is much larger than the third vertebral scale, with its posterolateral corner reaching the posterior margin of the third costal, which is a unique feature among bothremydids. The third vertebral scale is roughly square in shape. The intervertebral sulci cross the first neural medially, just anterior to the fourth neural and just posterior to the fifth neural.

PLASTRON

The plastron of THUg 3338 lacks the epiplastra and entoplastron; the anterior edge of both hyoplastra is damaged. In MDEt 25, the right hypoplastron and xiphiplastron are preserved. The hyoplastron is slightly widened at its anterior part. The mesoplastron is not completely known, but it seems to be slightly wider than long. The posterior lobe is narrow with nearly straight lateral margins. There is a shallow notch on the lateral border of the xiphiplastron at the lateral limit of the femoroanal scute sulcus in both THUg 3338 and MDEt 25. The anal notch is V-shaped, deep, and broad in THUg 3338. It has more curved borders in MDEt 25, which may be sexual dimorphism or may indicate a separate taxon. Both hypoplastron and xiphiplastron are much longer than wide. On the dorsal side of the xiphiplastron, the pubic scar is an elongated oval; it is less oblique and is placed closer to the lateral margin of the plastron than in Taphrosphys sulcatus ( Gaffney, 1975a) . The ischiac scar is roughly triangular in shape. It is posterior to the anterior margin of the anal notch, which differs from the condition in Foxemys , but it is not as close to the posterior margin of the plastron, as in Taphrosphys .

The humeropectoral sulcus is preserved on the right hyoplastron of THUg 3338; it is posterior to the epihyoplastral suture, as in Taphrosphys and Chedighaii , but it probably crosses the entoplastron. The pectoroabdominal sulcus is preserved on the left hyoplastron; it is nearly straight and placed well anterior to the mesoplastron. The abdominofemoral sulcus is slightly convex anteriorly, crossing the midline at about one-third of the length of the hypoplastron. The femoroanal sulcus crosses the midline near the middle of the xiphiplastron.

Chedighaii barberi (fig. 264, table 23)

MATERIAL AVAILABLE: FMNH P 26055, shell ( Schmidt, 1940: figs. 1–5); FMNH P2769, shell ( Zangerl, 1948: fig. 3); FMNH P27372, shell ( Zangerl, 1948: figs. 4, 7, 9,10); FMNH 27370, shell ( Zangerl, 1948: figs. 5, 6, 11, 12, pl. 4, fig. 1); FMNH 27331, shell; FMNH P27405, shell; FMNH PR 247, shell ( Gaffney and Zangerl, 1968: figs. 2, 3); ANSP 15902, shell ( Gaffney and Zangerl, 1968: figs. 408); YPM 3608, shell ( Gaffney and Zangerl, 1968: figs. 9–12); ALAB PV 2001.2, partial shell.

CARAPACE

Shell surface texture is the ‘‘pelomedusoid’’ pattern ( Broin, 1977). The carapace is broadly rounded and low, as in Taphrosphys . The lateral margins converge posteriorly, with the widest point at the level of the seventh to eighth peripherals, then becoming narrower posteriorly.

The posterior end of the carapace tapers to a blunt curve at the rear. It is wider than long ( FMNH P27369, FMNH P27370, FMNH P27372, FMNH PR 247) or as long as wide ( FMNH P26055). This is more like the shell of Taphrosphys ( Gaffney, 1975a) but different from that of Galianemys and Cearachelys , in which the shell is an anteroposteriorly elongated oval. The nuchal emargination is present in all specimens with this part available, although its size and shape vary. In FMNH P26055, FMNH PR 247, and FMNH P27370, the nuchal emargination is narrow and deep, while FMNH P27369 and YPM 3608 have a shallower emargination.

