The Paleocene Teberemt Formation south of the Adrar des Iforas Mountains, between Saguirilidad and In Fargas, Mali, yielded a nearly complete skull of a new genus and species of side-necked turtle, Acleistochelys maliensis. Acleistochelys is a member of the family Bothremydidae Baur, 1891, because: 1 the fossa precolumellaris is absent, 2 the foramen stapedio-temporale faces anteriorly, 3 the eustachian tube is separated from the stapes by bone, and 4 an exoccipital-quadrate contact is present. Within the Bothremydidae, Acleistochelys belongs to the tribe Taphrosphyini because: 1 the maxilla-quadratojugal contact is absent, 2 the palate is dorsally arched, 3 there is only a small contribution of the palatine to the triturating surfaces, and 4 the septum orbitotemporale is at least partially open. Acleistochelys is most closely related to Azabbaremys because both share a narrow vomer lacking a posterior attachment to the palatines.
The specimen was found in a marine limestone associated with crocodiles, echinoids, and mollusks.
Introduction
The Bothremydidae is an extinct group of side-necked turtles recently reviewed and expanded in content by Gaffney et al. 2006. The purpose of the present paper is to name and describe a new bothremydid, based on CNRST SUNY 199 from the Paleocene of Mali Tapanila et al., 2004. The skull description follows the outline in Gaffney et al. 2006: appendix 1 for ease of comparison with other bothremydids. Based on the phylogenetic analysis of Gaffney et al. 2006, Acleistochelys maliensis is a member of the tribe Taphrosphyini as characterized in Gaffney et al. 2006. Figures, descriptions, and references to all of the bothremydid taxa referred to in this paper can be found in Gaffney et al. 2006. The taxonomy also follows that reference. The skull here named Acleistochelys maliensis is included in the data set of Gaffney et al. 2006: appendix 3 and is shown in cladograms in figs. 288–314 as CNRST SUNY 199. The reader should see this work for further discussion of this phylogenetic analysis.
Geology
Geological Setting
During the latest Cretaceous Campanian–Maastrichtian through middle Eocene, the Trans-Saharan Seaway inundated portions of central West Africa Petters, 1979. In northern Mali, three transgressive–regressive cycles are recorded by a relatively thin <100 m sequence of sedimentary strata dominated by sandstones, shales, limestones, and phosphates e.g., Radier, 1959; Bellion et al., 1989; Moody and Sutcliffe, 1993; Tapanila et al., 2004. These units were deposited in terrestrial, marginal marine, and open marine settings within and along the margins of the Trans-Saharan Seaway, and they preserve a rich vertebrate and invertebrate fossil record.
The Late Cretaceous–Paleogene Malian outcrops that are the focus of this study fig. 1 are located in a narrow, half-graben structure, known as the Gao Trench or “Detroit Soudanais.” Paleontologic investigations of invertebrate faunas demonstrate that the Trans-Saharan Seaway may have provided at times a dispersal route via the Gao Trench for marine fauna between the Tethys Sea and South Atlantic Ocean Kogbe, 1981; Moody and Sutcliffe, 1993. Today this vast, desolate region is characterized by expansive low-relief plateaus that are capped by resistant limestone beds. Small shifting dune fields and vast stretches of desert pavement cover this region and obscure regional stratigraphic correlations.
Sedimentology of Mali 17
Description
The turtle specimen described herein was recovered from locality Mali 17, located south of the Adrar des Iforas Mountains, between Saguirilidad and In Fargas. Mali 17 is situated within a 6–10 m thick sequence of richly fossiliferous phosphatic sandstones, shales, and limestones. The sedimentology at Mali 17 is characterized by a basal 1 m thick white to pale yellow phosphatic sandstone unit 1 with nodular chert concretions that grades upward into a thin 0.5 m, dark gray sandy shale unit 2 containing ray tooth plates and abundant bioturbation fig. 1b. A 3 m thick chalky, micritic limestone with thin shale partings unit 3 overlies this unit and preserves multiple articulated and associated turtle remains, including the specimen described in this report. Other well-preserved vertebrate and invertebrate fossils were also recovered from this laterally extensive bed, including the remains of crocodiles, encrusting oysters Ostrea multicostata, nautiloids Deltoidonautilus sp., gastropods Gistortia sp., and echinoids e.g., Linthia sudanensis, Oriolampas michelini. The entire succession is capped by a distinctive 2 m thick, cliff-forming, blocky, recrystallized limestone bed unit 4. This unit preserves rare, moldic nautiloids and abundant trace fossils.
Interpretation
Mali 17 represents an upward fining, retrogradational marine sequence. Increasing water depth is interpreted based on reduction of siliciclastics and increasing purity of carbonates upward through the sequence. The fauna recovered from Mali 17 are consistent with relatively low-energy, normal marine conditions.
Strata of Mali 17 likely belong to the middle to upper portion of the Teberemt Formation of Moody and Sutcliffe 1991, 1993, which they interpreted as Paleocene. Paleogene echinoderms, such as Oriolampas michelini and Linthia sudanensis, recovered from the turtle-bearing unit 3 preclude a Cretaceous age assignment, while putative Eocene Ypresian vertebrates documented from stratigraphically higher phosphate conglomerates fig. 1A; ~68 m level at nearby Tamaguilelt Pascal and Traore, 1989; Patterson and Longbottom, 1989; Moody and Sutcliffe, 1993; O'Leary et al., 2006 provide an upper age limit. Moody and Sutcliffe 1993 also correlated the Teberemt Formation with a similar stratigraphic sequence on the west side of the Tilemsi valley, near Tichet, which Bellion et al. 1989 dated as late Paleocene using benthic foraminifera and ostracodes. Based on faunal analysis and detailed outcrop correlation with strata exposed below and above Mali 17, we tentatively agree with the Paleocene age assignment of Moody and Sutcliffe 1991, 1993 for these deposits, although an early Eocene age cannot necessarily be precluded.