The carapace of Chedighaii barberi is composed of one nuchal, six or seven neurals, eight pairs of costals, 11 pairs of peripherals, one suprapygal, and one pygal, as in other Pelomedusoides. The nuchal is six-sided and short; it is clearly wider than long. Six neurals are present in FMNH P26055, FMNH P27370, FMNH P27372, FMNH P27331, and ANSP 15902, while FMNH PR 247, FMNH P27369, and YPM 3608 have seven neurals. The first neural is rectangular and only slightly longer than the second neural, except YPM 3608, which has a longer first neural. In most other bothremydids, the first neural is much longer than the second one ( Cearachelys , Galianemys , Kurmademys , Foxemys , Polysternon ). The second to fifth neurals are six-sided with short anterolateral sides. The sixth neural is as wide as long; it is five-sided when it is the last neural or six-sided when seven neurals are present. In both cases the sixth neural is as wide as the fifth one. The seventh neural, when present, is clearly smaller than the sixth. The first costal is longer than the second and third costals together, as in Taphrosphys . The sixth to eighth pairs of costals meet on the midline, separating the last neural from the suprapygal, when six neurals are present. In specimens with seven neurals, the seventh and eighth costals meet on the midline. The suprapygal is triangular, as in other Pelomedusoides. The first peripheral is four-sided, with the posterior margin being much shorter than the anterior one. The size of the peripherals increases from front to back, reaching the maximum size at the eighth and ninth peripherals. The tenth and eleventh peripherals are much smaller than the ninth one. The tenth and eleventh peripherals are also small in Taphrosphys , but the peripherals are generally narrower in it ( Gaffney 1975a). The pygal is four-sided and nearly square, as in many Pelome-

TABLE 23

Comparison of Chedighaii barberi Shells dusoides, with the posterior margin being longer than the anterior one.

The scale sulci on the carapace are preserved and figured in Schmidt (1940: fig. 2, FMNH P26055) and Zangerl (1948: pl. 4, fig. 1, AMNH P27370). They are partly visible in ANSP 15902 ( Gaffney and Zangerl, 1968: fig. 4) and YPM 3608 ( Gaffney and Zangerl, 1968: fig. 9). The general pattern of the scales of Chedighaii barberi agrees with that in other Pelomedusoides. The cervical scale is absent, as in all Pelomedusoides. The vertebral scales are all wider than long in FMNH P27370, FMNH P27372, and FMNH P27331, with strongly angled lateral margins. In FMNH P26055, ANSP 15902, and YPM 3608, the first to fourth vertebrals are narrower, with only slightly angled lateral margins. The first marginal is much wider than long,as in Galianemys but in contrast to Taphrosphys and Cearachelys in which it is square. All marginals are restricted to the peripherals, as in other bothremydids.

On the inner side of the carapace, the axillary buttress scar lies on the lateral half of the first costal, near its posterior margin ( Zangerl, 1948: fig. 7; Gaffney and Zangerl, 1968: figs. 5, 10). In Taphrosphys , the axillary scar is even more posteriorly placed ( Gaffney, 1975a: fig. 4). The inguinal scar is less than half the width of the fifth costal, as in Taphrosphys . The iliac scar is placed on the posterior part of the seventh costal and eighth costals, overlapping slightly onto the suprapygal ( Zangerl, 1948: fig. 9), as in Taphrosphys .

PLASTRON

The plastron in Chedighaii is shorter than the carapace, with the anterior margin posterior to that of the carapace, unlike Cearachelys and the possible Galianemys shells. The anterior lobe is short and wide, with a semicircular outline in FMNH P26055, ANSP 15902, FMNH PR 247, and FMNH P27370, in contrast to YPM 3608 and FMNH P27369, which have a more trapezoidal-shaped anterior lobe. The bridge is longer than the anterior lobe and shorter than the posterior lobe. The bridge is wide, as in Taphrosphys but in contrast to Galianemys , Foxemys , Polysternon , and Rosasia in which it is narrower. The posterior lobe is narrower than the anterior lobe, with the lateral margins tapering posteriorly, unlike Taphrosphys , which has the posterior lobe nearly as wide as the anterior one with nearly parallel lateral margins. The anal notch is deep and wide. It has an inverse V-shape in FMNH PR 247, FMNH P26055, and FMNH P27369 and a more U-shape in ANSP 15902.

The plastron is composed of 11 plates, as in other podocnemidoids. The epiplastron is similar to Taphrosphys , with a comparatively short midline suture, except in YPM 3608 ( Gaffney and Zangerl, 1968) and FMNH P26055. In Cearachelys the midline suture of the epiplastron is longer. The entoplastron is diamond-shaped and wider than long in most specimens (except ANSP 15902), as in Taphrosphys but differing from Galianemys in which it is as long as wide. The hyoplastron and hypoplastron are similar in length. The xiphiplastron is much longer than wide, as in most other bothremydids but in contrast to Taphrosphys in which this plate is equidimensional.

On the inner side of the plastron, the pubic scar ( Zangerl, 1948: fig. 11, pl. 4, fig. 2; Gaffney and Zangerl, 1968: figs. 8, 12) is oval and elongated, similar to that of Araiochelys . In Taphrosphys , the pubic scar is longer and narrower ( Gaffney, 1975a: fig. 11). The ischiac scar of Chedighaii barberi is similar to that of Podocnemis . It is triangular, wider than long, and placed far anterior to the posterior margin of the plastron. In Taphrosphys , the ischiac scar is nearly rounded and is situated near the posterior margin of the plastron.

The scale sulci on the plastron are only partly visible in FMNH P27369 ( Zangerl, 1948: fig. 3), FMNH PR 247 ( Gaffney and Zangerl, 1968: fig. 3), and YPM 3608 ( Gaffney and Zangerl, 1968: fig. 11). There are some visible on ANSP 15902 ( Gaffney and Zangerl, 1968: fig. 7) but only a few on the type specimen (although none figured, Schmidt, 1940: fig. 5). When all specimens from the Selma Formation at the Field Museum are taken together, all the scales are known for a composite reconstruction (fig. 264). The intergular scale is similar to that in Foxemys , extending about half the length of the entoplastron. The gular scale is triangular and wider than long, reaching the entoplastron but not crossing it. The humeropectoral sulcus is entirely posterior to the epihyoplastral suture, extending through the posterior third of the entoplastron. The pectoroabdominal sulcus is similar to that of Galianemys , crossing the mesoplastron. The abdominofemoral sulcus lies along the midpoint of the hypoplastron. The femoroanal sulcus is similar to that in Taphrosphys .

TRIBE TAPHROSPHYINI Taphrosphys sulcatus (fig. 265, table 24) MATERIAL AVAILABLE: YPM PU 18706, partial shell collected by Regensburg and Dilkes, August, 1960, MFL Hornerstown Formation, Sewell, New Jersey ( Gaffney, 1975a); YPM PU 18707, partial shell with associated limb and skull fragments, same data as in YPM PU 18706; YPM PU 18708, left xiphiplastron, same data as in YPM PU 18706; ANSP 15544, carapace with partial skull and limb elements, collected by Keith Madden and James Maddox from the MFL Hornerstown Formation, Inversand pits, Sewell, New Jersey ( Richards and Gallagher, 1974; White, 1972); NJSM 10410, a plastron, no data; AMNH 1347, nuchal and first peripherals, ‘‘Cretaceous, New Jersey’’ (label), Cope Collection; AMNH 1470 ( Hay, 1908: 119, fig. 119), neural bones, nuchal, and other shell fragments, ‘‘Barnsboro, 9/8 1870 ’’ (label), Cope Collection; AMNH 1472 (type of Taphrosphys molops Cope , figured in Cope, 1870: pl. 7, fig. 16; text-figs. 43, 44; Hay, 1908: figs. 112–116), nearly complete anterior lobe, incomplete xiphiplastron and other shell fragments, West Jersey Marl Company pits, Barnsboro, New Jersey, August 25, 1868, Cope Collection; AMNH 1474 ( Hay, 1908: figs. 117, 118), partial plastron with incomplete right and left hypoplastra, complete left xiphiplastron and other fragments, Birmingham, New Jersey, Cope Collection; AMNH 1477 ( Hay, 1908: fig. 120), shell fragments, West Jersey Marl Company pits, Barnsboro, New Jersey, 1855, Cope Collection; AMNH 1471 ( Hay, 1908: fig. 106), anterior plastral bones, mixed with AMNH 1470 (label), Cope Collection; AMNH 1467 (type of Taphrosphys leslianus Cope , figured in Hay, 1908: figs. 103–105), partial shell, Hornerstown, New Jersey, collected by Dr. S. Lockwood, Cope Collection; AMNH 1125 (only the right first costal is labeled as 1124) (type of Taphrosphys longinuchus Cope , figured in Hay, 1908: figs. 101, 102), partial shell with limb fragments (this specimen was apparently mixed with the type of Adocus agilis Cope, AMNH 1135 ), David Haine’s marl pit, Medford, New (from Gaffney, 1975a). Dorsal view of carapace (left) and ventral view of plastron (right). [J. Kane, del.]