Systematics
ORDER TESTUDINES LINNEAUS 1758 OR BATSCH 1788
INFRAORDER PLEURODIRA COPE, 1864
FAMILY BOTHREMYDIDAE BAUR, 1891
TRIBE TAPHROSPHYINI GAFFNEY, TONG, AND MEYLAN, 2006
SUBTRIBE TAPHROSPHYINA GAFFNEY, TONG, AND MEYLAN, 2006
Acleistochelys, new genus
Etymology
Aklystos, Greek for “sheltered”, or “hollowed”, in allusion to the small, paired pits on the triturating surface; chelys, Greek for “turtle”.
Diagnosis
A bothremydid pleurodire with these unique characters among the tribe Taphrosphyini: small pit formed by jugal, maxilla, and palatine on triturating surface; jugal exposed on triturating surface; accessory ridge present on anterior triturating surface; wide palatine-basisphenoid contact separating pterygoids on midline; supraoccipital-quadrate contact present; basioccipital narrowly enters condylus occipitalis; palatine-jugal contact in small septum orbitotemporale. Other distinguishing characters are: skull relatively long and narrow; fossa pterygoideus deep and narrow as in Nigeremys but in contrast to Taphrosphys; foramen posterius canalis carotici interni formed by pterygoid, basisphenoid, and quadrate in contrast to Phosphatochelys; small remnant of septum orbitotemporale present consisting of ventral parietal process as in Phosphatochelys but in contrast to Taphrosphys.
Discussion
See table 1 for a comparison of genera in the tribe Taphrosphyini.
Table 1
Genera of the Tribe Taphrosphyini
Acleistochelys maliensis, new species
Type Specimen
CNRST SUNY 199 figs. 2–Figure 3Figure 45, partial skull including fragments of the lower jaw, lacking some of the left temporal region, and both posterior parts of the skull roof. Skull measurements are in table 2. Associated shell fragments include the anterior margin of the nuchal fig. 6, neurals one and four fig. 7, and peripheral two fig. 7. A partial cervical vertebra figs. 8, 9 and some pelvic fragments are present.
Figure 2
Acleistochelys maliensis n.gen., n.sp. Partially restored skull based on CNRST SUNY 199. A, dorsal; B, ventral; C, right lateral.
Figure 4
Acleistochelys maliensis n.gen., n.sp. Photographs of CNRST SUNY 199. A. dorsal; B, ventral; C, anterior; D, right lateral; E, posterior; F, left lateral.
Figure 5
Acleistochelys maliensis n.gen., n.sp. Line drawings of CNRST SUNY 199. A. dorsal; B, ventral; C, anterior; D, right lateral; E, posterior; F, left lateral.
Table 2
Cranial Measurements of Acleistochelys maliensis in centimeters
Figure 6
Acleistochelys maliensis n.gen., n.sp. CNRST SUNY 199. Anterior margin of nuchal bone, dashed line indicates midline, sulcus separates the first pair of marginal scales. Upper, dorsal view, right half restored; lower, ventral view.
Figure 7
Acleistochelys maliensis n.gen., n.sp. CNRST SUNY 199. Dorsal views of: A, posterolateral part of ?peripheral two; B, neural one; C, neural four.
Description
Prefrontal
Preservation: Both prefrontals are present: the right one is nearly complete, the left is missing some of its lateral edge.
Contacts: The prefrontal in Acleistochelys has the same contacts as in Azabbaremys, Nigeremys, and Arenila: prefrontal on midline, maxilla anteroventrally, and frontal posteriorly. There is no palatine or parietal contact.
Structures: The dorsal plate of the prefrontal in Acleistochelys is similar to that in Azabbaremys, but not strongly convex dorsally as in Azabbaremys. The dorsal margin of the apertura narium externa in Acleistochelys is well posterior to the ventral margin in contrast to Azabbaremys in which the dorsal margin is almost directly above the ventral margin.
The size of the fossa nasalis and sulcus olfactorius are similar in both Acleistochelys and Azabbaremys. The ventral process of the prefrontal is narrow in Acleistochelys.
Frontal
Preservation: The right frontal in CNRST SUNY 199 is nearly complete; it is missing only its posterolateral corner. The left one is missing most of its lateral half.
Contacts: The frontal in Acleistochelys contacts the prefrontal anteriorly, the other frontal medially, the postorbital posterolaterally, and the parietal posteriorly, all as in Azabbaremys, Nigeremys, and Arenila, and in contrast to Phosphatochelys and Rhothonemys, which lack a postorbital contact.
Structures: The frontal in Acleistochelys enters the orbital margin, similar to that seen in Azabbaremys and Nigeremys.
Parietal
Preservation: Both parietals are present but damaged. The dorsal plate on the left side retains only its anteromedial portion. The right dorsal plate is more complete laterally, but it also lacks all of the temporal margin. The processus inferior parietalis, however, is nearly complete on both sides, the left one having some breakage ventrally.
Contacts of dorsal plate: The dorsal plate in CNRST SUNY 199 preserves the midline parietal contact and the medial part of the frontal contact anteriorly. The right parietal has the anterolateral contact with the postorbital preserved.
Structures of dorsal plate: The degree of temporal emargination in Acleistochelys, although not determinable exactly, was probably not as extensive as in Nigeremys, because the remaining broken margin is relatively thick and extends about as far as the natural margin in Nigeremys and probably Arenila, although that skull also lacks most of the parietal dorsal plate.