Jersey, March 15, 1870, Cope Collection; AMNH 1468 ( Cope, 1870: 166, fig. 45; not that two consecutive figures in Cope [1870] are both labeled fig. 45 but illustrate different specimens; Hay, 1908: figs. 99, 100), shell fragments, Barnsboro, New Jersey, Cope Collection; AMNH 1469, shell fragments, West Jersey Marl Company pits, Barnsboro, New Jersey, 1869, Cope Collection; AMNH 2524 ( Cope, 1870: 165, fig. 45), nuchal bone, Barnsboro, New Jersey, in the Rutgers University Cook Collection until 1970; AMNH 1126, plastron fragments, incomplete ilium and humerus, and other shell fragments, type of Taphrosphys strenuous , ‘‘?Hornerstown Formation (Upper Greensand), Barnesboro, Gloucester Col, New Jersey, Cope Collection’’ (label); AMNH 1128, shell fragments, including one neural and several fragments of costals, ‘‘ Taphrosphys molops, Cope Collection’’ (label); AMNH 14754, incomplete nuchal and first peripherals, one neural, incomplete xiphiplastron, and other shell fragments; Taphrosphys sulcatus , shell fragments, MFL Hornerstown Formation, HT-1, Hornerstown type locality, Hornerstown, Monmouth Co., New Jersey, coll. E. S. Gaffney.

CARAPACE

The best previously published reconstructions of Taphrosphys sulcatus are those based on AMNH 1125 by Hay (1908: figs. 101, 102) and YPM PU 18706 by Gaffney (1975a). The two best preserved shells, YPM PU 18706 and YPM PU 18707, were described in Gaffney (1975a). Most of the material of Taphrosphys , particularly in the AMNH, YPM, NJSM, and ANSP collections, is fragmentary. Taphrosphys sulcatus has prominent surface texture on both carapace and plastron, consisting of irregular raised polygons separated by a network of deep furrows. This kind of sculpture (state 2, character 175) is different from the more common ‘‘pelomedusoid’’ texture (state 1) defined by Broin (1977) and found in other bothremydids. The carapace of Taphrosphys is a short oval, with a narrow anterior end and the lateral margins diverging posteriorly, reaching their widest point at the level of the seventh peripherals and then narrowing to a blunt posterior end.

TABLE 24

Comparison of Taphrosphyini Shells

This shape is very similar to Chedighaii , unlike the oval carapace in Galianemys , Cearachelys , Kurmademys , and Foxemys .

The nuchal is trapezoidal and as long as wide in AMNH 1125 and YPM PU 18706, but wider than long in ANSP 15544 ( Gaffney, 1975a). Another complete nuchal is preserved in AMNH 1347 and an incomplete nuchal is preserved in AMNH 1467 and AMNH 14754. The nuchal emargination is absent in all these specimens with the nuchal preserved; the front margin of the nuchal is straight or slightly convex anteriorly. The neural series consists of seven continuous neurals, based on ANSP 15544, YPM PU 18706, and YPM PU 18707 ( Gaffney, 1975a), as in Foxemys , Polysternon , and Rosasia . The first neural is clearly longer than the second one. It is four-sided with the convex lateral margins, as in Foxemys and Polysternon . The second to fifth neurals are six-sided with short anterolateral sides and are longer than wide. The sixth neural is nearly equidimensional. The seventh neural is five-sided and much smaller than the sixth neural. AMNH 1463 has the last neural preserved and the seventh and eighth costals meeting on the midline. The first costal is longer than twice the length of the second costal, as in Chedighaii and Araiochelys . The seventh and eighth costals meet on the midline, separating the neurals from the suprapygal. The suprapygal is triangular and nearly as long as wide. The pygal, preserved in YPM PU 18707, is wider than long with a convex posterior margin.

On the visceral surface of the shell, the axillary buttress scar is on the lateral half of the first costal, near the posterior margin of the plate. The first rib head is smaller than the second one. The inguinal buttress reaches about half the width of the fifth costal. The iliac scar covers the posterior part of the seventh and eighth costals and slightly overlaps onto the suprapygal in AMNH 1468, YPM PU 18706, and YPM PU 18707, but in AMNH 1125, the iliac scar reaches the anterolateral margin of the suprapygal but does not overlap onto it.

The cervical scale is absent, as in all Pelomedusoides. As in the reconstruction in Hay (1908: fig. 101) and Gaffney (1975a: fig. 1), the first vertebral is narrow; it is narrower than the second and third vertebrals, especially in YPM PU 18706. The second and third vertebrals are wider than long and the fourth vertebral is narrower. The marginal scales are restricted to the peripheral bones, as in all Pelomedusoides.

PLASTRON

The plastron of Taphrosphys sulcatus is composed of 11 plates, as in other bothremydids. The plastron is shorter than the carapace; its front margin does not reach the anterior margin of the carapace, as in Chedighaii , Polysternon , and Foxemys but in contrast to Galianemys and Cearachelys .