The septum orbitotemporale see Gaffney et al., 2006: fig. 78 in the tribe Taphrosphyini, may be reduced or absent Gaffney et al., 2006: character 28. This is best seen in Phosphatochelys Gaffney et al., 2006: fig. 202. However, the group that lacks the septum that is, Taphrosphys, Labrostochelys, Rhothonemys, Ummulisani also lacks the ventral parietal process seen in Phosphatochelys Gaffney et al., 2006: fig. 202. This process is a remnant of the lateral wall of the septum orbitotemporale and reaches the palatine above the base of the processus trochlearis pterygoidei lateral to the sulcus palatinopterygoideus. This process also occurs in Acleistochelys, although there are differences from Phosphatochelys: In Acleistochelys it is thinner and longer, and contacts the palatine only, whereas in Phosphatochelys it is wider and shorter and contacts the palatine anteriorly and the pterygoid posteriorly. The entire ventral portion of the septum orbitotemporale remnant differs in Acleistochelys and Phosphatochelys in the position of the jugal and palatine see Jugal, Palatine. Whether this structure should be considered homologous in Acleistochelys and Phosphatochelys is a fascinating question.
Contacts of processus inferior parietalis: In Acleistochelys the vertical wall of the parietal contacts the pterygoid ventrally, the prootic posteroventrally, and the supraoccipital posteriorly. Although somewhat damaged, the processus inferior parietalis in Acleistochelys seems to be relatively thin, as in Azabbaremys, but still seems to contact the palatine, a condition that is unclear in Azabbaremys.
Structures of processus inferior parietalis: The foramen interorbitale is relatively large in Acleistochelys, as in Azabbaremys and Phosphatochelys, and the processus inferior parietalis is narrow, also as in Azabbaremys. The foramen nervi trigemini, preserved on both sides, is formed by the usual bones: parietal anterodorsally, prootic posterodorsally, and pterygoid ventrally.
Jugal
Preservation: Neither jugal is complete, both are missing their posterodorsal margins. The left jugal preserves the ventral margin, and the right preserves the anterodorsal margin. The medial process on both sides is present.
Contacts of lateral plate: The jugal in Acleistochelys has a long anteroventral contact with the maxilla as in Azabbaremys and at least an anterodorsal contact with the postorbital, but the length of the bone is indeterminate. The jugal contacts the quadrate posteroventrally in a suture that is more horizontal than in Azabbaremys, showing that at least some of the jugal in Acleistochelys lies above the quadrate as in Taphrosphys, in contrast to abutting against it, as in Azabbaremys. Likely contacts with the quadratojugal are indeterminate due to the absence of that bone and that part of the jugal margin.
Structures of lateral plate: The jugal in Acleistochelys is relatively large and, to the extent preserved, similar to that in Azabbaremys in position and shape. The jugal forms the posteroventral margin of the orbit and is prevented from reaching the cheek margin by a quadrate-maxilla contact of the type seen in Taphrosphys, Labrostochelys, and Azabbaremys.
Contacts of medial process: The medial process of the jugal in Acleistochelys contacts the maxilla anteriorly, the pterygoid posteriorly, and the palatine medially, as in Azabbaremys and other Taphrosphyini. The jugal also contacts the palatine in the remnant of the septum orbitotemporale, lateral to the sulcus palatinopterygoideus. This is in contrast to the condition in Azabbaremys, which lacks that palatine contact and partial wall of the septum orbitotemporale. Phosphatochelys, however, does have a condition similar to that in Acleistochelys.
Structures of medial process: The medial process of the jugal is widely exposed in the orbital floor, more so than in Azabbaremys Gaffney et al., 2006: fig. 280. The jugal forms a dorsally opening concavity in the floor of the fossa orbitalis, as in Azabbaremys but in contrast to AMNH 30521 in which the pocket is formed by the maxilla Gaffney et al., 2006: fig. 211.
On the ventral surface the jugal in Acleistochelys is unique among Taphrosphyini in being exposed on the triturating surface, although the extent is much less than in other bothremydids Gaffney et al, 2006: fig. 11. However, in Azabbaremys the jugal extends onto the margin of the triturating surface Gaffney et al., 2006: fig. 215B in a condition that isn't very far from that in Acleistochelys. As in the group consisting of Bothremys and its close relatives, the exposure of the jugal in Acleistochelys is apparently correlated with the development of a small pit, formed by the jugal, maxilla, and palatine. The jugal forms the tip and posterior wall with the maxilla anteriorly and laterally and the palatine medially. The triturating surface in Acleistochelys is narrow, as in other members of the Taphrosphyini, and in contrast to the wide surface of the Bothremydini, which accommodates the larger pit in that group. The pit in Acleistochelys is smaller and shallower, with poorly defined anterior and lateral margins in comparison to the ones in the Bothremydini. Nonetheless, it is a structure completely absent in any living turtle, but that arose independently within the Bothremydidae and an undescribed Moroccan cryptodire Gaffney et al., 2006. Its function, other than cracking the odd walnut, remains enigmatic.
Quadratojugal
Preservation: Small fragments of each quadratojugal are suspended from each quadrate. The right one is more complete, but has broken anterior and dorsal margins.
Contacts: The quadratojugal fragment contacts the quadrate ventrally and the squamosal posteriorly, preventing quadrate exposure in the temporal margin. Although it may be slight, all bothremydids with the area preserved have a quadratojugal-squamosal contact. As in other members of the Taphrosphyini, there is no quadratojugal contact.
Structures: Although it is not preserved, the quadratojugal presumably reached the temporal margin.
Squamosal
Preservation: Both squamosals are present. The right one is more complete and lacks only a portion of its lateral margin.
Contacts: The squamosal in Acleistochelys has the usual bothremydid contacts: quadrate anteriorly, opisthotic medially, and quadratojugal anterodorsolaterally.
Structures: The squamosal in Acleistochelys projects posteriorly to a greater extent than in Azabbaremys or any other Taphrosphyini except Labrostochelys. As in Azabbaremys, it lacks the posteroventral vertical flange seen in other Taphrosphyini. It also agrees with Azabbaremys and differs from other Taphrosphyini in lacking the lateral tubercle see Gaffney et al., 2006: Characters 25 and 26. The antrum postoticum is moderate in size in Acleistochelys, compared to its absence in Azabbaremys, so the squamosal probably contains a cavity for it.
Postorbital
Preservation: Only the anterior part of the right postorbital is present in Acleistochelys, and it is broken into two pieces separated by missing bone.
Contacts: As preserved, the postorbital contacts the parietal posteromedially, the frontal anteromedially, and the jugal anteroventrally.
Structures: The medial process of the postorbital that usually forms the dorsolateral part of the septum orbitotemporale, is not formed in Acleistochelys, in agreement with all other Taphrosphyini except Nigeremys and Arenila. There is a low, transverse ridge marking the position of the septum, slightly larger than in Azabbaremys.
The postorbital forms the posterodorsal margin of the orbit as in other bothremydids. Its posterior extent is indeterminate.
Premaxilla
Preservation: Both premaxillae are present in CNRST SUNY 199: the right is nearly complete, the left lacks some of the horizontal plate.
Contacts: As in other bothremydids the premaxilla contacts the maxilla posterolaterally and the other premaxilla medially. As in Labrostochelys, Azabbaremys, Nigeremys, and Arenila, the premaxilla contacts the vomer posteromedially. In contrast to Azabbaremys, the premaxilla in Acleistochelys rises to nearly reach the prefrontal along the margin of the aperture narium externa.
Structures on dorsal surface: In contrast to Azabbaremys, Acleistochelys has a small, median tubercle formed on the midline of the ventral margin of the aperture narium externa.
Structures on ventral surface: The entire triturating surface of Acleistochelys is very distinct from that in the other Taphrosphyini, and the premaxillary portion is no exception. The skull in Acleistochelys is narrower than in the other members of this group, except Taphrosphys and Labrostochelys, and the anterior outline of the labial ridge reflects this. The labial ridge formed by the premaxilla is more acute in Acleistochelys with respect to the midline than in all other members of the Taphrosphyini. The ridge is thicker than in Taphrosphys, but not as thick as in Azabbaremys. The ridge has an acute ventral margin as in Azabbaremys. The labial ridge in Acleistochelys is slightly convex ventrally in anterior view, distinct from the sharper, more defined ventral protuberance seen in Azabbaremys.
Acleistochelys has a well-developed accessory ridge on the triturating surface that is morphologically unique among pleurodires. The ridge is expressed anteriorly on the premaxillae and flattens out posteriorly, disappearing about halfway down the maxillary length. The ridge is unusual in being broad anteroposteriorly on the premaxilla, which is defined anteriorly by a narrow trough and posteriorly by a median concavity on the posterior half of the premaxilla. There is no midline concavity as in Azabbaremys and other members of the Taphrosphyini. The premaxilla of Acleistochelys is much longer than that in Azabbaremys. A long premaxilla also occurs in Nigeremys and Arenila, but the morphology of the triturating surface is much simpler in those species, lacking an accessory ridge. The strong corrugations unique to Azabbaremys are completely absent in Acleistochelys.
Maxilla
Preservation: Both maxillae are present. The left one lacks the anterior part of the horizontal plate, the right one lacks the posteroventral end of the vertical plate. Together all of the maxilla morphology is available.
Contacts of vertical plate: As in other Taphrosphyini, the maxilla of Acleistochelys contacts the premaxilla anteromedially, the prefrontal anterodorsally, the jugal posterodorsally, and the quadrate posteroventrally.
Structures of vertical plate: The maxilla barely enters the margin of the apertura narium externa in contrast to the wider condition in other Taphrosphyini. The maxilla forms the anteroventral margin of the orbit, but it is less extensive than in Azabbaremys. The ventral margin of the maxilla trends slightly ventrally to meet the quadrate; there is no indication of a cheek emargination.
Contacts of horizontal plate: On the dorsal surface the maxilla contacts the jugal posteriorly and the palatine posteromedially. On the ventral surface the maxilla contacts the premaxilla anteromedially in a long suture, but does not reach the vomer.
Structures of horizontal plate: The labial ridge of the maxilla in Acleistochelys is thinner than that ridge in Azabbaremys and Nigeremys, but much thicker than in Ummulisani and Labrostochelys. The accessory ridge developed on the premaxilla extends onto the anterior part of the maxilla, but flattens posteriorly and disappears well anterior to the posterior end of the triturating surface. The triturating surface is parallel-sided with no distinct lingual ridge marking its medial limit.
The triturating surface has a shallow but distinct pit at its posterior-most margin, formed by the maxilla anteriorly, the palatine medially, and the jugal posteriorly see Jugal. The shallow trough marking the medial edge of the labial ridge, much deeper anteriorly, leads into the pit.
The maxilla forms the anterior part of the floor of the fossa orbitalis. It forms part of the distinct ridge that forms the orbital margin. The maxilla forms most of a dorsally opening concavity or fossa in the anterior part of the orbital floor. This is probably homologous to the ventral pocket seen in Ummulisani Gaffney et al., 2006: fig. 211B, “fossa orbitalis ventral pocket”, but the fossa in Acleistochelys is much smaller than the one in Ummulisani. A ventral concavity also occurs in other members of the Taphrosphyini, but the morphology is variable. Azabbaremys also has a concavity formed by the maxilla, but it is farther posterior than in Acleistochelys.