The plastron of Taphrosphys sulcatus has a very short and semicircular anterior lobe, as in Chedighaii . The bridge is longer than the anterior lobe and shorter than the posterior one, also as in Chedighaii . The bridge is relatively wide, as in Chedighaii and Araiochelys , not as narrow as in most other bothremydids. The posterior lobe is relatively wide, nearly as wide as the anterior lobe, with nearly parallel lateral margins, in contrast to Chedighaii in which the posterior lobe tapers posteriorly. The anal notch is rounded and very broad.

The epiplastron has a short midline suture, much shorter than the entoplastral length. The entoplastron is complete in AMNH 1472 and YPM PU 18706. It is diamond-shaped and clearly wider than long. The mesoplastron is not preserved in any of the specimens. The xiphiplastron is broad, being nearly as wide as long ( YPM PU 18708, AMNH 1125), or slightly longer than wide ( AMNH 1474 and YPM PU 18706), in contrast to Bothremydini in which it is much longer than wide.

The plastron in Taphrosphys sulcatus is covered by 13 scales, as in all Pelomedusoides. The intergular, preserved in YPM PU 18706 ( Gaffney, 1975a: fig. 2) and AMNH 1472 ( Hay, 1908: fig. 116), is very large, completely separating the gulars and humerals, as in Taphrosphys congolensis and Ummulisani but in contrast to all other bothremydids (except Elochelys ). The gular scale, preserved in YPM PU 18706 ( Gaffney, 1975a: fig. 2) and AMNH 1471 ( Hay, 1908: fig. 106), is triangular; it reaches the entoplastron but does not cross it. The humeropectoral sulcus is behind the epihyoplastral suture, crossing the entoplastron, as in Chedighaii . As in Taphrosphys congolensis and Ummulisani , T. sulcatus has a very short abdominal scale, with the pectoroabdominal sulcus lying near the hyohypoplastral suture, in contrast to all other bothremydids (except ‘‘ T.’’ olssoni and ‘‘ T.’’ ambiguous, fig. 267).

On the inner side of the plastron, the pubic scar is long and narrow. The ischiac scar is very small and rounded, placed very near the margin of the anal notch ( Gaffney, 1975a).

Taphrosphys congolensis (fig. 266, table 24)

MATERIAL AVAILABLE: This description is based primarily on the description and illustrations of Dollo (1913) and Wood (1975), although one of us (E.S.G.) has seen the material.

CARAPACE

The shell surface of T. congolensis is covered by ‘‘a network of deeply inscribed anastomosing vermiculations’’, as described by Wood (1975: 139), as in Taphrosphys sulcatus . The carapace is known only from the anterior portion and the posterior third. The preserved part is very similar to T. sulcatus . On the lectotype of T. congolensis ( Dollo, 1913: pl. VII, figs. 1, 2; Wood, 1975: fig. 1), the nuchal and the first peripheral are represented only by fragments and their shapes are therefore undeterminable. The second peripheral is roughly square and the third one is longer than wide, as in T. sulcatus ( YPM PU 18706 and YPM PU 18707, Gaffney, 1975a: figs. 1, 6, 8). The first costal is as long as in T. sulcatus . On the inner side, the axillary buttress attaches to the third peripheral and the lateral half of the first costal, although its position appears to be more anterior than in T. sulcatus ( Gaffney, 1975a: fig. 4). A posterior portion of carapace ( MRAC 4795) consists of the last neural, right fifth to eighth costals, left seventh and eighth costals, suprapygal, pygal, and right eighth to eleventh peripherals ( Wood, 1975: pls. 1, 2). The preserved neural, between the sixth and seventh costals, is the seventh neural. It is therefore likely that seven neurals are present in T. congolensis , as in T. sulcatus . The last neural is small and short, as in T. sulcatus .

On the inner surface of the carapace, the inguinal buttress attaches to the eighth peripheral and the fifth costal. The iliac scar is on the seventh and eighth costals and covers a small part of the suprapygal, all similar to T. sulcatus .

PLASTRON

A nearly complete plastron ( MRAC 4794) and two more-or-less complete xiphiplastra are figured in Wood (1975: pls. 3, 4, 6). The general morphology of the plastron of T. congolensis is similar to that of T. sulcatus . However, the plastron of T. congolensis seems wider than that of T. sulcatus . The reconstruction by Wood (1975: fig. 2) shows a narrow bridge relative to that of T. sulcatus . The posterior lobe has the lateral margins slightly tapering posteriorly, instead of nearly parallel as in T. sulcatus . The xiphiplastron is slightly longer than wide in MRAC 4794, while the complete xiphiplastron in plate 6 ( Wood, 1975) is broader. Another diagnostic feature of T. congolensis pointed out by Wood (1975) is that its intergular is hexagonal and wider than in T. sulcatus . It is nearly as broad as long; while in other Taphrosphys species, the intergular is much longer than wide. The comparison of species referred by various authors to Taphrosphys is shown in table 24.