Vomer
The single vomer is present in CNRST SUNY 199. It seems to be complete, but is bent slightly to the right. Its completeness is an issue because as preserved it has the unusual morphology of being wide anteriorly where it has a solid contact with the premaxillae, and narrows posteriorly ending at a point before reaching the palatines in strong contrast to other turtles that have a solid palatine contact. The interesting feature of this morphology is that it also occurs in Azabbaremys Gaffney et al., 2006: fig. 216, but not in other members of the Taphrosphyini. It is one of the two characters holding Acleistochelys and Azabbaremys together, the other being an absence of the vomer-maxilla contact Gaffney et al., 2006: characters 44, 47. In the single specimens available of Azabbaremys and Acleistochelys there is no sign of a broken edge along the length of the vomer, and the vomers in each seem to be well preserved. However, Acleistochelys, in contrast to Azabbaremys, is missing the anterior margin of the palatines, so no supporting examination of that morphology is possible.
Palatine
Preservation: Both palatines are present and are complete except for their anterior margins, which are broken and lacking a finished edge. Their dorsal surfaces are visible but broken medially with some areas unclear.
Contacts: The palatine contacts the maxilla anterolaterally, the other palatine medially, the pterygoid posteriorly and laterally, and the jugal anterolaterally, all of which occur in other members of the Taphrosphyini. Acleistochelys is unique, however, in having a strong posteromedial contact with the basisphenoid, correlated with the very large size of the palatines and the relatively small size of the pterygoids. The related separation of the pterygoids on the midline is also unique to Acleistochelys among the members of the Taphrosphyini. It is possible that the palatines reach the basisphenoid in the poorly preserved Nigeremys, but the known sutures are not similar to those in Acleistochelys.
On the dorsal surface the palatine in Acleistochelys contacts the parietal just lateral to the sulcus palatinopterygoideus.
Structures on dorsal surface: The dorsal surface of the palatine in Acleistochelys forms the medial part of the floor for the fossa orbitalis and the medial part of the low, transverse ridge marking the remnant of the septum orbitotemporale see also Jugal. In all of the members of the Taphrosphyini, except Nigeremys and Arenila, the septum orbitotemporale is partially or completely open. In the case of Acleistochelys and Phosphatochelys more of the septum is present than in Taphrosphys, Azabbaremys, and Rhothonemys. In Acleistochelys and Phosphatochelys there is a ventral process of the parietal see Parietal that meets at least the palatine and forms a partial lateral limit to the sulcus palatinopterygoideus Gaffney et al., 2006: Introduction to Pleurodire Skull Morphology. In Acleistochelys the palatine contacts the parietal dorsally, the jugal anteroventrally and the pterygoid posteroventrally. In Phosphatochelys the palatine contacts the parietal dorsally and the pterygoid posteriorly. In Azabbaremys the palatine has no dorsal contacts and bears only a low ridge in the position of the septum orbitotemporale. The other members of the Taphrosphyini are either too incomplete or covered with matrix in the area, preventing palatine comprehension.
Structures on ventral surface: The palatine in Acleistochelys forms the high choanal arch, similar to that in Azabbaremys. Although the arched palate is actually higher in Azabbaremys than in Acleistochelys, in Acleistochelys the palatine curves ventrally along its lateral edge to form a parasagittal, vertical wall that is a posterior continuation of the apertura narium interna margin, a morphology very different from Azabbaremys and the other members of the Taphrosphyini.
Quadrate
Preservation: Both quadrates are present but incomplete in CNRST SUNY 199. The processus articularis, medial process, and central area of the cavum tympani are present on both sides, but the anterior margin, anterior part of the cavum tympani and some of the dorsal margin are missing on both sides.
Contacts on lateral surface: The anteroventral contact with the maxilla is present on the left side, and there are small fragments of the quadratojugal on both sides, but the contacts of the jugal and postorbital with relation to the quadrate are indeterminate. The squamosal is present on both sides in its unalterable position.
Structures on lateral surface: Little indication of the temporal extent remains on the quadrate, but the left side seems to show a natural edge similar to that in Azabbaremys indicating emargination of at least the extent seen in that form.
The cavum tympani is incomplete, but what is present shows the incisura columellae auris as a fully enclosed canal and the sulcus eustachii as a narrow groove posteroventral to the incisura columellae auris, as in Azabbaremys and other members of the Taphrosphyini. The area where the fossa precolumellaris should be absent is not preserved. The ventral shelf seen in many other members of the Bothremydidae is present. The antrum postoticum seems to be present on the left side and is moderate in size, in the scheme of Gaffney et al. 2006. This would agree with Nigeremys, Arenila, and Rhothonemys, in contrast to Azabbaremys, which has a completely filled incisura columellae auris.
Contacts on dorsal and anterior surface: The quadrate contacts the prootic anteromedially, the opisthotic posteromedially and the squamosal posterolaterally as in other members of the Taphrosphyini. However, in contrast to all other members of this group indeterminate in Nigeremys and Arenila there is a supraoccipital-quadrate contact visible on both sides in Acleistochelys. This contact is a Bothremydidae synapomorphy, reversed in the Taphrosphyini, so its presence in Acleistochelys is either a retained primitive condition for Bothremydidae or a reversal.
Structures on dorsal and anterior surface: The foramen stapedio-temporale in Acleistochelys opens anteriorly and lies close to the foramen nervi trigemini see Gaffney et al., 2006, for distribution of these characters.
Contacts on ventral surface: As in other members of the Taphrosphyini, the quadrate in Acleistochelys contacts the pterygoid anteromedially, the basisphenoid medially, and the basioccipital posteromedially. The extent of contact with these elements varies a great deal within this group Gaffney et al., 2006: fig. 11. In Acleistochelys these contacts are roughly similar to those seen in Labrostochelys, in contrast to the much narrower basisphenoid contact of Azabbaremys.