Ummulisani rutgersensis (figs. 268, 269)

MATERIAL AVAILABLE: AMNH 30562, a nearly complete plastron associated with skull and jaws.

PLASTRON

The plastron of Ummulisani has the usual 11 bones and 13 scales and is very similar to the plastron of Taphrosphys . The epiplastra are slightly larger relatively in Ummulisani than in Taphrosphys , and the anterior plastral lobe is relatively wider. The entoplastron in Ummulisani is wider than long, as in Taphrosphys . The hyo- and hypoplastra are very similar in both genera. The mesoplastron in Ummulisani lacks the lateral edges, but it seems to be larger than the mesoplastron in Chedighaii . The xiphiplastron in Ummulisani has the narrow posterior projection and broad, semicircular anal notch of Taphrosphys , but it is slightly narrower posteriorly.

On the visceral surface, the buttresses are mostly broken away. The pubic scar is long and narrow in Ummulisani , as in Taphrosphys . The ischiac scar is round as in Taphrosphys , but it is not right on the edge of the xiphiplastron as in Taphrosphys .

The plastral scales of Ummulisani are similar to those in Taphrosphys . There is a large intergular that occupies most of the entoplastron, widely separating the gular and humeral scales. In contrast to Taphrosphys , the gular scales extend onto the entoplastron a short distance in Ummulisani . The abdominal scales are narrow and the femorals are large, as in Taphrosphys .

PELOMEDUSOIDES INDETERMINATE

? Galianemys sp. (table 22, figs. 271–274)

MATERIAL AVAILABLE: AMNH 30550, nearly complete shell (figs. 273, 274), Cenomanian, Kem Kem, Morocco; AMNH 30551, nearly complete shell (figs. 271, 272), Cenomanian, Kem Kem, Morocco. There are differences between these two shells (table 22), and it is possible they represent two taxa (see below). The named turtles known from skulls occurring in the Kem Kem Campanian of Morocco are Galianemys emringeri , Galianemys whitei , Hamadachelys , and Dirqadim , but a phylogenetic analysis including these shells (fig. 294) places them in the tribe Cearachelyini , suggesting that they may be the shells for one or both Galianemys species.

CARAPACE

The two shells have an oval outline, slightly expanded posteriorly. There is no nuchal emargination. AMNH 30551 has eight neurals with costal 8 meeting on the midline. AMNH 30550 has six neurals with costals 7 and 8 meeting on the midline (see table 22). In AMNH 30551 the nuchal is

(right). [E.S. Gaffney, del.]

slightly wider than it is long, with a long anterior margin. The first neural is six-sided with short posterolateral sides contacting the second costal as in Cearachelys and different from all other bothremydids. The second neural is four-sided and short, as in Cearachelys . The third to fifth neurals are six-sided with short anterolateral sides and are longer than wide. The sixth and seventh neurals are wider than long and the eighth neural is reduced and pentagonal in shape; it does not contact the suprapygal. The first costal is similar to that of Cearachelys , being shorter than the length of the second and third costals together. The eighth pair of costals meets on the midline in AMNH 30551 and the seventh and eighth pairs of costals meet on the midline in AMNH 30550 between the neural series and the suprapygal, in contrast to Cearachelys in which the costals are completely separated by the neural series. The suprapygal is roughly triangular in shape, as in other Pelomedusoides; it is wider than long.

The carapace scales are similar to the generalized scale pattern of other Pelomedusoides. The cervical is absent. The second and third vertebrals are roughly as long as wide, different from Cearachelys in which these two scales are wider than long. The marginals are restricted to the peripherals as in all Pelomedusoides.

PLASTRON

The plastron is sutured to the carapace. The plastron is long; its anterior margin reaches the anterior margin of the carapace, as in Cearachelys . The anterior lobe is short and wide at the base, with a semicircular outline, as in most bothremydids but different from Cearachelys , which is longer. The mesoplastron is laterally placed and longer than wide. The bridge is long, longer than the posterior lobe, in contrast to Cearachelys , in which the bridge is shorter than the posterior lobe. The anal notch is V-shaped.