Structures on ventral surface: The fossa pterygoidea is absent in all members of the Taphrosphyini except Nigeremys, Arenila, and Acleistochelys. In Acleistochelys the fossa is deep, with a strong anterolateral wall of quadrate and the foramen posterius canalis carotici interni in the anterior part of its roof. In the cladogram of Gaffney et al. 2006: fig. 288 the deep and narrow fossa pterygoidea evolves three times independently: in Acleistochelys, in Nigeremys + Arenila, and in Foxemys + Polysternon. Azabbaremys and Acleistochelys are only weakly resolved. Moving Acleistochelys to the group Nigeremys + Arenila results in a cladogram only one step longer than the shortest.
The foramen posterius canalis carotici interni in Acleistochelys is formed by the pterygoid anteriorly, the quadrate posterolaterally, and the basisphenoid posteromedially. Within the members of the Taphrosphyini, only Taphrosphys agrees with Acleistochelys in having these three bones form the foramen. However, the formation of the fossa pterygoidea is apparently associated with the more dorsal position of the foramen posterius canalis carotici interni Gaffney et al., 2006. In Azabbaremys the foramen is formed by the pterygoid and quadrate. The excavation of the fossa pterygoidea could result in the exposure of the more dorsal portion of the basisphenoid, so the canalis carotici interni would appear to have moved dorsally. Despite the attraction of this elegant hypothesis, the presence of the same condition in Taphrosphys, which completely lacks a fossa pterygoidea, makes it dubious.
The condylus mandibularis in Acleistochelys is slightly anterior to the position of the condylus occipitalis. This is coded as “near or on plane of basioccipital-basisphenoid suture” in Gaffney et al. 2006: Character 60. The condylus mandibularis in Acleistochelys is just posterior to the suture. In Azabbaremys the condylus mandibularis is slightly more posterior but still similar in position. This is in contrast to the more posterior position of the condylus mandibularis in Nigeremys + Arenila.
Contacts on posterior surface: As in the other members of the Taphrosphyini, the quadrate in Acleistochelys contacts the squamosal dorsolaterally, the opisthotic dorsomedially, and the exoccipital medially.
Structures on posterior surface: The fenestra postotica in Acleistochelys is an oval opening completely enclosed by the quadrate ventrally and the opisthotic dorsally. There is no subdivision of the fenestra postotica.
Pterygoid
Preservation: Both pterygoids are present, the left one is nearly complete lacking some of the edge of the processus trochlearis pterygoidei and some of the crista pterygoidea. The right pterygoid lacks the lateral part of the processus trochlearis pterygoidei.
Contacts on ventral surface: The pterygoid in Acleistochelys is unusually shaped. The common transverse contact with the palatine seen in most turtles is sharply angled anterolateral to posteromedial to such an extent that the nearly universal medial contact of the pterygoids is lost in Acleistochelys. Some bothremydids do have a very short pterygoid contact see Gaffney et al., 2006, but they do not lose the contact. Acleistochelys has a long anterolateral process of the pterygoid extending along the palatine to reach the jugal near the foramen palatinum posterius, as in nearly all turtles. The pterygoid in Acleistochelys contacts the basisphenoid medially and the quadrate posterolaterally.
Structures on ventral surface: The processus trochlearis pterygoidei is present on both sides of CNRST SUNY 199. The right one is broken, but most of the process seems to be present on the left. As in other members of the Taphrosphyini, the processus trochlearis pterygoidei is relatively small compared with other bothremydids and podocnemidids. Although some of the thin edge and much of the flange is broken away, the preserved size seems to be close to its original size because the preserved bone is very thin and tapered along the broken margin.
The fossa pterygoidea see Quadrate is deep and narrow, as in Nigeremys + Arenila. It is formed by the pterygoid anteriorly, the quadrate posterolaterally, and the basisphenoid posteromedially. The foramen posterius canalis carotici interni see Quadrate lies in the anterior part of the roof of the fossa pterygoidea and is formed by the pterygoid, quadrate, and basisphenoid. On both sides there is a small foramen, anterior to the foramen posterius canalis carotici interni but within the fossa pterygoidea, that leads into a canal running anteriorly in the pterygoid. This foramen appears to be the foramen nervi vidiani, which occurs in this position in some other bothremydids Gaffney et al., 2006: figs. 276, 277.
The foramen palatinum posterius is formed anteromedially by the palatine and posterolaterally by the pterygoid. In most bothremydids this suture is transverse, but in Acleistochelys it is sharply oriented posteromedially and the foramen palatinum posterius has retained its position in the suture but not the palate in general.
Contacts on dorsal surface: The pterygoid in Acleistochelys contacts the palatine anterodorsally and the jugal anteriorly at the base of the processus trochlearis pterygoidei, but due to the near absence of the septum orbitotemporale there is no postorbital contact. The crista pterygoidea contacts the parietal dorsally in its anterior part, then the quadrate posterior to the foramen nervi trigemini.
Structures on dorsal surface: The dorsal surface of the pterygoid in Acleistochelys has the relatively narrow crista pterygoidea, similar to that in Azabbaremys. The sulcus palatinopterygoideus is relatively open laterally in Acleistochelys, but its floor is formed by the palatine anteriorly and the pterygoid posteriorly. The pterygoid also forms a low wall along the lateral margin of the sulcus, from the processus trochlearis pterygoidei to the narrow dorsal process marking the anterior remnant of the sulcus palatinopterygoideus.
Supraoccipital
Preservation: The ventral body of the supraoccipital in CNRST SUNY 199 is largely complete, although some is missing in cracks on the left side, and there is a slight amount of lateral distortion. The crista supraoccipitalis, however, is broken away and the dorsal part of the supraoccipital, although present, is displaced.
Contacts: The supraoccipital of Acleistochelys has the usual contacts found in members of the Taphrosphyini: parietals dorsally and anteriorly, prootic anterolaterally, opisthotic posterolaterally, and exoccipitals posteroventrally. In addition, and unique among this group, there is a quadrate contact laterally see Quadrate. This is clear and visible on both sides, formed in the usual bothremydid manner by a lateral lappet of supraoccipital.