The plastron scale pattern differs between the two shells. The intergular scale is narrow

(right). [J. Kane, del.]

in AMNH 30551 and it covers about a third of the entoplastron. The gular scale is larger, extending onto the entoplastron, and is much wider than the intergular, a uniquecondition among bothremydids. In AMNH 30550, the intergular is wide, larger than the gular scales, which do not extend onto the entoplastron. In both, the humeropectoral sulcus is entirely posterior to the epihyoplastral suture and just behind the entoplastron, as in Cearachelys . The pectoroabdominal sulcus crosses the mesoplastron in both shells, as in Cearachelys and Euraxemys .

Three genera of turtles have been collected in the Cenomanian Kem Kem beds: Galianemys ( Bothremydidae ), Hamadachelys ( Podocnemididae ), and Dirqadim ( Euraxemydidae ), all based on isolated skulls. This turtle fauna has a strong affinity with the Early Cretaceous turtle fauna of the Santana Formation. The shells from Kem Kem described above are similar to bothremydids in the shape of the anterior lobe of the plastron. The shells from Kem Kem share with Cearachelys the first neural being six-sided with short posterolateral sides contacting the second costal and the four-sided second neural. This character is not known in any other Pelomedusoides. This suggests that the shells may belong to Galianemys . The phylogenetic analysis supports this hypothesis by placing these shells in the tribe Cearachelyini . The comparisons between the shells of Galianemys and Cearachelys are shown in table 22.

There is evidence that the? Galianemys shells are not the same taxon. AMNH 30551 has eight neurals with costal 8 meeting on the midline, and a very narrow intergular scale flanked by large gulars that extend onto the entoplastron. This is in contrast to AMNH 30550 that has six neurals with costals 7 and 8 meeting on the midline, and a wide intergular flanked by smaller gulars that do not extend onto the entoplastron. A fine kettle of fish.

(right). [J. Kane, del.]

PHYLOGENETIC ANALYSIS

METHODOLOGY

The phylogenetic analyses were made with PAUP* version 4.0b10 ( Swofford, 2002) using the parsimony algorithm. All characters were run unweighted and unordered, although there is a good argument for ordering at least two characters (characters 20 and 94, see Character Descriptions for discussion). Characters were entered and cladograms examined using MacClade version 4.05 ( Maddison and Maddison, 2000). The data matrix (appendix 3) used as the basis of this analysis has 41 taxa and 175 characters (of these, only 171 are parsimony informative for the core analysis, cladogram 1 [fig. 288], due to the presence of some shell characters that vary only in taxa absent in the core analysis; see below). The PAUP analysis of this matrix, using both heuristic and stepwise addition (random), results in one shortest cladogram (fig. 288) of 382 steps with a consistency index of 0.60, a retention index of 0.81, a rescaled consistency index of 0.49, and a homoplasy index of 0.39. Bootstrapping was run using 100 replicates. Bremer decay indices were obtained using TreeRot version 2 ( Sorenson, 1999).

One caution in using the dataset provided in appendix 3 is the amount of missing data. Taxa with large numbers of missing characters are a concern because the analytic program supplies the missing characters as most consistent with the shortest cladogram, thus ignoring the likely interference of contradictory characters commonly present. Even so, inclusion of taxa with missing data can produce useful results ( Kearney and Clark, 2003). Table 25 lists the amount of missing data for each taxon.

As shown in the decay analysis, there are some groups that lose resolution in the consensus of the 38 cladograms that are one step longer than the MPC. This consensus tree (fig. 289) shows that the Bothremydini loses resolution, with the Foxemydina becoming a multichotomy with the Taphrosphyini and remaining Bothremydini . Zolhafah and Rosasia form a multichotomy with the remaining Bothremydini . The genus Bothremys is paraphyletic with respect to Chedighaii . All the other tribes hold up, but the two species of Galianemys form a multichotomy with Cearachelys . Within the Taphrosphyini , there is a loss of resolution in the basal members, and the species in the genus Taphrosphys are paraphyletic with respect to a Phosphatochelys + Ummulisani + Rhothonemys trichotomy. Araripemys and the Pelomedusidae form a trichotomy with the remaining Pelomedusoides. All of these alternatives are quite reasonable, as they are held together by few characters. Two particular characters that seem a little more subjective than the others, at least in part, are characters 31 and 110 (see Character Descriptions for discussion). Deleting these characters produces 38 trees with the consensus cladogram seen in figure 290.

The 40 taxa in the core analysis are all represented by good skull material; of these, 21 also have associated shells (fig. 291A) and 20 have associated lower jaws (fig. 291B). The shells are relatively well distributed taxonomically with at least one skull-shell association in each tribe of Bothremydidae (see also fig. 314, character 164). In the dataset skull characters are 70 % (122) of the 175 characters, shell characters are 21 % (38), and the remaining 9 % (14) are other postcranial characters. When all postcranial characters are deleted, the same single cladogram (fig. 288) results, showing the dominance of cranial characters in the dataset and their critical importance to the analysis. The addition of a selected group of eight shell-only taxa to the dataset (indicated by asterisk in fig. 292 and in appendix 3) also results in one cladogram, cladogram 2 (fig. 292). However, these shell-only taxa are not representative of the majority of named shell-only taxa. They have been chosen because they have few missing data and do not generate large numbers of trees. Some shell-only taxa (see section on Dubious Taxa) are represent- ed by very incomplete specimens and have been excluded from this analysis and the resulting cladograms.

The influence of shells on the analysis can be observed by examing the results following the exclusion of all cranial characters. The basic dataset (appendix 3) was modified by

TABLE 25 Missing Data Taxa with asterisks are shell-only specimens added to the analysis in figure 292. Taxa with two asterisks have high missing data values, are treated in figures 294 and 295, and are discussed in text.

dropping all cranial characters and all skull-only taxa, with the result seen in figure 293A. This restricted dataset has 28 taxa and 39 characters and produced 2704 trees in a PAUP parsimony analysis with no ordering or weighting. The Pleurodira, Megapleurodira, and Eupleurodira are resolved along with a few lower level bothremydid groups, but most Pelomedusoides form a multichotomy. Admittedly, only 39 characters for 28 taxa are too few to expect complete resolution, but we have used as many shell characters as possible in the analysis. The poor showing of shell characters here seems to be the result of the general conservatism of Pelomedusoides shells. It is possible to find shell characters for alpha-level diagnoses and differentiation in this group. However, even these characters may be questioned because individual variation is so poorly known, and more inclusive comparisons are difficult given our current knowledge of skull-shell associations. Furthermore, most descriptive work on shells emphasize the external suture and scale relations, while many characters can be found in the more complex internal morphology.

In any case, for whatever reason, with the present state of knowledge of this group, the shell alone provides little help in resolving relationships among Pelomedusoides. However, when included with the cranial characters, shell characters do help resolve some groups (see Character Descriptions). In an effort to try squeezing something more out of the shell-only dataset, it has been reanalyzed by weighting using the rescaled consistency index option in PAUP, which is a way of reducing the weight of the more homoplastic characters in favor of the more consistent characters. The resulting consensus of 250 equally parsimonious trees (fig. 293B) is interesting in that it unites almost all bothremydids (shown by asterisks) and excludes the Podocnemididae from the bothremydids.

In order to try to incorporate as much information as possible from the more poorly known shell-only taxa, some have been added singly to the dataset. Most of these are discussed in the Incertae Sedis section following Systematics. Nearly all of these taxa are shells or partial shells, with some lacking as much as 95 % of the characters (for effects of missing data and possible solutions, see Wilkinson, 1995, 2003; Strauss et al., 2003; and Wiens, 2003). Some of these may represent shells of named skulls used in the basic list (e.g., Galianemys and AMNH 30550, AMNH 30551). Adding these taxa to the basic list one at a time allows some resolution of their phylogenetic position and a determination of a level of incertae sedis within a higher taxon without large numbers of trees obscuring the result.

BASIC TAXA OUTSIDE PELOMEDUSOIDES

Although the Pelomedusoides is the focus of this paper, outgroups are significant in the phylogenetic analysis, and there are taxa that have been identified as pleurodires in the literature that are of questionable status and require discussion.

Synapsida/ Diapsida

In order to root the turtle taxa, the main groups of amniotes outside turtles are included as a single taxon. We consider turtles to be the sister group of diapsids, not within diapsids or within pareiasaurs/procolophonids. There are polymorphic codings for some characters, and certain characters that are turtle synapomorphies, like plastral features, are coded ‘‘?’’, so this should not be considered a useful exercise in determining turtle relationships within amniotes. For further literature on these taxa and for other analyses of amniote relationships, see Lee (1997) and Rieppel and Reisz (1999). We are not dealing with the relationships of extinct groups like pareiasaurs and procolophonids to turtles, because to do so would not alter relationships within turtles. We do not consider the hypothesis that turtles are within diapsids (e.g., Braga and Rieppel, 1997) to have merit, but even accepting this would not alter relationships within turtles as analyzed here. Proganochelys still comes out as the sister group to cryptodires plus pleurodires.