Structures: The crista supraoccipitalis is broken off close to its base, but the broken edge is relatively high and thick, more so than in Azabbaremys. The foramen magnum is higher than wide and oval, as in Azabbaremys.
Exoccipital
Preservation: Both exoccipitals are present and nearly complete in CNRST SUNY 199, with a slight amount of dorsoventral distortion.
Contacts: As in other bothremydids the exoccipital of Acleistochelys contacts the supraoccipital dorsally, the opisthotic dorsolaterally, the quadrate ventrolaterally and the basioccipital ventrally.
Structures: The condylus occipitalis in Acleistochelys is formed mostly by the two exoccipitals with the basioccipital entering the condylus articularis surface slightly. This is unusual for most advanced bothremydids, which have the basioccipital excluded from the articular surface and restricted to the neck of the condylus occipitalis.
The foramen nervi hypoglossi is a single, large opening in Acleistochelys, in contrast to the usual two or more smaller openings seen in other bothremydids. The foramen nervi hypoglossi is just dorsolateral to the neck of the condylus occipitalis, opening ventrolaterally. In Azabbaremys the foramen is smaller, paired, and placed more dorsally.
The foramen jugulare posterius in Acleistochelys is closed by the dorsal and ventral plates of the exoccipital with a thin process of quadrate between them, as in Azabbaremys.
Basioccipital
Preservation: The basioccipital in CNRST SUNY 199 is nearly complete but visible only ventrally.
Contacts: As in other bothremydids, the basioccipital of Acleistochelys contacts the basisphenoid anteriorly, the quadrate laterally, and the exoccipital posterodorsally.
Structures: The basioccipital slightly enters the articular surface of the condylus occipitalis see Exoccipital. There is a relatively deep, median concavity, facing posteroventrally, formed by basioccipital and basisphenoid, not seen in Azabbaremys but present in the Nigeremys + Arenila group. The basioccipital of Acleistochelys is relatively short as in Azabbaremys.
Prootic
Preservation: Both prootics are present and relatively well preserved, although the right one is a little chewed on ventrally.
Contacts: As in the other members of the Taphrosphyini, the prootic of Acleistochelys contacts the parietal dorsomedially, the quadrate laterally, the supraoccipital posterodorsally, and the pterygoid ventrally. Uniquely among this group, there is no opisthotic contact due to the supraoccipital lappet contacting the quadrate see Quadrate. It is extremely likely that an opisthotic-prootic contact underlies the supraoccipital lappet as in the other bothremydids.
Structures: The prootic of Acleistochelys has an almost cryptodiran anterior overhang, above the foramen nervi trigemini and foramen stapedio-temporale. Neither foramen is visible in dorsal view and both are close together Gaffney et al., 2006. The foramen nervi trigemini is formed by the prootic posterodorsally, the parietal anterodorsally, the pterygoid anteroventrally, and a small sliver of quadrate seems to enter it posteroventrally. The quadrate rarely enters the foramen nervi trigemini, but, as in Phosphatochelys and Taphrosphys at least, the sulcus cavernosus is exposed in the enlarged foramen and the quadrate forms part of the sulcus cavernosus.
The foramen stapedio-temporale is poorly preserved on the right side, but is clearly visible on the left. It lies in the prootic-quadrate suture, but is more dorsal than that foramen in Azabbaremys. The foramen has an irregular shape in CNRST SUNY 199, possibly due to erosion of the thin bone along its lateral margins. The foramen is at the lateral end of a groove that extends ventromedially into the sulcus cavernosus and foramen nervi trigemini, as seen in Azabbaremys and some other bothremydids that have this area well preserved. Despite the deep fossa pterygoidea on the ventral surface of the skull, there is no exposure of the prootic.
Opisthotic
Preservation: Both opisthotics are present; the left one is missing its dorsomedial part.
Contacts: As in the other bothremydids, the opisthotic of Acleistochelys contacts the supraoccipital anteromedially, the quadrate anterolaterally, the squamosal posterolaterally, and the exoccipital posteromedially. There is no external prootic contact due to the supraoccipital-quadrate contact.
Structures: The fenestra postotica is formed by the opisthotic dorsally and by the quadrate ventrally. It is completely enclosed as in other members of the group Taphrosphys and its close relatives. The fenestra is oval, with a strong horizontal shelf overhanging it dorsally, and similar in shape to that seen in Azabbaremys.
Basisphenoid
Preservation: The basisphenoid in CNRST SUNY 199 is complete but not visible dorsally.
Contacts: The basisphenoid of Acleistochelys is similar in size and shape to that in Azabbaremys and Nigeremys, but has a broad anterior contact with the palatines, uniquely among bothremydids. Anterolaterally the basisphenoid contacts the pterygoids, posteriorly the basioccipital, and posterolaterally the quadrate.
Structures: The basisphenoid in Acleistochelys is triangular in shape with a lateral process curving anterolaterally into the fossa pterygoidea. In the fossa pterygoidea the basisphenoid forms part of the foramen posterius canalis caroticus interni, along with the pterygoid and quadrate see Pterygoid. The basisphenoid forms the anterior margin of a midline concavity see Basioccipital.
Postcranial Morphology
There are a large number of shell fragments associated with the type skull, as well as some postcranial bones. Among the more useful shell elements are a first neural, a fourth neural, part of the nuchal, and an anterior peripheral fragment. The carapace elements are all relatively thick; the first neural averages 2.1 centimeters for an element 10.9 centimeters in length.
The first neural bone in Acleistochelys is four-sided with the first vertebral-second vertebral sulcus running transversely across it as in other Pelomedusoides. The orientation of the first neural is unclear, but the most likely anterior direction has been chosen by comparison with Taphrosphys. It seems to be slightly thinner from side to side in this direction, as in Taphrosphys.
The fourth neural is identified as such because it lacks an intervertebral sulcus, making it either number two or number four, and its anterior sutural surface does not mate with neural one so it must be neural four. On the ventral surface of the neurals is the longitudinal ridge that is fused to the neural spine of the thoracic centra. Some fragments of thoracic centra are present but do not articulate with the neurals.
The nuchal bone of CNRST SUNY 199 is represented by the anterior margin from the midline to the left contact with the first peripheral. The midline sulcus is present and shows that a cervical scale is absent, as in all other Pelomedusoides. As with the neurals, the bone is very thick, but restoring the width of the neural at 27 cms results in a very large shell when extrapolated, so the thickness may be just a function of size. The nuchal in Acleistochelys has a transverse anterior margin, as in Taphrosphys, not embayed as in Chedighaii. The anterior edge of the nuchal is rounded and relatively obtuse rather than thin and acute as in Taphrosphys. Laterally the edge becomes even thicker in Acleistochelys, also in contrast to Taphrosphys in which it is acute and thin for its entire width.
One of the many carapace fragments may be the posterior position of peripheral two. It has the posterior sutural surface preserved, but lacks its anterior half and medial margin. The sulci for marginal two-marginal three and vertebral one-marginal three are present and shows that the marginals are distal to the peripheral-costal suture, as in other bothremydids. As with the nuchal, the free margin of the peripheral is thick and obtuse, with a rounded edge.
A single cervical, lacking its neural spine, is present in CNRST SUNY 199. It is procoelus, as in cervicals 2-8 of all other Pelomedusoides Williams, 1950. Unfortunately, cervical vertebrae of bothremydids are very poorly known, and only fragments exist of Taphrosphys, Bothremys, and Chedighaii. The more complete ones are a centrum lacking the entire neural arch, FMNH PR 268, either Bothremys or Chedighaii; and a nearly complete cervical, FMNH 247, Bothremys. There are some neural spines known for Taphrosphys sulcatus, but as this area is broken off in CNRST SUNY 199, they are not relevant. The Acleistochelys cervical is not determinable as to position; it is probably more posterior in the series based on general size and shape comparison with Pelusios and Pelomedusa. The Acleistochelys and Bothremydini centra differ significantly from pelomedusids in being much higher and shorter and with a deep ventral process. The Acleistochelys cervical is very similar to the Bothremydini cervicals differing from FMNH PR 247 in being longer with a narrower ventral process. The Acleistochelys cervical is nearly identical to FMNH PR 268. The differences with PR 247 could be the result of relative position in the column. The bothremydid cervicals have prominent transverse processes, also found in recent pleurodires. The central articulations in the bothremydids are nearly circular; the ones in pelomedusids vary from circular to wider than high. The base of the neural arch in the three bothremydid cervicals is much narrower than in pelomedusids. So the Acleistochelys cervical agrees with the Bothremydini cervicals, but the systematic significance of this is unknown at present.
The Acleistochelys carapace elements show an irregular surface texture that is similar to but not as well defined as that seen in Taphrosphys sulcatus. The texture consists of anastomosing furrows forming raised humps.
Summary
Acleistochelys maliensis is a new bothremydid side-necked turtle from the Paleocene Teberemt Formation of Mali. It is based on a single individual with a well-preserved skull and a few postcranial fragments. Characters of the skull indicate that it is most closely related to Azabbaremys moragjonesi also from the Paleocene Teberemt Formation of Mali. Although the phylogenetic analysis Gaffney et al. 2006 places these two as sister taxa, based on the narrow, posteriorly unattached vomer, the skulls differ in many morphologic features. Compared with Azabbaremys, Acleistochelys has a narrower skull, a narrower triturating surface with a small pit and no corrugations and an accessory ridge, a jugal exposed on the triturating surface, a wide palatine-basisphenoid contact separating the pterygoids, a supraoccipital-quadrate contact, a fossa pterygoidea, and a foramen posterius canalis carotici interni formed by the pterygoid, basisphenoid, and quadrate.
The narrow skull, narrow and smooth triturating surfaces, and fossa pterygoidea present in Acleistochelys also occur in Arenila, another Taphrosphyini from the Maastrichtian of Egypt. In the cladogram of Gaffney et al. 2006: fig. 288 the deep and narrow fossa pterygoidea evolves three times independently: in Acleistochelys, in Nigeremys + Arenila, and in Foxemys + Polysternon. Azabbaremys and Acleistochelys are only weakly resolved. Moving Acleistochelys to the group Nigeremys + Arenila results in a cladogram only one step longer than the shortest. So the possibility of a close relationship of Acleistochelys with Arenila and its close relative, Nigeremys, should be kept in mind.
The presence in the near-shore marine sediments of Mali of two large species of closely related bothremydid pleurodires is not unusual. The near-shore marine phosphates of the North African Paleogene have yielded over a dozen taxa of pleurodires in recent years Gaffney et al. 2006, and it was clearly an area of bothremydid diversification in the Paleogene.
Acknowledgments
We thank M. Haidara and Y. Maiga of the Centre Nationale de la Recherche Scientifique et Technologique and M. Dembelé of the Institut des Sciences Humaines, Bamako, Mali. The project was accomplished with the assistance of several collaborators from the U.S.A. and Mali: M. Doumbia, M. Gama, S. Keita, R. McIntosh, M. J. Novacek, M. Sanogo, and B. Traoré. The authors would like to thank the L.S.B. Leakey Foundation, which funded the fieldwork resulting in the discovery of Acleistochelys. We greatly appreciate the efforts of our two reviewers, C. J. Bell and P. Meylan, for taking the time to improve the text. Figures 2–Figure 3Figure 45 are the work of Coleen Lodge assisted by Frank Ippolito. The senior author would like to thank his colleagues, P. Meylan and H. Tong for discussions of pleurodire systematics.
